CONTROLLER AND METHOD FOR PREDICTING FAULTS IN HEATING, VENTILATION, OR AIR CONDITIONING (HVAC) SYSTEMS - Patent application

Abstract

An oil management controller for a heating, ventilation, or air conditioning (HVAC) system includes a processing circuit. The processing circuit is configured to analyze operational data of the HVAC system using a machine learning model to predict a variable state or condition of oil used by the HVAC system. The processing circuit is configured to identify an oil deficiency based on the variable state or condition of the oil. The processing circuit is configured to automatically initiate corrective action in response to identifying the oil deficiency.

Description

[Technical field]

[0001] The present disclosure relates generally to the field of operating building equipment, and more specifically to predicting the condition of building equipment using artificial intelligence. [Background technology]

[0002] In order for building equipment (e.g., heating, ventilation, or air conditioning (HVAC) equipment) to operate effectively and minimize deterioration of the building equipment, various operating conditions of the building equipment must be monitored and taken into account. However, conventional building systems leave many operating conditions unmonitored, which can lead to rapid deterioration of the building equipment and increased costs over time. Summary of the Invention

[0003] One embodiment of the disclosure is an oil management controller for a heating, ventilation, or air conditioning (HVAC) system. The controller includes a processing circuit. The processing circuit is configured to analyze operational data of the HVAC system using a machine learning model to predict a variable state or condition of oil used by the HVAC system. The processing circuit is configured to identify an oil starvation based on the variable state or condition of the oil. The processing circuit is configured to automatically initiate a corrective action in response to identifying the oil starvation.

[0004] The oil variable or condition is the amount of oil in the HVAC system. Identifying an oil shortage includes determining that the amount of oil in the HVAC system is below a threshold amount. Corrective action includes providing more oil to the HVAC system to increase the amount of oil in the HVAC system.

[0005] In some embodiments, the variable state or condition of the oil is the viscosity of an oil-refrigerant mixture in the HVAC system. Identifying the oil starvation includes determining that the viscosity of the oil-refrigerant mixture is below a threshold viscosity. Corrective action includes providing more oil to the HVAC system to increase the viscosity of the oil-refrigerant mixture in the HVAC system.

[0006] In some embodiments, the HVAC unit is capable of operating at different operating speeds, the variable state or condition of the oil is the viscosity of the oil-refrigerant mixture in the HVAC unit, and the corrective action includes setting an upper limit on the operating speed of the HVAC unit based on the viscosity of the oil-refrigerant mixture.

[0007] In some embodiments, the HVAC apparatus is coupled to a refrigerant loop that circulates an oil-refrigerant mixture between the HVAC apparatus and one or more other devices coupled to the refrigerant loop, and the corrective action includes operating the HVAC apparatus to circulate the oil-refrigerant mixture within the refrigerant loop, thereby returning oil from the one or more other devices to the HVAC apparatus.

[0008] In some embodiments, the machine learning model is a convolutional neural network (CNN) model having an input layer, one or more hidden layers, and an output layer. Analyzing the operational data includes providing the operational data as an input to an input layer of the CNN model and obtaining a prediction of an oil variable state or condition at an output layer of the CNN model.

[0009] In some embodiments, the machine learning model is a recurrent neural network (RNN) model. Analyzing the operational data includes providing a time series of values ​​of the operational data as an input to the RNN model and obtaining a prediction of a variable state or condition of the oil as an output of the RNN model.

[0010] In some embodiments, the processing circuitry is configured to generate the machine learning model using a set of training data obtained from the simulation model.

[0011] Another embodiment of the disclosure is a method for operating a heating, ventilation, or air conditioning (HVAC) system using a machine learning model. The method includes obtaining training data indicative of conditions affecting oil used by the HVAC system and a variable state or condition of the oil. The method includes generating a machine learning model by performing a training process based on the training data. The machine learning model is trained to predict the variable state or condition of the oil based on the conditions affecting the oil. The method includes predicting, using the machine learning model, whether the variable state or condition of the oil will violate a threshold. The method includes automatically initiating a corrective action in response to predicting that the variable state or condition of the oil will violate the threshold.

[0012] In some embodiments, the oil variable state or condition is an amount of oil in the HVAC system, the threshold is a minimum threshold for the amount of oil in the HVAC system, and the corrective action includes providing more oil to the HVAC system to increase the amount of oil in the HVAC system.

[0013] In some embodiments, the oil variable state or condition is a viscosity of the oil-refrigerant mixture in the HVAC system, the threshold value is a minimum threshold value for the viscosity of the oil-refrigerant mixture, and the corrective action includes providing more oil to the HVAC system to increase the viscosity of the oil-refrigerant mixture in the HVAC system.

[0014] In some embodiments, the HVAC unit is capable of operating at different operating speeds, the variable state or condition of the oil is the viscosity of the oil-refrigerant mixture in the HVAC unit, and the corrective action includes setting an upper limit on the operating speed of the HVAC unit based on the viscosity of the oil-refrigerant mixture.

[0015] In some embodiments, the HVAC apparatus is coupled to a refrigerant loop that circulates an oil-refrigerant mixture between the HVAC apparatus and one or more other devices coupled to the refrigerant loop, and the corrective action includes operating the HVAC apparatus to circulate the oil-refrigerant mixture within the refrigerant loop, thereby returning oil from the one or more other devices to the HVAC apparatus.

[0016] In some embodiments, the machine learning model is a convolutional neural network (CNN) model having an input layer, one or more hidden layers, and an output layer. Determining whether an oil variable state or condition violates a constraint using the machine learning model includes providing the operational data as an input to an input layer of the CNN model, and obtaining a prediction of the oil variable state or condition at an output layer of the CNN model.

[0017] In some embodiments, the machine learning model is a recurrent neural network (RNN) model. Determining whether an oil variable state or condition violates a constraint using the machine learning model includes providing a time series of values ​​of the operational data as an input to the RNN model, and obtaining a prediction of the oil variable state or condition as an output of the RNN model.

[0018] In some embodiments, obtaining the training data includes obtaining a simulation model that simulates operation of the HVAC system and changes in an oil variable or condition over time, and obtaining the training data includes running the simulation model to generate the training data.

[0019] Another embodiment of the present disclosure is an environmental control system for a building. The system includes a heating, ventilation, or air conditioning (HVAC) device operable to affect environmental conditions of the building. The system includes a controller including a processing circuit. The processing circuit is configured to analyze operational data of the HVAC device using a machine learning model to predict a variable state or condition of oil used by the HVAC device. The processing circuit is configured to identify an oil shortage based on the variable state or condition of the oil. The processing circuit is configured to automatically initiate a corrective action in response to identifying the oil shortage.

[0020] In some embodiments, the oil variable state or condition is an amount of oil in the HVAC system, identifying an oil shortage includes determining that the amount of oil in the HVAC system is below a threshold amount, and corrective action includes providing more oil to the HVAC system to increase the amount of oil in the HVAC system.

[0021] In some embodiments, the variable state or condition of the oil is the viscosity of an oil-refrigerant mixture in the HVAC system. Identifying the oil starvation includes determining that the viscosity of the oil-refrigerant mixture is below a threshold viscosity. Corrective action includes providing more oil to the HVAC system to increase the viscosity of the oil-refrigerant mixture in the HVAC system.

[0022] In some embodiments, the HVAC unit is capable of operating at different operating speeds, the variable state or condition of the oil is the viscosity of the oil-refrigerant mixture in the HVAC unit, and the corrective action includes setting an upper limit on the operating speed of the HVAC unit based on the viscosity of the oil-refrigerant mixture.

[0023] Another embodiment of the present disclosure is a controller for operating a compressor motor in a heating, ventilation, or air conditioning (HVAC) system according to some embodiments. The controller includes a processing circuit. The processing circuit is configured to obtain a machine learning model that predicts an amplitude set point for a current provided to the motor. The amplitude set point affects vibration of the motor. The processing circuit is configured to analyze operational data of the motor using the machine learning model to predict an amplitude set point for the current. The processing circuit is configured to operate the motor based on the amplitude set point.

[0024] In some embodiments, operating the motor based on the amplitude set point includes providing the amplitude set point to an inverter, the inverter configured to provide current to the motor.

[0025] In some embodiments, inputs to the machine learning model include at least one of an axis error between the direct and quadrature axes of the motor, a frequency of the current, an actual noise level associated with the motor, or a desired noise level associated with the motor.

[0026] In some embodiments, the machine learning model is trained to learn the correlation between the operational data and the amount of noise produced by the motor, which is used as a proxy to predict motor vibration.

[0027] In some embodiments, the processing circuitry is further configured to generate a machine learning model using the set of training data obtained from the simulation model.

[0028] In some embodiments, the machine learning model is a recurrent neural network (RNN) model. Analyzing the operational data includes providing a time series of values ​​of the operational data as an input to the RNN model and obtaining a prediction of the amplitude setting value as an output of the RNN model.

[0029] In some embodiments, the current is alternating current (AC). The frequency of the AC affects the rotational speed of the motor, and the amplitude of the AC affects the torque applied by the motor.

[0030] Another embodiment of the present disclosure is a controller for predicting faults in a variable refrigerant flow (VRF) system. The controller comprises a processing circuit. The processing circuit is configured to analyze operational data of the VRF system using a machine learning model to predict a fault classification of the VRF system. The fault classification identifies a fault condition affecting the VRF system. The processing circuit is configured to identify a VRF device of the VRF system associated with the fault condition. The processing circuit is configured to automatically initiate corrective action to address the fault condition in response to identifying the device and the fault condition.

[0031] In some embodiments, the fault classification includes a severity metric associated with the fault condition. The severity metric indicates an impact of the fault condition on the VRF system. Corrective action is determined based on the severity metric.

[0032] In some embodiments, the machine learning model is a recurrent neural network (RNN) model. Analyzing the operational data includes providing a time series of values ​​of the operational data as an input to the RNN model and obtaining a prediction of the fault classification as an output of the RNN model.

[0033] In some embodiments, the processing circuitry is further configured to generate a machine learning model using the set of training data obtained from the simulation model.

[0034] In some embodiments, the fault classification identifies multiple fault conditions affecting a VRF system, the multiple fault conditions being associated with multiple VRF devices of the VRF system.

[0035] In some embodiments, the fault condition is at least one of a refrigerant leak, frosting of the outdoor unit, a clogged indoor fan, a clogged indoor filter, a clogged heat exchanger, a clogged outdoor fan, a demagnetized motor, or an oil leak from the compressor.

[0036] Another embodiment of the disclosure is a method for operating a motor of a compressor in a heating, ventilation, or air conditioning (HVAC) system, according to some embodiments. The method includes obtaining a machine learning model that predicts an amplitude set point for a current provided to the motor. The amplitude set point affects vibration of the motor. The method includes analyzing operational data of the motor using the machine learning model to predict the amplitude set point for the current. The method includes operating the motor based on the amplitude set point.

[0037] In some embodiments, operating the motor based on the amplitude set point includes providing the amplitude set point to an inverter, the inverter configured to provide current to the motor.

[0038] In some embodiments, inputs to the machine learning model include at least one of an axis error between the direct and quadrature axes of the motor, a frequency of the current, an actual noise level associated with the motor, or a desired noise level associated with the motor.

[0039] In some embodiments, the machine learning model is trained to learn the correlation between the operational data and the amount of noise produced by the motor, which is used as a proxy to predict motor vibration.

[0040] In some embodiments, the method includes generating a machine learning model using a set of training data obtained from the simulation model.

[0041] In some embodiments, the machine learning model is a recurrent neural network (RNN) model. Analyzing the operational data includes providing a time series of values ​​of the operational data as an input to the RNN model, and obtaining a prediction of the amplitude setting value as an output of the RNN model.

[0042] In some embodiments, the current is alternating current (AC). The frequency of the AC affects the rotational speed of the motor, and the amplitude of the AC affects the torque applied by the motor.

[0043] Those skilled in the art will appreciate that the summary is merely illustrative and is not intended to be limiting in any manner. Other aspects, features, and advantages of the devices and / or processes described herein, as defined solely by the claims, will become apparent in the detailed description set forth herein and taken in conjunction with the accompanying drawings. [Brief description of the drawings]

[0044] Various objects, aspects, features, and advantages of the present disclosure will become more apparent and be better understood by reference to the detailed description taken in conjunction with the accompanying drawings, in which like reference numbers generally indicate identical, functionally similar, and / or structurally similar elements.

[0045] [Figure 1] FIG. 1 is a diagram of a building with an HVAC system, according to some embodiments. [Diagram 2] FIG. 2 is a block diagram of a waterside system that may be used to service the heating or cooling load of the building of FIG. 1 according to some embodiments. [Diagram 3] FIG. 2 is a block diagram of an airside system that may be used to service the heating or cooling loads of the building of FIG. 1 according to some embodiments. [Figure 4]FIG. 2 is a block diagram of a building management system (BMS) that can be used to monitor and control the building of FIG. 1 according to some embodiments. [Diagram 5] FIG. 2 is a block diagram of another BMS that can be used to monitor and control the building of FIG. 1 according to some embodiments. [Figure 6A] FIG. 1 is a diagram of a variable refrigerant flow (VRF) system having one or more outdoor VRF units and multiple indoor VRF units, according to some embodiments. [Figure 6B] FIG. 1 is a diagram of a variable refrigerant flow (VRF) system having one or more outdoor VRF units and multiple indoor VRF units, according to some embodiments. [Figure 7A] 1 is a schematic diagram of a VRF system, according to some embodiments. [Figure 7B] FIG. 1 is a block diagram of a VRF system, according to some embodiments. [Figure 8] FIG. 2 is a block diagram of a controller for predicting oil properties according to some embodiments. [Figure 9A] 1 is an illustration of a recurrent neural network (RNN) structure according to some embodiments. [Figure 9B] 1 is an illustration of a neural network (NN) architecture, according to some embodiments. [Figure 10] FIG. 1 is a flow diagram of a process for monitoring oil properties using an AI model, according to some embodiments. [Figure 11A] 1 is a graph illustrating the change in RMSE based on several iterations in an exemplary model training process for an artificial intelligence (AI) model, according to some embodiments. [Figure 11B] 11B is a graph illustrating the change in loss based on the number of iterations associated with the AI ​​model of FIG. 11A in accordance with some embodiments. [Figure 12A] 11B is a graph illustrating a prediction of compressor oil volume generated by the AI ​​model of FIG. 11A in accordance with some embodiments. [Figure 12B] 11B is a graph illustrating a prediction of accumulator oil volume generated by the AI ​​model of FIG. 11A in accordance with some embodiments. [Figure 12C] 11B is a graph illustrating a prediction of oil viscosity generated by the AI ​​model of FIG. 11A in accordance with some embodiments. [Figure 13A] FIG. 1 is a block diagram of a VRF system including a compressor vibration controller according to some embodiments. [Figure 13B] FIG. 13B is a block diagram of the VRF system of FIG. 13A in more detail, according to some embodiments. [Figure 14A] 1 is a graph illustrating Kirchhoff's Law, according to some embodiments. [Figure 14B] 1 is a graph illustrating αβ conversion for a motor according to some embodiments. [Figure 14C] 1 is a graph illustrating a direct quadrature transform (dq) for a motor according to some embodiments. [Figure 15] 4 is a graph illustrating the relationship between crank angle and torque for different types of compressors according to some embodiments. [Figure 16] FIG. 13B is a block diagram of the compressor vibration controller of FIG. 13A in more detail according to some embodiments. [Figure 17A] 1 is an illustration of a neural network for predicting a value of current to provide to a compressor motor, according to some embodiments. [Figure 17B] 1 is an illustration of another neural network for predicting a value of current to provide to a compressor motor, according to some embodiments. [Figure 18] FIG. 1 is a flow diagram of a process for predicting an AC signal amplitude to provide to a compressor using an AI model, according to some embodiments. [Figure 19] FIG. 1 is a block diagram of a VRF system, according to some embodiments. [Figure 20]FIG. 2 is a block diagram of a VRF failure controller for predicting failures in a VRF system, according to some embodiments. [Figure 21] FIG. 1 is a diagram of a neural network for predicting fault classification in a VRF system, according to some embodiments. [Figure 22] FIG. 1 is a flow diagram of a process for predicting a fault classification in a VRF system using an AI model, according to some embodiments. [Figure 23] FIG. 1 is a block diagram of a motor efficiency controller, according to some embodiments. [Figure 24] FIG. 1 is a flow diagram of a process for predicting the efficiency of a motor in a VRF system using an AI model, according to some embodiments. [Figure 25A] 1 is an illustration of a recurrent neural network structure according to some embodiments. [Figure 25B] 1 is an illustration of a long short-term memory model structure according to some embodiments. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0046] overview Reference is generally made to diagrams, systems, and methods for predicting the performance of a building's variable refrigerant flow (VRF) system and utilizing artificial intelligence (AI) in operating the VRF system and VRF system components, according to some embodiments. In particular, the present disclosure utilizes AI to predict the characteristics of oils used in VRF systems, as well as to predict various conditions and characteristics of motors and compressors within the VRF system.

[0047] However, it should be understood that the systems and methods described herein are not limited to VRF systems. Rather, the VRF system is shown and described for example purposes only as one potential implementation of the present disclosure. The systems and methods described herein can be applied to various systems that require oil to be provided to the device (e.g., other climate control systems), as well as other types of systems, including compressors, motors, any type of device that uses oil, and / or any type of device that may experience vibration or disturbance during operation. For example, the systems and methods described herein can be applied to various heating, ventilation, or air conditioning (HVAC) systems and devices (e.g., various air conditioners, variable air volume (VAV) systems, residential air conditioning (RAC) systems, etc.).

[0048] As referred to herein, AI and AI models can be used to describe a variety of different models that can be used in predicting state and other information associated with devices in a VRF system. In some embodiments, a recurrent neural network (RNN) model is utilized to generate the prediction. RNNs are a class of artificial neural networks in which connections between nodes form a directed graph along a temporal sequence. More specifically, a long short-term memory (LSTM) model can be utilized in generating the prediction. LSTM is a particular type of artificial RNN architecture that is primarily used in deep learning. LSTMs can classify and process entire sequences of time series data and can make predictions based on the time series data. Advantageously, LSTMs can account for delays of unknown duration between significant events in a time series. In some embodiments, other types of AI models, such as convolutional neural networks (CNNs), are utilized in generating the predictions. Thus, it should be appreciated that various types of AI models can be utilized in generating the predictions.

[0049] As defined herein, oil characteristics, which are used interchangeably herein with the term "oil characteristics," can refer to a particular property of the oil. In other words, oil characteristics can be a variable state or condition of the oil. Oil characteristics of a VRF system (i.e., variable state or condition of the oil) can include, for example, the amount of oil in one or more compressors of the VRF system, the amount of oil in an accumulator of the VRF system, the viscosity of the oil, etc. In some embodiments, the viscosity of the oil-refrigerant mixture is estimated instead of or in addition to simply the viscosity of the oil. In this case, the oil-refrigerant mixture can be output by a compressor of the VRF system as a result of the oil being integrated into the compressed refrigerant typically output by the compressor. When the VRF system is operated, the oil characteristics can change over time, thereby leading to changes in the operation of building devices (e.g., compressors, valves, oil separators, etc.) in the VRF system. For example, if a compressor using oil operates at a higher speed, the amount of oil in the compressor can decrease and the viscosity of the oil can decrease due to the higher operating temperature of the compressor.

[0050] Specifically, with respect to a VRF system, a conventional time-based oil return system may periodically interrupt the heating and / or cooling provided by the VRF system, reducing the overall efficiency of the VRF system. In a refrigeration cycle, oil drift and other related problems may result in a particular indoor or outdoor heat exchanger being utilized for a long period of time, which is often detrimental to the overall operating condition of the VRF system. Furthermore, if a VRF device (e.g., a compressor) cannot receive sufficient oil, the VRF device may be at high risk of failure. Although a conventional oil return system can mitigate failures and other catastrophic problems associated with inaccurate oil return, there is often a significant impact on the efficiency of the VRF system. For example, with a periodic oil return, after a transition to an oil return state, the compressor of the VRF system may need to reduce its operating speed and / or restart, thereby resulting in efficiency loss of the heating / cooling system.

[0051] As described in more detail below, issues associated with traditional oil return systems (e.g., time-based oil return) can be addressed through the use of AI. AI can be used to predict oil volume and oil viscosity in various building devices. Based on the prediction, the AI ​​can determine the optimal time to perform an oil return when the VRF system is in cooling / heating operation. Advantageously, the prediction performed by the AI ​​can be made without the use of oil sensors. Not having to utilize oil sensors to detect oil conditions (e.g., oil volume, oil viscosity, etc.) can reduce costs due to fewer parts to purchase and maintain.

[0052] In some embodiments, an AI model can be utilized to manage vibrations associated with a compressor of a VRF system. Excessive vibrations of a compressor can result in rapid degradation due to higher operating, maintenance, and replacement costs, and thus increased costs over time. To manage vibrations, a specific AI model (e.g., an RNN model) can be trained to predict target currents associated with the direct (D) and quadrature (Q) axes. The prediction of the target currents can be based on inputs associated with the compressor, such as, for example, the frequency provided by the inverter, the actual noise level, the q-axis feedback current, and the axis error between the axes. The predicted D-axis and Q-axis currents can be used to determine the correlated vibrations of the compressor. If the predicted current and / or the predicted vibrations are too high (e.g., the predicted current is higher than a threshold current value), the operation of the compressor can be modified to avoid excessive vibrations, thereby avoiding rapid degradation of the compressor. Specifically, constant input current and speed control can be implemented to reduce vibrations.

[0053] In some embodiments, an AI model may be utilized to predict fault conditions of a compressor in a VRF system. A fault condition may refer to an operating condition of the compressor that is beyond a preferred operating condition. For example, fault conditions of a VRF system may include refrigerant leaks, frosting of the outdoor unit, clogged indoor fan, dirty indoor filters, dirty heat exchangers, dirty outdoor fans, motor demagnetization, compressor oil leaks, and other conditions that result in imperfect efficiency of the compressor. To predict the conditions, an AI model (e.g., an RNN model) may be trained to map values ​​of inputs associated with the compressor to a fault classification. The fault classification may include an indication of what fault conditions (if any) are identified for the VRF system based on a set of input data. For example, the AI ​​model may utilize inputs such as compressor speed, ambient temperature, discharge temperature, suction pressure, discharge pressure, indoor fan mode, outdoor fan step, etc. Based on the fault classification, various corrective actions may be initiated in response. As defined herein, a corrective action may refer to any action taken to address a fault and / or some undesirable condition. For example, corrective actions may include scheduling maintenance, alerting building operators to the fault, disabling particular devices (e.g., particular compressors), operating particular devices, etc. In this manner, faults affecting compressors can be addressed quickly and efficiently.

[0054] In some embodiments, the AI ​​model is utilized to predict the efficiency of a motor in a VRF system (or other system). Motor efficiency can directly impact costs over a period of time, especially if the motor is operated at a high level of inefficiency. As the efficiency of a motor decreases, costs may increase as additional resources (e.g., electricity, water, etc.) are consumed to generate the desired output. Additionally, the motor may deteriorate at a faster rate as a result of having to operate under more intensive operating conditions (e.g., at higher revolutions per minute) to generate the desired output. In some embodiments, the current efficiency of a motor may be expressed as a percentage of how efficient the motor is compared to its maximum efficiency value when the motor was originally installed. Motor efficiency may be determined based on the amount of resources required as input to effect a particular change. For example, the efficiency of an electric motor in a compressor may be defined by the amount of electricity required to compress a given amount of gas (e.g., air). In an example, a 50% efficiency of an electric motor may indicate that the motor is half as efficient as when it was originally installed, thereby requiring twice the power to compress a given amount of gas. To predict the efficiency of a motor, an AI model can be trained to map various inputs (e.g., the amount of power consumed by the motor, the ambient temperature, the motor's revolutions per minute (RPM) value, etc.) to an efficiency value for the motor.

[0055] It should be noted that while the AI ​​models used to predict vibration, fault conditions, and motor efficiency are described separately, the present disclosure also contemplates one or more aggregate AI models predicting one or more of the aforementioned predictive objectives. For example, a single AI model may be trained to predict currents associated with vibration and fault conditions that may affect a compressor. These and other features of the present disclosure are discussed in more detail below.

[0056] Construction of HVAC systems and building management systems 1-5, there are shown several Building Management Systems (BMS) and HVAC systems in which the systems and methods of the present disclosure may be implemented, according to some embodiments. In a brief overview, FIG. 1 shows a building 10 with an HVAC system 100. FIG. 2 is a block diagram of a waterside system 200 that may be used to provide services to the building 10. FIG. 3 is a block diagram of an airside system 300 that may be used to provide services to the building 10. FIG. 4 is a block diagram of a BMS that may be used to monitor and control the building 10. FIG. 5 is a block diagram of another BMS that may be used to monitor and control the building 10.

[0057] Buildings and HVAC Systems 1, a perspective view of a building 10 is shown. The building 10 is served by a BMS. A BMS is generally a system of devices configured to control, monitor, and manage equipment in or around a building or building area. A BMS may include, for example, an HVAC system, a security system, a lighting system, a fire alarm system, any other system capable of managing building functions or devices, or any combination thereof.

[0058] The BMS serving the building 10 includes an HVAC system 100. The HVAC system 100 may include a number of HVAC devices (e.g., heaters, chillers, air handling units, pumps, fans, thermal energy storage, etc.) configured to provide heating, cooling, ventilation, or other services for the building 10. For example, the HVAC system 100 is shown to include a waterside system 120 and an airside system 130. The waterside system 120 may provide a heated or cooled fluid to an air handling unit of the airside system 130. The airside system 130 may use the heated or cooled fluid to heat or cool an airflow provided to the building 10. Exemplary waterside and airside systems that may be used in the HVAC system 100 are described in more detail with reference to FIGS. 2-3.

[0059] The HVAC system 100 is shown to include a chiller 102, a boiler 104, and a rooftop air handling unit (AHU) 106. The waterside system 120 may use the boiler 104 and the chiller 102 to heat or cool a working fluid (e.g., water, glycol, etc.) and circulate the working fluid to the AHU 106. In various embodiments, the HVAC devices of the waterside system 120 may be located in or around the building 10 (as shown in FIG. 1 ) or at an off-site location such as a central plant (e.g., a chiller plant, a steam plant, a heat plant, etc.). The working fluid may be heated in the boiler 104 or cooled in the chiller 102 depending on whether heating or cooling is required in the building 10. The boiler 104 may add heat to the circulated fluid, for example, by burning a combustible material (e.g., natural gas) or by using an electric heating element. The chiller 102 may place the circulated fluid in a heat exchange relationship with another fluid (e.g., a refrigerant) in a heat exchanger (e.g., an evaporator) to absorb heat from the circulated fluid. The working fluid from the chiller 102 and / or the boiler 104 may be transported to the AHU 106 via piping 108.

[0060] The AHU 106 may place the working fluid in heat exchange relationship with an airflow passing through the AHU 106 (e.g., via one or more stages of cooling and / or heating coils). The airflow may be, for example, outside air, return air from within the building 10, or a combination of both. The AHU 106 may transfer heat between the airflow and the working fluid to provide heating or cooling of the airflow. For example, the AHU 106 may include one or more fans or blowers configured to pass the airflow over or through a heat exchanger that contains the working fluid. The working fluid may then return to the chiller 102 or boiler 104 via piping 110.

[0061] The airside system 130 may deliver the airflow supplied by the AHU 106 (i.e., supply airflow) to the building 10 via the air supply duct 112 and may provide return air from the building 10 to the AHU 106 via the air return duct 114. In some embodiments, the airside system 130 includes multiple variable air volume (VAV) units 116. For example, the airside system 130 is shown to include a separate VAV unit 116 for each floor or zone of the building 10. The VAV units 116 may include dampers or other flow control elements that may be operated to control the amount of supply airflow provided to individual zones of the building 10. In other embodiments, the airside system 130 delivers the supply airflow (e.g., via the supply duct 112) to one or more zones of the building 10 without the use of intermediate VAV units 116 or other flow control elements. The AHU 106 may include various sensors (e.g., temperature sensors, pressure sensors, etc.) configured to measure attributes of the supply airflow. The AHU 106 may receive input from sensors located within the AHU 106 and / or within the building zone, and may adjust the flow rate, temperature, or other attributes of the supply air flow through the AHU 106 to achieve setpoint conditions for the building zone.

[0062] Waterside System 2, a block diagram of a waterside system 200 is shown, according to some embodiments. In various embodiments, the waterside system 200 may supplement or replace the waterside system 120 in the HVAC system 100, or may be implemented separately from the HVAC system 100. When implemented in the HVAC system 100, the waterside system 200 may include a subset of the HVAC devices in the HVAC system 100 (e.g., the boiler 104, the chiller 102, pumps, valves, etc.) and may operate to supply heated or cooled fluid to the AHU 106. The HVAC devices of the waterside system 200 may be located in the building 10 (e.g., as components of the waterside system 120) or at an off-site location, such as a central plant.

[0063] In FIG. 2, the waterside system 200 is shown as a central plant having multiple subplants 202-212. The subplants 202-212 are shown to include a heater subplant 202, a heat recovery chiller subplant 204, a chiller subplant 206, a cooling tower subplant 208, a high temperature thermal energy storage (TES) subplant 210, and a low temperature thermal energy storage (TES) subplant 212. The subplants 202-212 consume resources (e.g., water, natural gas, electricity, etc.) from a utility to provide thermal energy loads (e.g., hot water, chilled water, heating, cooling, etc.) for a building or campus. For example, the heater subplant 202 can be configured to heat water in a hot water loop 214 that circulates hot water between the heater subplant 202 and the building 10. The chiller subplant 206 can be configured to cool water in a chilled water loop 216 that circulates chilled water between the chiller subplant 206 and the building 10. Heat recovery chiller subplant 204 may be configured to transfer heat from chilled water loop 216 to hot water loop 214 to provide additional heating for the hot water and additional cooling for the chilled water. Condenser water loop 218 may absorb heat from the chilled water in chiller subplant 206 and reject the absorbed heat in cooling tower subplant 208 or transfer the absorbed heat to hot water loop 214. High temperature TES subplant 210 and low temperature TES subplant 212 may store high temperature and low temperature thermal energy, respectively, for subsequent use.

[0064] Hot water loop 214 and chilled water loop 216 may deliver heated and / or chilled water to an air handler (e.g., AHU 106) located on the roof of building 10 or to individual floors or zones (e.g., VAV units 116) of building 10. The air handler pushes air through a heat exchanger (e.g., heating coil or cooling coil) through which water flows to provide heating or cooling of the air. The heated or cooled air can be delivered to individual zones of building 10 to provide the thermal energy load of building 10. The water then returns to subplants 202-212 to receive further heating or cooling.

[0065] Although subplants 202-212 are shown and described as providing heated water and chilled water for circulation to the building, it is understood that any other type of working fluid (e.g., glycol, CO2, etc.) may be used instead of or in addition to water to provide the thermal energy load. In other embodiments, subplants 202-212 may provide heating and / or cooling directly to a building or campus without the need for an intermediate heat transfer fluid. These and other variations on waterside system 200 are within the teachings of the present disclosure.

[0066] Each of the subplants 202-212 may include various devices configured to facilitate the functioning of the subplant. For example, the heater subplant 202 is shown to include a number of heating elements 220 (e.g., boilers, electric heaters, etc.) configured to add heat to hot water in the hot water loop 214. The heater subplant 202 is also shown to include a number of pumps 222 and 224 configured to circulate the hot water in the hot water loop 214 and control the flow rate of the hot water through the individual heating elements 220. The chiller subplant 206 is shown to include a number of chillers 232 configured to remove heat from the cold water in the cold water loop 216. The chiller subplant 206 is also shown to include a number of pumps 234 and 236 configured to circulate the cold water in the cold water loop 216 and control the flow rate of the cold water through the individual chillers 232.

[0067] The heat recovery chiller subplant 204 is shown to include a number of heat recovery heat exchangers 226 (e.g., refrigeration circuits) configured to transfer heat from the cold water loop 216 to the hot water loop 214. The heat recovery chiller subplant 204 is also shown to include a number of pumps 228 and 230 configured to circulate hot and / or cold water through the heat recovery heat exchangers 226 and control the flow rate of water through the individual heat recovery heat exchangers 226. The cooling tower subplant 208 is shown to include a number of cooling towers 238 configured to remove heat from condenser water in the condenser water loop 218. The cooling tower subplant 208 is also shown to include a number of pumps 240 configured to circulate condenser water in the condenser water loop 218 and control the flow rate of condenser water through the individual cooling towers 238.

[0068] Hot TES subplant 210 is shown to include a hot TES tank 242 configured to store hot water for later use. Hot TES subplant 210 may also include one or more pumps or valves configured to control the flow of hot water to or from hot TES tank 242. Cold TES subplant 212 is shown to include a cold TES tank 244 configured to store cold water for later use. Cold TES subplant 212 may also include one or more pumps or valves configured to control the flow of cold water to or from cold TES tank 244.

[0069] In some embodiments, one or more pumps in waterside system 200 (e.g., pumps 222, 224, 228, 230, 234, 236, and / or 240) or pipelines in waterside system 200 include isolation valves associated therewith. The isolation valves may be integrated with the pumps or located upstream or downstream of the pumps to control the flow of fluids in waterside system 200. In various embodiments, waterside system 200 may include more, fewer, or different types of devices and / or subplants based on the particular configuration of waterside system 200 and the type of load served by waterside system 200.

[0070] Airside System 3, a block diagram of an airside system 300 is shown, according to some embodiments. In various embodiments, the airside system 300 may supplement or replace the airside system 130 in the HVAC system 100, or may be implemented separately from the HVAC system 100. When implemented in the HVAC system 100, the airside system 300 may include a subset of the HVAC devices in the HVAC system 100 (e.g., AHU 106, VAV unit 116, ducts 112-114, fans, dampers, etc.) and may be located in or around the building 10. The airside system 300 may operate to heat or cool an airflow provided to the building 10 using heated or cooled fluid provided by the waterside system 200.

[0071] In FIG. 3, the airside system 300 is shown to include an economizer-type air handling unit (AHU) 302. The economizer-type AHU varies the amount of outside air and return air that the air handling unit uses for heating or cooling. For example, the AHU 302 may receive return air 304 from a building zone 306 via a return air duct 308 and deliver supply air 310 to the building zone 306 via a supply air duct 312. In some embodiments, the AHU 302 is a rooftop unit located on the roof of the building 10 (e.g., the AHU 106 as shown in FIG. 1) or otherwise positioned to receive both return air 304 and outside air 314. The AHU 302 may be configured to operate an exhaust damper 316, a mixing damper 318, and an outside air damper 320 to control the amount of outside air 314 and return air 304 that are combined to form the supply air 310. Any return air 304 that does not pass through the mixing damper 318 may be exhausted from the AHU 302 through the exhaust damper 316 as exhaust air 322.

[0072] Each of the dampers 316-320 may be operated by an actuator. For example, the exhaust damper 316 may be operated by an actuator 324, the mixing damper 318 may be operated by an actuator 326, and the outside air damper 320 may be operated by an actuator 328. The actuators 324-328 may communicate with the AHU controller 330 via a communication link 332. The actuators 324-328 may receive control signals from the AHU controller 330 and may provide feedback signals to the AHU controller 330. The feedback signals may include, for example, an indication of current actuator or damper position, an amount of torque or force exerted by the actuator, diagnostic information (e.g., results of diagnostic tests performed by the actuators 324-328), status information, commissioning information, configuration settings, calibration data, and / or other types of information or data that may be collected, stored, or used by the actuators 324-328. The AHU controller 330 may be an economizer controller configured to control the actuators 324-328 using one or more control algorithms (e.g., a state-based algorithm, an extremum-seeking control (ESC) algorithm, a proportional-integral (PI) control algorithm, a proportional-integral-derivative (PID) control algorithm, a model predictive control (MPC) algorithm, a feedback control algorithm, etc.).

[0073] 3, the AHU 302 is shown to include a cooling coil 334, a heating coil 336, and a fan 338 disposed within the supply air duct 312. The fan 338 can be configured to force the supply air 310 through the cooling coil 334 and / or the heating coil 336 and provide the supply air 310 to the building zone 306. The AHU controller 330 can communicate with the fan 338 via a communication link 340 to control the flow rate of the supply air 310. In some embodiments, the AHU controller 330 controls the amount of heating or cooling applied to the supply air 310 by modulating the speed of the fan 338.

[0074] The cooling coil 334 may receive chilled fluid from the waterside system 200 (e.g., from the chilled water loop 216) via piping 342 and may return chilled fluid to the waterside system 200 via piping 344. A valve 346 may be disposed along the piping 342 or piping 344 to control the flow rate of chilled fluid through the cooling coil 334. In some embodiments, the cooling coil 334 includes multiple stages of cooling coils that can be activated and deactivated independently (e.g., by the AHU controller 330, by the BMS controller 366, etc.) to regulate the amount of cooling applied to the supply air 310.

[0075] The heating coil 336 may receive heated fluid from the waterside system 200 (e.g., from the hot water loop 214) via piping 348 and may return heated fluid to the waterside system 200 via piping 350. Valves 352 may be disposed along the piping 348 or the piping 350 to control the flow rate of heated fluid through the heating coil 336. In some embodiments, the heating coil 336 includes multiple stages of heating coils that can be activated and deactivated independently (e.g., by the AHU controller 330, by the BMS controller 366, etc.) to regulate the amount of heating applied to the supply air 310.

[0076] Each of the valves 346 and 352 may be controlled by an actuator. For example, the valve 346 may be controlled by the actuator 354 and the valve 352 may be controlled by the actuator 356. The actuators 354-356 may communicate with the AHU controller 330 via communication links 358-360. The actuators 354-356 may receive control signals from the AHU controller 330 and provide feedback signals to the controller 330. In some embodiments, the AHU controller 330 receives a measurement of the supply air temperature from a temperature sensor 362 located in the supply air duct 312 (e.g., downstream of the cooling coil 334 and / or the heating coil 336). The AHU controller 330 may also receive a measurement of the temperature of the building zone 306 from a temperature sensor 364 located in the building zone 306.

[0077] In some embodiments, the AHU controller 330 operates the valves 346 and 352 via the actuators 354-356 to regulate the amount of heating or cooling provided to the air 310 (e.g., to achieve a setpoint temperature for the supply air 310 or to maintain the temperature of the supply air 310 within a setpoint temperature range). The position of the valves 346 and 352 affects the amount of heating or cooling provided to the supply air 310 by the cooling coil 334 or heating coil 336, and may correlate to the amount of energy consumed to achieve a desired supply air temperature. The AHU 330 may control the temperature of the supply air 310 and / or building zone 306 by activating or deactivating the coils 334-336, adjusting the speed of the fan 338, or a combination of both.

[0078] 3, the airside system 300 is shown to include a building management system (BMS) controller 366 and a client device 368. The BMS controller 366 may include one or more computer systems (e.g., servers, supervisory controllers, subsystem controllers, etc.) that act as a system level controller, application or data server, head node, or master controller for the airside system 300, the waterside system 200, the HVAC system 100, and / or other controllable systems that service the building 10. The BMS controller 366 may communicate with multiple downstream building systems or subsystems (e.g., HVAC system 100, security system, lighting system, waterside system 200, etc.) via communication links 370 according to similar or different protocols (e.g., LON, BACnet, etc.). In various embodiments, the AHU controller 330 and the BMS controller 366 may be separate (as shown in FIG. 3) or integrated. In an integrated implementation, the AHU controller 330 may be a software module configured for execution by a processor of the BMS controller 366.

[0079] In some embodiments, the AHU controller 330 receives information (e.g., commands, set points, operating boundaries, etc.) from the BMS controller 366 and provides information (e.g., temperature measurements, valve or actuator positions, operating status, diagnostics, etc.) to the BMS controller 366. For example, the AHU controller 330 may provide to the BMS controller 366 temperature measurements from the temperature sensors 362-364, equipment on / off status, equipment operating capacity, and / or any other information that may be used by the BMS controller 366 to monitor or control variable states or conditions within the building zone 306.

[0080] The client device 368 may include one or more human-machine interfaces, or client interfaces (e.g., graphical user interfaces, report interfaces, text-based computer interfaces, client-enabled web services, web servers serving pages to web clients, etc.) for controlling, viewing, or otherwise interacting with the HVAC system 100, its subsystems, and / or devices. The client device 368 may be a computer workstation, a client terminal, a remote or local interface, or any other type of user interface device. The client device 368 may be a fixed terminal or a mobile device. For example, the client device 368 may be a desktop computer, a computer server with a user interface, a laptop computer, a tablet, a smartphone, a PDA, or any other type of mobile or non-mobile device. The client device 368 may communicate with the BMS controller 366 and / or the AHU controller 330 via a communication link 372.

[0081] Building Management Systems 4, a block diagram of a building management system (BMS) 400 is shown, according to some embodiments. The BMS 400 may be implemented in the building 10 to automatically monitor and control various building functions. The BMS 400 is shown to include a BMS controller 366 and a number of building subsystems 428. The building subsystems 428 are shown to include a building electrical subsystem 434, an information and communication technology (ICT) subsystem 436, a security subsystem 438, an HVAC subsystem 440, a lighting subsystem 442, a lift / escalator subsystem 432, and a fire safety subsystem 430. In various embodiments, the building subsystems 428 may include fewer, additional, or alternative subsystems. For example, the building subsystems 428 may also or alternatively include a refrigeration subsystem, an advertising or signage subsystem, a cooking subsystem, a vending subsystem, a printer or copy service subsystem, or any other type of building subsystem that monitors or controls the building 10 using controllable devices and / or sensors. In some embodiments, the building subsystems 428 include the waterside system 200 and / or the airside system 300 described with reference to Figures 2-3.

[0082] Each of the building subsystems 428 may include any number of devices, controllers, and connections to complete its individual functions and control activities. The HVAC subsystem 440 may include many of the same components as the HVAC system 100 described with reference to Figures 1-3. For example, the HVAC subsystem 440 may include chillers, boilers, any number of air handling units, economizers, field controllers, supervisory controllers, actuators, temperature sensors, and other devices for controlling the temperature, humidity, airflow, or other variable conditions within the building 10. The lighting subsystem 442 may include any number of lighting fixtures, ballasts, lighting sensors, dimmers, or other devices configured to controllably adjust the amount of light provided to the building space. The security subsystem 438 may include occupancy sensors, video surveillance cameras, digital video recorders, video processing servers, intrusion detection devices, access control devices and servers, or other security related devices.

[0083] 4, the BMS controller 366 is shown to include a communication interface 407 and a BMS interface 409. The interface 407 may facilitate communication between the BMS controller 366 and external applications (e.g., monitoring and reporting applications 422, enterprise control applications 426, remote systems and applications 444, applications residing on client devices 448, etc.) to enable user control, monitoring, and adjustment over the BMS controller 366 and / or subsystems 428. The interface 407 may also facilitate communication between the BMS controller 366 and client devices 448. The BMS interface 409 may facilitate communication between the BMS controller 366 and building subsystems 428 (e.g., HVAC, lighting security, lifts, power distribution, business, etc.).

[0084] The interfaces 407, 409 can be or include wired or wireless communication interfaces (e.g., jacks, antennas, transmitters, receivers, transceivers, wired terminals, etc.) for data communication with the building subsystem 428 or other external systems or devices. In various embodiments, communication over the interfaces 407, 409 can be direct (e.g., local wired or wireless communication) or can be over a communication network 446 (e.g., a WAN, the Internet, a cellular network, etc.). For example, the interfaces 407, 409 can include an Ethernet card and port for transmitting and receiving data over an Ethernet-based communication link or network. In another example, the interfaces 407, 409 can include a Wi-Fi transceiver for communicating over a wireless communication network. In another example, one or both of the interfaces 407, 409 can include a cellular phone communication transceiver or a mobile phone communication transceiver. In one embodiment, the communication interface 407 is a power plumbing communication interface and the BMS interface 409 is an Ethernet interface. In other embodiments, both the communication interface 407 and the BMS interface 409 are Ethernet interfaces or are the same Ethernet interface.

[0085] 4, the BMS controller 366 is shown to include a processing circuit 404 that includes a processor 406 and a memory 408. The processing circuit 404 may be communicatively coupled to a BMS interface 409 and / or a communication interface 407 such that the processing circuit 404 and its various components can transmit and receive data via the interfaces 407, 409. The processor 406 may be implemented as a general purpose processor, an application specific integrated circuit (ASIC), one or more field programmable gate arrays (FPGAs), a group of processing components, or other suitable electronic processing components.

[0086] Memory 408 (e.g., memory, memory unit, storage device, etc.) may include one or more devices (e.g., RAM, ROM, flash memory, hard disk storage, etc.) for storing data and / or computer code for completing or facilitating the various processes, layers, and modules described herein. Memory 408 may be or include volatile or non-volatile memory. Memory 408 may include database components, object code components, script components, or any other type of information structure for supporting the various activities and information structures described herein. According to some embodiments, memory 408 is communicatively coupled to processor 406 via processing circuitry 404 and includes computer code for executing (e.g., by processing circuitry 404 and / or processor 406) one or more processes described herein.

[0087] In some embodiments, the BMS controller 366 is implemented in a single computer (e.g., one server, one housing, etc.). In various other embodiments, the BMS controller 366 may be distributed across multiple servers or computers (e.g., may be in distributed locations). Additionally, while FIG. 4 shows applications 422 and 426 residing external to the BMS controller 366, in some embodiments, the applications 422 and 426 may be hosted within the BMS controller 366 (e.g., in memory 408).

[0088] 4, memory 408 is shown to include an enterprise integration layer 410, an automatic measurement and verification (AM&V) layer 412, a demand response (DR) layer 414, a fault detection and diagnosis (FDD) layer 416, an integrated control layer 418, and a later building subsystem integration 420. Layers 410-420 may be configured to receive inputs from building subsystems 428 and other data sources, determine optimal control actions for building subsystems 428 based on the inputs, generate control signals based on the optimal control actions, and provide the generated control signals to building subsystems 428. The following paragraphs describe some of the general functions performed by each of layers 410-420 in BMS 400.

[0089] The enterprise integration layer 410 can be configured to provide information and services to client or local applications to support various enterprise-level applications. For example, the enterprise control application 426 can be configured to provide a subsystem spanning control to a graphical user interface (GUI) or any number of enterprise-level business applications (e.g., accounting systems, user identification systems, etc.). The enterprise control application 426 can also, or alternatively, be configured to provide a configuration GUI for configuring the BMS controller 366. In yet other embodiments, the enterprise control application 426 can operate with layers 410-420 to optimize building performance (e.g., efficiency, energy use, comfort, or safety) based on inputs received at the interface 407 and / or the BMS interface 409.

[0090] The building subsystem integration layer 420 may be configured to manage communications between the BMS controller 366 and the building subsystems 428. For example, the building subsystem integration layer 420 may receive sensor data and input signals from the building subsystems 428 and provide output data and control signals to the building subsystems 428. The building subsystem integration layer 420 may also be configured to manage communications between the building subsystems 428. The building subsystem integration layer 420 translates communications (e.g., sensor data, input signals, output signals, etc.) across multiple multi-vendor / multi-protocol systems.

[0091] The demand response layer 414 can be configured to optimize resource usage (e.g., electricity usage, natural gas usage, water usage, etc.) and / or monetary costs of such resource usage in response to meeting demands of the building 10. The optimization can be based on time-of-use prices, curtailment signals, energy availability, or other data received from utility providers, distributed energy generation systems 424, energy storage 427 (e.g., high temperature TES 242, low temperature TES 244, etc.), or other sources. The demand response layer 414 can receive inputs from other layers of the BMS controller 366 (e.g., building subsystem integration layer 420, integrated control layer 418, etc.). Inputs received from other layers can include environmental or sensor inputs such as temperature, carbon dioxide levels, relative humidity levels, air quality sensor outputs, occupancy sensor outputs, room schedules, etc. The inputs can also include inputs such as electricity usage (e.g., expressed in kWh), heat load measurements, pricing information, forecasted pricing, smoothed pricing, curtailment signals from utilities, etc.

[0092] According to some embodiments, the demand response layer 414 includes control logic for responding to the data and signals it receives. These responses may include communicating with control algorithms in the integrated control layer 418, changing control strategies, changing set points, or activating / deactivating building devices or subsystems in a controlled manner. The demand response layer 414 may also include control logic configured to determine when to utilize stored energy. For example, the demand response layer 414 may determine to start using energy from the energy storage 427 just before the start of a peak usage period.

[0093] In some embodiments, the demand response layer 414 includes a control module configured to actively initiate control actions (e.g., automatically change set points) to minimize energy costs based on or based on one or more inputs representing demand (e.g., prices, curtailment signals, demand levels, etc.). In some embodiments, the demand response layer 414 uses equipment models to determine an optimal set of control actions. The equipment models may include, for example, thermodynamic models that describe the inputs, outputs, and / or functions performed by various sets of building equipment. The equipment models may represent collections of building equipment (e.g., subplants, chiller arrays, etc.) or individual devices (e.g., individual chillers, heaters, pumps, etc.).

[0094] The demand response layer 414 may further include or utilize one or more demand response policy definitions (e.g., databases, XML files, etc.). The policy definitions may be edited or adjusted by a user (e.g., via a graphical user interface) so that the control actions initiated in response to a demand input may be tailored to the user's application, desired comfort level, specific building equipment, or based on other concerns. For example, the demand response policy definitions may specify which equipment may be turned on or off in response to a particular demand input, how long a system or equipment should be turned off, what setpoints may be changed, what the allowable setpoint adjustment range is, how long to hold a high demand setpoint before returning to a normally scheduled setpoint, how close to a capacity limit, which equipment modes to utilize, energy transfer rates (e.g., maximum rates, alarm rates, other rate boundary information, etc.) to and from energy storage devices (e.g., thermal storage tanks, battery banks, etc.), and when to dispatch on-site generation of energy (e.g., via fuel cells, motor-generator sets, etc.).

[0095] The integrated control layer 418 can be configured to make control decisions using data inputs or outputs of the building subsystem integration layer 420 and / or subsequent demand response 414. The subsystem integration provided by the building subsystem integration layer 420 allows the integrated control layer 418 to integrate control activities of the subsystems 428 such that the subsystems 428 operate as a single integrated super system. In some embodiments, the integrated control layer 418 includes control logic that uses inputs and outputs from multiple building subsystems to provide greater comfort and energy savings compared to the comfort and energy savings that the separate subsystems could provide alone. For example, the integrated control layer 418 can be configured to use inputs from a first subsystem to make energy saving control decisions for a second subsystem. The results of these decisions can be communicated back to the building subsystem integration layer 420.

[0096] The integrated control layer 418 is shown logically below the demand response layer 414. The integrated control layer 418 can be configured to enhance the effectiveness of the demand response layer 414 by allowing the building subsystems 428 and their respective control loops to be controlled in coordination with the demand response layer 414. This configuration can advantageously reduce disruptive demand response behavior compared to conventional systems. For example, the integrated control layer 418 can be configured to ensure that a demand response, i.e., driving up adjustments to the setpoint of the chilled water temperature (or another component that directly or indirectly affects the temperature) does not result in an increase in fan energy (or other energy used to cool the space) that results in a greater total energy usage for the building than was saved in the chiller.

[0097] The integrated control layer 418 can be configured to provide feedback to the demand response layer 414 so that the demand response layer 414 ensures that constraints (e.g., temperature, lighting levels, etc.) are properly maintained even when a requested load shedding is in progress. Constraints may also include set points or sensed limits related to safety, equipment operating limits and performance, comfort, fire codes, electrical codes, energy codes, etc. The integrated control layer 418 is also logically below the fault detection and diagnostics layer 416 and the automatic measurement and verification layer 412. The integrated control layer 418 can be configured to provide calculated inputs (e.g., aggregates) to these higher levels based on outputs from two or more building subsystems.

[0098] The automatic measurement and verification (AM&V) layer 412 can be configured to verify that control strategies commanded by the integrated control layer 418 or the demand response layer 414 are operating properly (e.g., using data aggregated by the AM&V layer 412, the integrated control layer 418, the building subsystem integration layer 420, the FDD layer 416, or otherwise). The calculations performed by the AM&V layer 412 can be based on building system energy models and / or equipment models for individual BMS devices or subsystems. For example, the AM&V layer 412 may compare model predicted outputs to actual outputs from the building subsystems 428 to determine the accuracy of the models.

[0099] The fault detection and diagnosis (FDD) layer 416 can be configured to provide continuous fault detection for the building subsystems 428, building subsystem devices (i.e., building equipment), and control algorithms used by the demand response layer 414 and the integrated control layer 418. The FDD layer 416 may receive data inputs from the integrated control layer 418, directly from one or more building subsystems or devices, or from another data source. The FDD layer 416 may automatically diagnose and respond to detected faults. Responses to detected or diagnosed faults may include providing a user with a warning message, a maintenance scheduling system, or a control algorithm configured to attempt to repair or avoid the fault.

[0100] The FDD layer 416 can be configured to use detailed subsystem inputs available at the building subsystem integration layer 420 to output a specific identification of the faulty component or cause of the fault (e.g., loose damper interlocking). In other exemplary embodiments, the FDD layer 416 is configured to provide a "fault" event to the integrated control layer 418, which executes control strategies and policies in response to the received fault event. According to some embodiments, the FDD layer 416 (or policies executed by an integrated control engine or business rules engine) may shut down systems or direct control activities around the faulty device or system to reduce energy waste, extend equipment life, or ensure appropriate control responses.

[0101] The FDD layer 416 can be configured to store or access a variety of different system data stores (or data points for raw data). The FDD layer 416 may use the contents of some of the data stores to identify equipment level faults (e.g., a particular chiller, a particular AHU, a particular terminal unit, etc.) and other contents to identify component or subsystem level faults. For example, the building subsystem 428 may generate temporal (i.e., time series) data indicative of the performance of the BMS 400 and its various components. The data generated by the building subsystem 428 may include measurements or calculations that exhibit statistical characteristics and provide information about how a corresponding system or process (e.g., a temperature control process, a flow control process, etc.) is performing in terms of error from its set point. These processes can be inspected by the FDD layer 416 to expose when a system begins to degrade in performance and alert a user to repair the fault before the fault becomes more severe.

[0102] 5, a block diagram of another building management system (BMS) 500 is shown, according to some embodiments. The BMS 500 can be used to monitor and control the HVAC system 100, the waterside system 200, the airside system 300, the building subsystems 428, as well as other types of BMS devices (e.g., lighting devices, security devices, etc.) and / or HVAC equipment devices.

[0103] The BMS 500 provides a system architecture that facilitates automatic device discovery and device model distribution. Device discovery can occur on multiple levels of the BMS 500, across multiple different communication buses (e.g., system bus 554, zone buses 556-560 and 564, sensor / actuator bus 566, etc.) and across multiple different communication protocols. In some embodiments, device discovery is accomplished using active node tables that provide status information for devices connected to each communication bus. For example, each communication bus may be monitored for new devices by monitoring the corresponding active node table for new nodes. Once a new device is detected, the BMS 500 may begin interacting with the new device (e.g., sending control signals using data from the device) without user interaction.

[0104] Some devices in the BMS 500 are represented on the network using equipment models. Equipment models define equipment object attributes, view definitions, schedules, trends, and associated BACnet value objects (e.g., analog values, binary values, multi-state values, etc.) used for integration with other systems. Some devices in the BMS 500 have their own equipment models stored. Other devices in the BMS 500 have equipment models stored externally (e.g., in other devices). For example, the zone coordinator 508 may store an equipment model for the bypass damper 528. In some embodiments, the zone coordinator 508 automatically creates an equipment model for the bypass damper 528 or other devices on the zone bus 558. Other zone coordinators may also create equipment models for devices connected to the zone bus. Equipment models for devices may be automatically created based on the type of data points the device exposes on the zone bus, the device type, and / or other device attributes. Some examples of automatic equipment discovery and equipment model distribution are described in more detail below.

[0105] 5, the BMS 500 is shown to include a system manager 502, several zone coordinators 506, 508, 510, and 518, and several zone controllers 524, 530, 532, 536, 548, and 550. The system manager 502 can monitor data points within the BMS 500 and report the monitored variables to various monitoring and / or control applications. The system manager 502 can communicate with client devices 504 (e.g., user devices, desktop computers, laptop computers, mobile devices, etc.) via data communication links 574 (e.g., BACnet IP, Ethernet, wired or wireless communication, etc.). The system manager 502 can provide a user interface to the client devices 504 via the data communication links 574. The user interface can enable a user to monitor and / or control the BMS 500 via the client devices 504.

[0106] In some embodiments, the system manager 502 is connected to the zone coordinators 506-510 and 518 via a system bus 554. The system manager 502 can be configured to communicate with the zone coordinators 506-510 and 518 via the system bus 554 using a Master-Slave Token Passing (MSTP) protocol or any other communication protocol. The system bus 554 can also connect the system manager 502 to other devices such as a fixed capacity (CV) rooftop unit (RTU) 512, an input / output module (IOM) 514, a thermostat controller 516 (e.g., a TEC 5000 series thermostat controller), and a network automation engine (NAE) or third party controller 520. The RTU 512 can be configured to communicate directly with the system manager 502 and can be directly connected to the system bus 554. Other RTUs can communicate with the system manager 502 via intermediate devices. For example, the wired input 562 may connect the third party RTU 542 to a thermostat controller 516 that connects to the system bus 554 .

[0107] The system manager 502 can provide a user interface for any device that includes a device model. Devices such as the zone coordinators 506-510 and 518 and the thermostat controller 516 can provide their device models to the system manager 502 via the system bus 554. In some embodiments, the system manager 502 automatically creates device models for connected devices that do not include a device model (e.g., the IOM 514, the third party controller 520, etc.). For example, the system manager 502 can create a device model for any device that responds to a device tree request. The device models created by the system manager 502 can be stored within the system manager 502. The system manager 502 can then use the device models created by the system manager 502 to provide a user interface for devices that do not include their own device models. In some embodiments, the system manager 502 stores view definitions for each type of device connected via the system bus 554 and uses the stored view definitions to generate a user interface for the devices.

[0108] Each zone coordinator 506-510 and 518 can be connected to one or more of the zone controllers 524, 530-532, 536, and 548-550 via zone buses 556, 558, 560, and 564. The zone coordinators 506-510 and 518 can communicate with the zone controllers 524, 530-532, 536, and 548-550 via the zone buses 556-560 and 564 using the MSTP protocol or any other communication protocol. The zone buses 556-560 and 564 can also connect the zone coordinators 506-510 and 518 to other types of devices, such as variable air volume (VAV) RTUs 522 and 540, switched bypass (COBP) RTUs 526 and 552, bypass dampers 528 and 546, and PEAK controllers 534 and 544.

[0109] The zone coordinators 506-510 and 518 can be configured to monitor and command various zoning systems. In some embodiments, each zone coordinator 506-510 and 518 monitors and commands a separate zoning system and is connected to the zoning systems via a separate zone bus. For example, the zone coordinator 506 can be connected to the VAV RTU 522 and the zone controller 524 via a zone bus 556. The zone coordinator 508 can be connected to the COBP RTU 526, the bypass damper 528, the COBP zone controller 530, and the VAV zone controller 532 via a zone bus 558. The zone coordinator 510 can be connected to the PEAK controller 534 and the VAV zone controller 536 via a zone bus 560. The zone coordinator 518 can be connected to the PEAK controller 544, the bypass damper 546, the COBP zone controller 548, and the VAV zone controller 550 via a zone bus 564.

[0110] A single model of zone coordinators 506-510 and 518 can be configured to handle multiple different types of zoning systems (e.g., VAV zoning systems, COBP zoning systems, etc.). Each zoning system can include an RTU, one or more zone controllers, and / or bypass dampers. For example, zone coordinators 506 and 510 are shown as Verasys VAV Engines (VVEs) connected to VAV RTUs 522 and 540, respectively. Zone coordinator 506 is directly connected to VAV RTU 522 via zone bus 556, and zone coordinator 510 is connected to third party VAV RTU 540 via wired input 568 provided to PEAK controller 534. Zone coordinators 508 and 518 are shown as Verasys COBP Engines (VCEs) connected to COBP RTUs 526 and 552, respectively. The zone coordinator 508 is directly connected to the COBP RTU 526 via a zone bus 558 , while the zone coordinator 518 is connected to the third party COBP RTU 552 via a wired input 570 provided to the PEAK controller 544 .

[0111] Zone controllers 524, 530-532, 536, and 548-550 can communicate with individual BMS devices (e.g., sensors, actuators, etc.) via a sensor / actuator (SA) bus. For example, VAV zone controller 536 is shown connected to networked sensors 538 via SA bus 566. Zone controller 536 can communicate with networked sensors 538 using MSTP protocol or any other communication protocol. Although only one SA bus 566 is shown in FIG. 5, it should be understood that each zone controller 524, 530-532, 536, and 548-550 can be connected to a different SA bus. Each SA bus can connect the zone controllers with various sensors (e.g., temperature sensors, humidity sensors, pressure sensors, light sensors, occupancy sensors, etc.), actuators (e.g., damper actuators, valve actuators, etc.), and / or other types of controllable devices (e.g., coolers, heaters, fans, pumps, etc.).

[0112] Each zone controller 524, 530-532, 536, and 548-550 can be configured to monitor and control a different building zone. The zone controllers 524, 530-532, 536, and 548-550 can monitor and control the various building zones using inputs and outputs provided via their SA bus. For example, the zone controller 536 can use temperature inputs (e.g., the measured temperature of the building zone) received from the networked sensor 538 via the SA bus 566 as feedback in a temperature control algorithm. Zone controllers 524, 530-532, 536, and 548-550 may use various types of control algorithms (e.g., state-based algorithms, extremum-seeking control (ESC) algorithms, proportional-integral (PI) control algorithms, proportional-integral-derivative (PID) control algorithms, model predictive control (MPC) algorithms, feedback control algorithms, etc.) to control variable states or conditions (e.g., temperature, humidity, airflow, lighting, etc.) within or around building 10.

[0113] Variable Refrigerant Flow System 6A-6B, a variable refrigerant flow (VRF) system 600 is shown, according to some embodiments. The VRF system 600 is shown to include a number of outdoor VRF units 602 and a number of indoor VRF units 604. The outdoor VRF units 602 may be located outside of a building and may operate to heat or cool a refrigerant. The outdoor VRF units 602 may consume electricity to convert the refrigerant between a liquid phase, a vapor phase, and / or a superheated vapor phase. The indoor VRF units 604 may be distributed throughout various building zones within the building and may receive heated or cooled refrigerant from the outdoor VRF units 602. Each indoor VRF unit 604 may provide temperature control for the particular building zone in which the indoor VRF unit is located.

[0114] A major advantage of a VRF system is that some indoor VRF units 604 can operate in cooling mode and other indoor VRF units 604 can operate in heating mode. For example, each of the outdoor VRF units 602 and indoor VRF units 604 can operate in heating mode, cooling mode, or off mode. Each building zone can be independently controlled and can have different temperature setpoints. In some embodiments, each building has up to three outdoor VRF units 602 located on the outside of the building (e.g., on the rooftop) and up to 128 indoor VRF units 604 distributed throughout the building (e.g., various building zones).

[0115] There are many different configurations for the VRF system 600. In some embodiments, the VRF system 600 is a two-pipe system where each outdoor VRF unit 602 connects to a single refrigerant return line and a single refrigerant outlet line. In a two-pipe system, all of the outdoor VRF units 602 operate in the same mode, since only one of the heated or cooled refrigerants can be provided through a single refrigerant outlet line. In other embodiments, the VRF system 600 is a three-pipe system where each outdoor VRF unit 602 connects to a refrigerant return line, a hot refrigerant outlet line, and a cold refrigerant outlet line. In a three-pipe system, both heating and cooling can be provided simultaneously through dual refrigerant outlet lines.

[0116] VRF system 600 may represent an example of a VRF system that may utilize AI to predict oil quantity and viscosity to ensure that components (e.g., outdoor VRF unit 602, indoor VRF unit 604, etc.) are operating properly. Specifically, the components may require oil for proper lubrication to ensure that rapid degradation of the components is avoided. Specifically, VRF system 600 may utilize the systems and methods described throughout FIGS. 7-10 to ensure that all components have an adequate supply of oil and that the oil provided is of the proper viscosity.

[0117] Referring now to FIG. 7A, an illustration of a VRF system 700 is shown, according to some embodiments. In some embodiments, the VRF system 700 is similar and / or the same as the VRF system 600 described with reference to FIG. 6A and FIG. 6B. More specifically, the VRF system 700 can illustrate how oil is utilized in a VRF system. It should be noted that the VRF system 700 is shown merely for the purpose of an example of how a VRF system may operate. The components, relationships, and / or other features of the VRF system 700 can be customized and configured based on a particular implementation. For example, the VRF system 700 can include more or fewer compressors 701 than are shown in FIG. 7A.

[0118] The VRF system 700 is shown to include a compressor 701, a heat exchanger 702, a double-tube heat exchanger 703, an oil separator 704, and an accumulator 705. To operate properly, the compressor 701 may require oil to ensure that the components of the compressor 701 are properly lubricated. Without oil, the components may deteriorate quickly and the compressor 701 may not be able to provide adequate cooling / heating to the zones. The heat exchangers 702 and 703 may transfer heat between the fluids (e.g., oil and refrigerant). The oil separator 704 may separate the oil from the refrigerant and / or other fluids in the VRF system 700. Specifically, during operation, the compressor 701 may leak and / or otherwise allow some oil to be mixed into the refrigerant output by the compressor 701. If the oil is not back-extracted from the oil / refrigerant mixture, the oil may be provided to components beyond VRF system 700 (e.g., indoor AHUs) which may result in rapid loss of oil within VRF system 700. Thus, oil separator 704 can distill the oil from the fluid mixture and provide the oil to accumulator 705 for temporary storage. Accumulator 705 can store the oil and can be accessed as needed to recover oil for other components of VRF system 700.

[0119] The VRF system 700 is also shown to include a strainer 706, a distributor 707, a reversing valve 708, a capillary tube 709, and a microcomputer controlled expansion valve 710. The strainer 706 can remove impurities (e.g., dirt, debris, etc.) from the oil and / or oil / refrigerant mixture that may inadvertently become integrated with the oil and / or oil / refrigerant mixture during operation of the VRF system 700. The impurities can result in malfunction of building equipment and can affect the properties of the oil in the VRF system 700 (e.g., by increasing or decreasing the viscosity of the oil). The distributor 707 can assist in distributing the fluid throughout the heat exchanger 702. The reversing valve 708 can change the direction of the refrigerant flow in the VRF system 700 to switch the VRF system 700 between heating and cooling modes. The capillary tube 709 can assist in reducing the temperature of the refrigerant in the VRF system 700 by affecting the pressure of the refrigerant. A microcomputer controlled expansion valve 710 can regulate the amount of refrigerant entering the components of the VRF system 700 .

[0120] The VRF system 700 is also shown to include a check valve 711, a solenoid valve 712, a check joint 713, a stop valve 714 for the liquid piping, a stop valve 715 for the gas (cold) piping, a stop valve 716 for the gas (hot / cold) piping, a refrigerant pressure sensor 717, another refrigerant pressure sensor 718, and a high pressure switch 719. The check valve 711 can help ensure that fluid is flowing in the correct direction within the VRF system 700 by restricting the fluid from flowing in a direction opposite to the desired flow direction. The solenoid valve 712 can regulate the flow of fluid within the VRF system 700. The check joint 713 can help regulate the stress of the components of the VRF system 700. The stop valves 714, 715, and 716 can respectively restrict the flow of fluid within the liquid piping, the gas (cold) piping, and the gas (hot / cold) piping shown in the illustration of FIG. 7A. With regard to refrigerant pressure sensors 717 and 718, refrigerant pressure sensor 717 can be a high pressure sensor while refrigerant pressure sensor 718 can be a low pressure sensor in VRF system 700. When the refrigerant returns to compressor 701, high pressure switch 719 can stop refrigerant from entering compressor 701 if the refrigerant pressure is too high or too low to prevent damage to compressor 701.

[0121] VRF system 700 is also shown to include various thermistors. In VRF system 700, the resistance across the thermistors can be based primarily on the temperature of the connected components. In the illustration of FIG. 7A, VRF system 700 is shown to include thermistors 720-725. Thermistor 720 is associated with the upper side of the first compressor 701. Thermistor 721 is associated with the upper side of the second compressor 701. Thermistor 722 is associated with the gas side of the heat exchanger 702. Thermistor 723 is associated with the liquid side of the heat exchanger 702. Thermistor 724 is associated with the subcooler bypass side. Thermistor 725 is associated with the autocharge of refrigerant.

[0122] Each pipe in the VRF system 700 is also labeled with a corresponding outer diameter OD and thickness T, which is Table 1 It is noted that the material used in the VRF system 700 throughout all piping is C1220T-O. [Table 1]

[0123] 7B, an illustration of a VRF system 750 is shown, according to some embodiments. In some embodiments, the VRF system 750 is similar and / or the same as the VRF system 700 described with reference to FIG. 7 and / or the VRF system 600 described with reference to FIGS. 6A and 6B. In particular, the VRF system 750 illustrates the flow of oil through a VRF system. As with the VRF system 700, the VRF system 750 is provided for example purposes only. The components, structure, and / or other characteristics of the VRF system 750 can be customized and configured depending on the implementation.

[0124] The VRF system 750 is shown to include suction piping 752. The suction piping 752 can provide refrigerant (e.g., refrigerant vapor) to a compressor 754 for use by one or more devices / systems (e.g., an indoor AHU). In some embodiments, some oil can be included in the fluid provided to the compressor 754 by the suction piping 752. Based on the received refrigerant, the compressor 754 can operate to compress the refrigerant into a higher pressure gas and output the higher pressure gas via the discharge piping 756. Because the compressor 754 can require oil to function properly, the compression process performed by the compressor 754 can result in some oil being mixed into the output high pressure gas, thereby resulting in an oil / refrigerant mixture.

[0125] The VRF system 750 is also shown to include an oil separator 758. Based on the received oil / refrigerant mixture, the oil separator 758 can operate to separate the oil from the refrigerant. After separation, the refrigerant can be provided to a number of devices / systems (e.g., an indoor AHU) via refrigerant line 762. The separated oil can be provided to an accumulator 760 via oil line 764. The accumulator 760 can act as a storage container for the oil separated by the oil separator 758. The accumulator 760 can have a number of maximum capacities that define a maximum amount of oil that can be stored in the accumulator 760.

[0126] The oil stored by the accumulator 760 may be returned to the compressor 754 via an oil return line 766. Given that the accumulator 760 has some non-zero amount of stored oil, the accumulator 760 may provide oil to any of the compressors 754 if a particular compressor 754 requires more oil. In some embodiments, the VRF system 750 includes a valve 768 that regulates the flow of oil to the compressor 754. In this case, the valve 768 may prevent excess oil from being provided to the compressor 754 and / or may otherwise regulate the oil provided to the compressor 754.

[0127] In some embodiments, the viscosity of the oil in VRF system 750 and the amount of oil in each of compressor 754 and accumulator 760 are estimated / predicted by an AI model, which may take inputs such as the operating speed of compressor 754, the ambient temperature proximate VRF system 750, the discharge temperature and pressure of discharge line 756, the suction pressure of suction line 752, the temperature of the refrigerant vapor, and / or the temperature of other gases in VRF system 750. Such inputs may be measured by sensors (e.g., temperature sensors, pressure sensors, etc.) throughout VRF system 750 and / or may be provided directly by components of VRF system 750 (e.g., compressor 754 may directly output their operating speed).

[0128] Based on the inputs, the AI ​​model may predict characteristics of the oil in the VRF system 750 (e.g., oil volume, oil viscosity, etc.). If the oil characteristics do not meet a predefined threshold (or some other constraint), one or more corrective actions may be initiated to address the deficiency of the oil characteristics to meet the predefined threshold. For example, if the viscosity of the oil in the VRF system 750 is too high or too low, the corrective action may include introducing new oil into the VRF system 750 and / or completely replacing the oil in the VRF system 750 at a particular time. As another example, if the oil volume in a particular compressor 754 is too low, corrective measures may be initiated to operate a particular compressor 754, a particular valve 768, and / or an accumulator 760 such that the particular compressor 754 can obtain more oil from the accumulator 760 at a particular time resulting in a relatively low impact on the efficiency of the VRF system 750. The AI ​​models and corrective actions that may be initiated are described in more detail below with reference to FIGS. 8-10.

[0129] Systems and methods for oil quantity and viscosity estimation - Patents.com 8-10, systems and methods for estimating and predicting oil properties in a VRF system are shown and described, according to some embodiments. It should be understood that the following description is described with reference to a VRF system for illustrative purposes only and should not be considered limiting. The systems and methods described throughout FIGs. 8-10 may be similarly applied to various systems that utilize oil (e.g., other building systems, vehicle systems, etc.) and are not meant to be limited to VRF systems.

[0130] The systems and methods described below can utilize artificial intelligence (AI) models to predict how oil properties will change over time based on various inputs. The AI ​​models can include any suitable type of AI model. For example, the AI ​​models can be or include long short-term memory (LSTM) models, other types of recurrent neural networks (RNNs), convolutional neural networks (CNNs), etc. The type of AI model utilized can be selected based on, for example, the accuracy of a given AI model, what the specific inputs / outputs are to be considered, user preferences, etc. In some embodiments, RNN models, such as LSTM models, are preferred due to the time series nature of oil. It should be noted that machine learning models may be referred to herein as synonymous with AI models.

[0131] The AI ​​model can be trained to predict certain oil characteristics based on a set of training data. The training data can be provided by a variety of sources. For example, a user can provide a set of inputs including a variety of variables that can help the AI ​​model determine how oil is being utilized in a VRF system, and a corresponding set of outputs based on the actual measured operating conditions of the system or a similar system. In this case, the inputs can include, for example, the operating speed of the compressor, the ambient temperature in the vicinity of the compressor, the compressor discharge temperature, the compressor suction pressure, the compressor discharge pressure, the gas temperature of the gas in the compressor, etc. As defined herein, the compressor discharge temperature can refer to the temperature measurement of the superheated refrigerant vapor in the VRF system, the suction pressure can refer to the suction pressure generated by the compressor during operation, and the discharge pressure can refer to the pressure generated at the output side of the compressor. The outputs can include, for example, oil viscosity, the oil volume in the compressor, the oil volume in the accumulator, etc. The AI ​​model can then be trained to use the inputs and the corresponding outputs to predict the value of the outputs based on the inputs. Of course, the inputs and outputs are given for example purposes and are not meant to limit the possible inputs to the AI ​​model.

[0132] In some embodiments, a simulation model is utilized to generate training data used to train the AI ​​model. The training data generated using the simulation model may be used separately or in addition to training data collected from other sources (e.g., from measured conditions of the actual system). The simulation model may be constructed to simulate changes in the oil return system over time based on various conditions. In other words, the simulation model may be constructed to digitally mimic the operation of the oil return system. The state of the simulation model (e.g., oil volume, oil viscosity, energy consumption, etc.) may be manipulated to generate training data representing a wide variety of conditions. The simulation model may be run multiple times to generate training data representing the evolution of the system over time, under a variety of different loads, using different building devices, under different weather / environmental conditions, at different times, etc. With respect to a VRF system, the simulation model may be, for example, a closed-loop Functional Mock-up Unit (FMU) model that may typically be used to operate a VRF system if an AI model is not used. Advantageously, by utilizing a simulation model, a large amount of training data may be generated in a shorter period of time compared to operating an actual system over time to generate the training data. Furthermore, the simulation model can be run to generate training data representing fringe scenarios (e.g., dangerously high loads, unsafe operating conditions, device failures, etc.) without imposing conditions on the actual system that could be dangerous and disruptive to occupant comfort in the actual building.

[0133] Once trained based on a set of training data, the AI ​​model can predict oil characteristics based on the inputs. For example, the AI ​​model may predict the viscosity of the oil, the amount of oil in the compressor, and the amount of oil in the accumulator. Based on the predicted oil characteristics, it can be determined whether the predicted oil characteristics comply with predefined thresholds (and / or other constraints) on the values ​​of the characteristics. The predefined thresholds can include any limits on the values, such as, for example, a threshold above or below which the value of the characteristic must be, a range of acceptable values ​​of the characteristic, etc. If the value of the oil characteristic meets the predefined threshold, the VRF system can continue with standard operation. However, if the value of the oil characteristic does not meet the predefined threshold, a corrective action can be generated and initiated. A corrective action, as defined herein, can refer to any action taken to address one or more oil characteristics that do not meet some predefined constraint / threshold. For example, corrective actions may include generating and transmitting notifications to user devices, scheduling a technician to perform maintenance on the VRF system and / or change the oil, generating control signals and operating building equipment (e.g., VRF devices) based on the control signals, disabling certain building devices, automatically injecting new oil into the VRF system, recording the threshold violation in a database, etc. The corrective action to initiate may be determined based on various factors such as the threshold violated, the amount of the threshold violated (e.g., the difference between the actual value of the oil property and the threshold), user preferences, and / or other applicable considerations. A violation of a threshold for an oil property may also indicate a deficiency in the oil in that the oil lacks some desired characteristic (e.g., a desired viscosity). As described herein, a violation of a threshold may refer to a value (e.g., a predicted value) being above the threshold in the case of a maximum value threshold, or below the threshold in the case of a minimum value threshold.

[0134] Referring now to FIG. 8, a block diagram of an oil management controller 800 for predicting oil characteristics is shown, according to some embodiments. In particular, the oil management controller 800 may predict oil characteristics in a VRF system. However, the oil management controller 800 may be applied to various other systems / devices (e.g., other HVAC systems, car systems, etc.) that require oil to operate properly. In some embodiments, the oil management controller 800 and / or its components are incorporated into the BMS controller 366 described with reference to FIGS. 3-4 and / or another controller. In some embodiments, the oil management controller 800 is used to operate some and / or all of the VRF systems described throughout FIGS. 7A-7B.

[0135] The oil management controller 800 is shown to include a communication interface 808 and a processing circuit 802. The communication interface 808 may include a wired or wireless interface (e.g., a jack, an antenna, a transmitter, a receiver, a transceiver, a wired terminal, etc.) for data communication with various systems, devices, or networks. For example, the communication interface 808 may include an Ethernet card and port for transmitting and receiving data over an Ethernet-based communication network, and / or a Wi-Fi transceiver for communicating over a wireless communication network. The communication interface 808 may be configured to communicate over a local area network or a wide area network (e.g., the Internet, a building WAN, etc.) and may use various communication protocols (e.g., BACnet®, IP, LON®, etc.).

[0136] The communication interface 808 may be a network interface configured to facilitate electronic data communication between the oil management controller 800 and various external systems or devices (e.g., the devices 822, the sensors 820, the user devices 824, etc.). For example, the oil management controller 800 may receive device feedback from the devices 822 via the communication interface 808.

[0137] The processing circuit 802 is shown to include a processor 804 and a memory 806. The processor 804 may be a general purpose or special purpose processor, an application specific integrated circuit (ASIC), one or more field programmable gate arrays (FPGAs), a group of processing components, or other suitable processing components. The processor 804 may be configured to execute computer code or instructions stored in the memory 806 or received from another computer readable medium (e.g., a CDROM, a network storage device, a remote server, etc.).

[0138] The memory 806 may include one or more devices (e.g., memory units, memory devices, storage devices, etc.) for storing data and / or computer code for completing and / or facilitating various processes described in this disclosure. The memory 806 may include random access memory (RAM), read only memory (ROM), hard drive storage, temporary storage, non-volatile memory, flash memory, optical memory, or any other suitable memory for storing software objects and / or computer instructions. The memory 806 may include database components, object code components, script components, or any other type of information structure for supporting various activities and information structures described in this disclosure. The memory 806 may be communicatively connected to the processor 804 via the processing circuitry 802 and may include computer code for executing (e.g., by the processor 804) one or more processes described herein. In some embodiments, one or more components of the memory 806 are part of a single component. However, each component of the memory 806 is shown independently for ease of explanation.

[0139] Memory 806 is shown to include a training data collector 810. The training data collector 810 can collect training data used to train the artificial intelligence model from one or more training data sources 818. In particular, the training data collector 810 can obtain training data associated with characteristics of oil in the VRF system. In some embodiments, the training data collector 810 transmits queries to the training data sources 818 to obtain the training data. In some embodiments, the training data collector 810 can passively receive training data from the training data sources 818 without having to actively request the training data.

[0140] Training data source 818 may include any data source capable of storing and / or providing training data to training data collector 810. For example, training data source 818 may be or include a user device (e.g., laptop, desktop computer, mobile device, tablet, etc.) capable of providing a stored training data set to training data collector 810. As another example, training data source 818 may be or include a database (e.g., a cloud database) that stores data associated with utilizing a standard VRF plant model with additional outputs such as oil volume, oil viscosity, etc. In such an example, the VRF plant model may be a standard model used to operate a VRF system. In this manner, the training data may include measurements from a VRF system in actual operation along with measurements of oil properties.

[0141] In some embodiments, the training data collector 810 utilizes a simulation model to generate some or all of the training data used by the model generator 812 to generate the AI ​​model. The simulation model can model how the actual system may operate under various conditions (e.g., weather conditions, heating / cooling loads, device limitations, etc.) and how the oil in the system may be consumed and / or otherwise change over time. In this manner, the training data collector 810 may not need to retrieve training data from the training data source 818 and may instead generate the training data within the oil management controller 800. In some embodiments, the simulation model is hosted by a third-party controller / device / system (e.g., a cloud computing system) that can provide the oil management controller 800 with training data generated as a result of running the simulation model. In either case, the simulation model can be used / run to generate various training data representing various operating conditions of the oil-utilizing system in a shorter period of time than waiting for the actual system to generate training data through operation. Additionally, the simulation model can be run to generate training data illustrating fringe scenarios under which the actual system may be dangerous to operate simply to generate the training data.

[0142] Based on the acquired training data, the training data collector 810 can combine the collected training data into a training data set and provide the training data set to the model generator 812. Based on the training data set, the model generator 812 can generate an AI model that models the oil characteristics over time. Specifically, the model generator 812 can train the AI ​​model to predict the oil characteristics based on specified inputs. For example, the model generator 812 can train the AI ​​model to predict values ​​of oil viscosity, oil volume in one or more compressors, and oil volume in an accumulator based on inputs of compressor speed, ambient temperature, discharge temperature, suction pressure, discharge pressure, and gas temperature.

[0143] The AI ​​model generated by the model generator 812 can be any of a variety of AI model structures. In some embodiments, the AI ​​model is an RNN model, such as an LSTM model. In this case, the original FMU plant model can be utilized to generate enough simulation data for the RNN model to analyze and train on in order for the RNN to operate properly on the VRF system. As the training time increases, the final RNN model can function much closer to the original plant model. In particular, some advantages of using an RNN model are that it can be faster and have higher stability compared to the FMU plant model. Furthermore, the trained RNN model can reduce the effect of oil return and improve the efficiency of the VRF system. For the LSTM model in particular, the LSTM model is an artificial RNN used for deep learning. The LSTM model can classify and process entire sequences of time series data and can make predictions even with unknown duration lags between important events in the time series. The LSTM model generated by the model generator 812 can include various structures depending on the implementation. For example, the LSTM model generated by the model generator 812 can include one sequence input layer, one dropout layer, two fully connected layers, and two LSTM layers.

[0144] In some embodiments, the AI ​​model is a CNN model. In this case, the CNN model may include, for example, an input layer, multiple hidden layers (e.g., a normalized linear unit layer, a pooling layer, a fully connected layer, a normalization layer, etc.), an output layer, etc. In some embodiments, the AI ​​model follows some other artificial intelligence model architecture. An example architecture of the AI ​​model is described in more detail below with reference to Figures 9A and 9B.

[0145] The model generator 812 may utilize various training techniques to generate the AI ​​model. For example, the model generator 812 may utilize a stochastic gradient descent with momentum approach, an adaptive moment estimation approach, a root mean square propagation approach, etc. For the root mean square propagation approach, the model generator 812 may utilize the root mean square error (RMSE) to measure how accurate the model predictions are against the training data provided by the training data collector 810. Specifically, the model generator 812 may monitor the RMSE over time based on the following formula:

number

[0146] The model generator 812 can provide the generated AI model to a prediction generator 814. The prediction generator 814 can use the AI ​​model to generate a prediction of the oil properties over time. To generate the prediction, the prediction generator 814 can operate to obtain values ​​for inputs required by the AI ​​model from a variety of sources. For example, the prediction generator 814 may obtain device feedback from devices 822, measured variables from sensors 820, and / or any other suitable source of input values.

[0147] Unit 822 can be or include any device capable of providing values ​​for inputs required by the AI ​​model. For example, in a VRF system, unit 822 can include compressors, heat exchangers, accumulators, etc. More specifically, if the AI ​​model requires compressor speed as an input, unit 822 can include one or more compressors that can provide operating speed as unit feedback to prediction generator 814.

[0148] The sensors 820 may be or include various sensors capable of measuring values ​​of inputs (i.e., variables) required by the AI ​​model. For example, the sensors 820 may include pressure sensors that measure suction pressure and / or discharge pressure. As another example, the sensors 820 may include temperature sensors that measure ambient temperature, discharge temperature, and / or gas temperature.

[0149] Based on the AI ​​model and the obtained input values, the prediction generator 814 can generate an oil property prediction by passing the obtained input values ​​to the AI ​​model. As a result of passing the obtained input values ​​to the AI ​​model, the AI ​​model can output values ​​for one or more oil properties (e.g., oil viscosity, amount of oil in a compressor, amount of oil in an accumulator, etc.). In this manner, the properties of the oil in the VRF system can be estimated without the need for additional sensors to measure the oil properties.

[0150] In some embodiments, the prediction generator 814 generates predictions for oil at multiple stages in the VRF system. For example, the prediction generator 814 may generate a prediction of the viscosity of the oil-refrigerant mixture output by the compressor and may predict the separate viscosity of just the oil after the oil is separated from the oil-refrigerant mixture. In such an example, predicting the viscosity of the oil-refrigerant mixture and the viscosity of just the oil may be useful in determining how the oil may benefit from a particular corrective action (e.g., the amount of oil to add / remove from the VRF system). By generating predictions for oil at multiple stages in the VRF system, oil starvation can be predicted and tracked over time throughout the entire VRF system, rather than at a single point in the VRF system.

[0151] The prediction generator 814 can provide predictions of the oil characteristics to the corrective action generator 816. The corrective action generator 816 can analyze the predicted oil characteristics to determine whether and what corrective action should be initiated. As defined above, a corrective action can refer to any action taken to address an oil characteristic that does not meet some predefined threshold. The corrective action can include, for example, distributing a notification / alert to a user device 824 to indicate to a user that a particular oil characteristic violates a predefined threshold, operating the device 822 in oil return control, disabling the device 822, automatically scheduling maintenance activities to be performed on the device 822, recording the threshold violation in a database, etc. The predefined threshold can be defined by a user, provided by a manufacturer, estimated based on the operating conditions of the devices in the VRF system, etc. For example, a manufacturer can define a range of acceptable oil viscosity levels within which the compressor should be operated. In this case, excessively viscous oil and / or low viscosity oil can result in more rapid deterioration of the compressor. As another example, a user may define a minimum oil quantity threshold that defines a minimum allowable amount of oil in the accumulator. As should be appreciated, the thresholds defined for oil characteristics can come from a variety of sources (e.g., manufacturer, user, forecast-based, etc.) and can include a variety of limit types (e.g., ranges, thresholds, exact values ​​that should represent the characteristic, etc.).

[0152] As a more specific example, consider a scenario in which the AI ​​model predicts values ​​of oil properties including the amount of oil in the compressor, the amount of oil in the accumulator, the viscosity of the oil, and / or the viscosity of the oil-refrigerant mixture flowing through the compressor. In an embodiment, the corrective action generator 816 may determine whether the amount of oil in the compressor and the accumulator is above a first and second minimum threshold, respectively. If the amount of oil in the compressor is low, the corrective action generator 816 may determine that the compressor should operate in oil return control to recover oil from an external source (e.g., an indoor VRF system) and / or from the accumulator. As defined herein, oil return control may refer to an operating mode in which the compressor operates at a high speed to return oil from the outdoor system. Oil return control may result in high energy consumption and required heating / cooling may be temporarily withheld while the compressor operates to recover oil. If the amount of oil in the accumulator falls below a second minimum threshold, corrective action generator 816 may determine that the oil amount is too low and that more oil should be added to the VRF system. Adding oil may result from running the compressor on oil return control, may include a user manually adding new oil to the VRF system, etc.

[0153] Similarly, if the corrective action generator 816 determines that the viscosity of the oil or oil-refrigerant mixture is outside of a predefined range, the corrective action generator 816 may determine that oil should be added or removed from the VRF system. Specifically, if the viscosity of the oil-refrigerant mixture is above the maximum boundary of the range, the corrective action generator 816 may determine that oil should be removed from the VRF system (i.e., reduce the amount of oil in the oil-refrigerant mixture) to reduce the viscosity. If the viscosity of the oil-refrigerant mixture is below the minimum boundary of the range, the corrective action generator 816 may determine that oil should be added to the VRF system (i.e., increase the amount of oil in the oil-refrigerant mixture) to increase the viscosity. With respect to the oil itself, if the viscosity of the oil is above the maximum boundary of the range, the corrective action generator 816 may determine that oil should be added to the VRF system to reduce the viscosity if the new oil is expected to have a lower viscosity compared to the existing oil. If the viscosity of the oil is below the minimum boundary, the corrective action generator 816 may determine that oil should be removed to increase the viscosity.

[0154] In some embodiments, the corrective action generator 816 compares the output of the AI ​​model over time to determine whether a particular oil characteristic is approaching a threshold violation, thereby including a deficiency. In this case, the corrective action generator 816 may compare the value of the oil characteristic output by the AI ​​model to a previous output value of the oil characteristic. If a particular oil characteristic is trending toward violating a threshold, the corrective action generator 816 may preemptively initiate corrective action before a violation occurs. For example, if the viscosity of the oil is increasing over time and will exceed an upper viscosity limit within a future time period based on the current trend, the corrective action generator 816 may initiate corrective action before the viscosity of the oil exceeds the maximum limit. Advantageously, preemptive initiation of corrective action can ensure that an equipment (e.g., a compressor) is operated under conditions associated with a violation of an oil characteristic threshold. Reducing such time can, in turn, reduce equipment degradation, reduce energy consumption, and provide other benefits.

[0155] In some embodiments, the corrective action generator 816 predicts the time to initiate a particular corrective action to reduce the impact on the equipment 822 and / or other devices / systems. For example, the corrective action generator 816 may predict the time to initiate oil return control to reduce the negative impact on the heating / cooling load required by the building. As described above, operating a compressor under oil return control may temporarily postpone the required heating / cooling during the oil return control. Thus, the corrective action generator 816 may predict the time when the impact on the required heating / cooling may be reduced (e.g., minimized). As another example, if new oil needs to be introduced into the system, the corrective action generator 816 may predict the time to temporarily disable the equipment 822 so that the new oil can be safely added to the system by an individual.

[0156] The corrective action generator 816 may utilize various techniques to predict the time to initiate a particular corrective action. For example, the corrective action generator 816 may track a particular variable over time and identify a lower range of values ​​that may result in a low amount of disruption to the system. As a particular example, the corrective action generator 816 may identify a range of values ​​associated with low heating / cooling needs such that the impact on environmental conditions within the building is reduced. Based on the identified range of values, the corrective action generator 816 may track the actual heating / cooling needs over time and, in response to identifying a period during which the actual heating / cooling needs are within the identified range, may initiate a corrective action during that period. In this manner, the corrective action generator 816 is effectively predicting a period during which initiating a corrective action will result in a low overall impact.

[0157] In some embodiments, the corrective action generator 816 can operate as a standard equipment controller when no corrective action is required (e.g., when the oil viscosity and oil amount are at the proper values). In other words, the corrective action generator 816 can generate a control signal for the equipment 822 to operate the equipment 822 to affect some variable state or condition (e.g., temperature, humidity, etc.) in the building. In some embodiments, the corrective action generator 816 can be configured to set boundary conditions for the equipment 822 based on the prediction provided by the prediction generator 814. For example, the corrective action generator 816 can set a maximum speed of the compressor based on a prediction of the oil viscosity. In that example, if the viscosity is within the proper range and is not approaching a range violation, the corrective action generator 816 can generate a control signal to operate the compressor at a higher speed because the oil viscosity is proper.

[0158] Any violation of the thresholds for the oil properties can be addressed by initiating some corrective action, which can ensure that the time during which the device 822 is operating under conditions associated with some oil property violation is reduced (e.g., minimized). Overall, in a VRF system, using predictions based on an AI model to initiate corrective actions can be beneficial in saving hardware costs, reducing the impact of oil return, and improving the efficiency of the VRF system.

[0159] 9A, an illustration of a recurrent neural network (RNN) structure 900 is shown, according to some embodiments. In particular, the RNN structure 900 may illustrate the structure of an RNN model (e.g., an LSTM model) that may be generated and utilized as an AI model as described above with reference to FIG.

[0160] RNN is a class of artificial neural networks in which connections between nodes form a directed graph along a time sequence. For a VRF system, an RNN model represented by the RNN structure 900 can be generated using simulation data collected based on an FMU plant model. As the training time increases, the RNN model can function more and more closely to the original plant model for a VRF system.

[0161] The RNN structure 900 illustrates a condensed network structure and how the condensed network structure can be "unfolded" to illustrate how the RNN structure 900 operates over a temporal sequence. Specifically, the condensed ("folded") structure and the unfolded structure are equivalent, and the unfolded structure illustrates in more detail the use of an RNN model over a temporal sequence.

[0162] The RNN structure 900 is shown to include inputs represented as x, which may be a vector containing inputs required by the RNN model. For a VRF system, the input vector x may include, for example, compressor speed, ambient temperature, discharge temperature, suction pressure, discharge pressure, gas temperature, etc. A weight vector U may be applied to x, and the results may be provided to the hidden layer vector h. Similarly, a weight vector V may be applied to the hidden layer vector of the previous time step. Based on the weighted inputs, and the weighted values ​​of the previous hidden layer vector, a function may be applied to determine a corresponding output that may result in output o after the weight vector W is applied. This process may be repeated for each time step in the temporal sequence. In other words, a new input vector x t can be obtained for time step t, and x t , previous state h t-1 , and the corresponding weight vectors U, V, and W, the output vector o t can be determined for time step t.

[0163] As a result of incorporating the RNN structure 900 into the RNN model generated and used by the oil management controller 800, the predictions of the RNN model may be modified over time as a result of previous time steps. As oil properties change over time as a result of changing conditions (e.g., changing environmental conditions, operating conditions, etc.), it may be useful to utilize the RNN model in particular because of its unique ability to account for changes over a time sequence, as opposed to being limited by the original training process, as is the case with some other neural network architectures.

[0164] 9B, an example of a neural network (NN) architecture 950 is shown, according to some embodiments. The NN architecture 950 may describe a general architecture that may be utilized by the AI ​​model described above with reference to FIG. 8 for a VRF system (e.g., VRF system 600). Specifically, the NN architecture 950 may illustrate how a neural network may generate a set of outputs based on a set of inputs associated with a VRF system. However, it should be noted that the NN architecture 950 is provided merely for the purpose of an example of a neural network architecture that may be utilized, and is not intended to be limiting to the neural network architecture that may be utilized by the AI ​​model described with reference to FIG. 8.

[0165] The NN architecture 950 is shown to receive compressor speed, ambient temperature, discharge temperature, suction pressure, discharge pressure, and gas temperature as inputs. Each input may be associated with a specific input node of an input layer in the NN architecture 950. In other words, some nodes in the input layer may correspond to some actual inputs as a one-to-one relationship. It should be understood that the inputs shown in FIG. 9B are provided for example purposes only. The NN architecture 950 may be modified to account for various different inputs depending on the implementation. For example, if gas temperature is not considered as an input, the input layer may include only five input nodes.

[0166] The NN architecture 950 is also shown to include a hidden layer that includes hidden nodes. In the NN architecture 950, the hidden layer is shown to include a single layer that includes a number of hidden nodes that correspond to the number of input nodes of the input layer. However, it should be noted that according to various embodiments, the hidden layer can include one or more layers that include a different number of hidden nodes that may or may not correspond to the number of input nodes. For example, in a convolutional neural network architecture, the NN architecture 950 can include multiple hidden layers (e.g., multiple convolutional layers) with a different number of hidden nodes. Furthermore, the nodes of each layer do not necessarily need to connect to all nodes of adjacent layers, as shown in FIG. 9B.

[0167] In the NN architecture 950, a weight W may be applied with respect to a connection between two nodes. In some embodiments, each connection between nodes includes a specific value for the particular connection. For example, the weight between input node 1 in the input layer and hidden node 1 in the hidden layer may be different from the weight between input node 1 and hidden node 2 in the hidden layer. In some embodiments, various connections between nodes may be associated with the same weight. For example, in an LSTM specific architecture, the weights associated with connections between input nodes and hidden nodes may be the same.

[0168] Based on each weighted value input to a particular node, a function may be applied to determine a composite value for the node. For example, for hidden node 1 of NN architecture 950, a function may be applied to the weighted input values ​​input to the node to determine a composite value for hidden node 1. The composite value of each node in a particular layer to determine the output of the particular layer. The output of a particular layer may correspond to the input to the subsequent layer along with the weights between the particular layer and the subsequent layer. This process may be repeated for each layer until the output layer is reached.

[0169] The NN architecture 950 is also shown to include an output layer including output nodes. Some output nodes in the output layer may correspond one-to-one to desired outputs of the NN model. For VRF systems in particular, the outputs may include oil viscosity, oil-refrigerant mix viscosity, oil volume in one or more compressors, and oil volume in an accumulator. Thus, the first, second, and third output nodes may correspond to oil viscosity, oil volume in one or more compressors, and oil volume in an accumulator, respectively. For the NN architecture 950, the composite value of output node 1 may correspond to oil viscosity, the composite value of output node 2 may correspond to oil volume in one or more compressors, and the composite value of output node 3 may correspond to oil volume in an accumulator. In this manner, predicted values ​​of the outputs may be generated by simply providing input values ​​to the NN architecture 950.

[0170] Referring now to FIG. 10, a flow diagram of a process 1000 for monitoring oil properties using an AI model is shown, according to some embodiments. The process 1000 can utilize the AI ​​model to predict values ​​of oil properties and initiate corrective action if the values ​​do not meet a predefined threshold. Although the process 1000 is described primarily with reference to building systems (e.g., VRF systems), the process 1000 can be applied to a variety of systems that include components / devices that require oil for proper operation. For example, the process 1000 can be applied to VRF systems, other HVAC systems, car systems, etc. In some embodiments, some and / or all steps of the process 1000 are performed by the oil management controller 800 described with reference to FIG. 8.

[0171] Process 1000 is shown to include obtaining training data describing relationships between conditions that affect oil used by building equipment of a building and properties of the oil (step 1002). The building equipment can include various devices that can affect variable states or conditions of the building and utilize oil for proper operation. For example, the building equipment can include compressors, AHUs, other sub-plants, etc. The training data can be obtained from various sources. For example, the training data can be obtained via direct input from a user, by accessing a database (e.g., a cloud database) that stores historical information associated with the operation of the building equipment, by using training data provided by a manufacturer of the building equipment, etc. In some embodiments, step 1002 includes generating the training data using a simulation model. If a simulation model is used, the simulation model can generate some and / or all of the training data obtained in step 1002. The simulation model can be configured to consider various aspects of the system, including the building equipment, such as, for example, the amount of oil used by the building equipment's devices during operation, how external weather conditions and / or other ambient conditions affect the system, various heating / cooling loads of the building, etc. During the generation of training data, variables associated with the simulation model can be manipulated to generate training data representing various scenarios. Using a simulation model in step 1002 can result in more training data being available in a shorter time compared to collecting data based on actual operation of the building equipment. Furthermore, using a simulation model in step 1002 can help obtain training data that describes fringe cases that may not typically be included in training data collected based on actual device operation. In some embodiments, step 1002 is performed by the training data collector 810.

[0172] Process 1000 is shown to include generating an artificial intelligence (AI) model based on the training data for modeling the oil characteristics (step 1004). The AI ​​model generated in step 1004 can be a variety of different AI models, such as, for example, an RNN model (e.g., an LSTM model), a CNN model, etc. The AI ​​model can be generated to relate conditions that affect the oil and the oil characteristics themselves. Specifically, the AI ​​model can be trained to relate certain inputs (e.g., compressor speed, ambient temperature, discharge temperature, suction pressure, discharge pressure, gas temperature, etc.) to certain outputs (e.g., oil viscosity, amount of oil in the compressor, amount of oil in the accumulator, etc.). In some embodiments, step 1004 can include training weights associated with connections between nodes of the AI ​​model to account for relationships between conditions and oil characteristics. In some embodiments, step 1004 is performed by the model generator 812.

[0173] Process 1000 is shown to include using the AI ​​model to generate a prediction of the oil property over time based on a set of model inputs (step 1006). As discussed above in step 1004, the AI ​​model can be trained to associate particular inputs with particular outputs. Thus, once trained, the AI ​​model can utilize values ​​of the inputs to predict corresponding values ​​of the outputs (i.e., oil properties). In some embodiments, step 1006 is performed by prediction generator 814.

[0174] Process 1000 is shown to include determining whether the prediction violates any constraints (step 1008). In some embodiments, the constraints are predefined constraints that define acceptable values ​​for the oil properties. For example, the constraints may include thresholds that the oil property values ​​should be above / below, acceptable ranges of values ​​that the oil property values ​​should be within, etc. Specifically, the constraints may be thresholds that should not be violated. As a specific example, a constraint on oil viscosity may be defined as a range of acceptable values ​​that the oil viscosity may be within. In that example, a violation may be identified if the predicted oil viscosity is above the maximum value of the range or below the minimum value of the range. If the predicted oil properties do not violate any constraints and therefore do not have a defect (step 1008, "No"), process 1000 may repeat starting from step 1006. In this case, a new set of predictions may be generated for subsequent time steps so that the oil properties can be monitored / tracked over time. However, if a constraint violation is identified (step 1008, “Yes”), process 1000 may proceed to step 1010. In some embodiments, a single constraint violation results in process 1000 proceeding to step 1010. In some embodiments, multiple constraint violations (e.g., two constraint violations, three constraint violations, etc.) may be required for process 1000 to proceed to step 1010. In some embodiments, step 1008 includes at least partially considering the severity of the particular constraint violation in determining whether to proceed to step 1010. For example, an oil viscosity exceeding a maximum viscosity of 0.001 Pascal-second may also require several other constraints to be violated in order for process 1000 to proceed to step 1010, whereas an oil viscosity exceeding a maximum viscosity of 1 Pascal-second may be sufficient on its own for process 1000 to proceed to step 1010. In some embodiments, step 1008 may include predicting whether an oil property will violate a constraint in a future time period, and if so, causing process 1000 to proceed to step 1010 to preemptively address the anticipated violation. In some embodiments, step 1008 is performed by corrective action generator 816.

[0175] Process 1000 is shown to include determining a corrective action based on which oil characteristic violated a constraint (step 1010). In other words, a corrective action may be determined to address the particular oil characteristic that is violating one or more constraints / thresholds. For example, if the viscosity of the oil violates a constraint (e.g., maximum allowable viscosity), the determined corrective action may be to transmit a notification to a user device to notify the user that servicing of the building equipment may be necessary to reduce the oil viscosity. As another example, if the oil amount in the compressor falls below a minimum threshold, the corrective action determined in step 1010 may be to operate the compressor on oil return control to recover more oil (e.g., from an accumulator, from an indoor VRF system, etc.). In some embodiments, step 1010 includes determining a particular time and / or period for the corrective action to occur. To determine the particular time and / or time period, step 1010 may include monitoring system conditions (e.g., device operating conditions, ambient conditions, etc.) to determine the time that will have the least impact on cost, heating / cooling efficiency, etc. In some embodiments, if the corrective action is transmitting a notification or if the constraint violation is severe, the determined time and / or time period may be the earliest possible time (e.g., immediately). In some embodiments, step 1010 is performed by corrective action generator 816.

[0176] Process 1000 is shown to include initiating a corrective action (step 1012). By initiating the corrective action determined in step 1010, one or more constraint / threshold violations identified in step 1008 may be addressed. In this manner, the overall time that one or more constraint / threshold violations are active may be reduced. Reducing the time that a constraint / threshold is violated may be beneficial in reducing deterioration of building equipment, helping to reduce costs (e.g., energy costs), increasing overall safety of the system, etc. In some embodiments, if step 1010 includes determining when a corrective action should be performed, step 1012 may include initiating the corrective action at the determined time. In some embodiments, step 1012 is performed by corrective action generator 816.

[0177] Experimental Results 11A-12C, results of an exemplary experiment are shown, according to some embodiments. The exemplary experiment in FIG. 11A-12C is provided for illustrative purposes only and is not intended to limit the present disclosure, but rather to demonstrate the utility of utilizing an AI model in predicting oil properties. The AI ​​model referenced below through FIG. 11A-12C is an RNN model trained for the purpose of predicting oil properties.

[0178] 11A and 11B, a pair of graphs illustrating the results of an AI model training process for an example experiment, according to some embodiments, are shown. FIG. 11A is shown to include a graph 1100 illustrating the change in RMSE based on several iterations of the example training process. FIG. 11B is shown to include a graph 1150 illustrating the change in loss based on the number of iterations. The example training process associated with FIG. 11A and 11B utilized 10 closed-loop test cases from a VRF model-based definition (MBD) oil viscosity plant model, with approximately 4000 seconds allocated to each test case, as training data. To determine the accuracy of the AI ​​model, a single test data set from the VRF MBD oil viscosity plant model, with approximately 4000 seconds allocated to each test case, was used for comparison.

[0179] Graph 1100 is shown to include a series 1102. Series 1102 can illustrate how the RMSE associated with an AI model changes as a result of additional iterations of a training process. Specifically, series 1102 illustrates a generally decreasing trend as the number of iterations increases. In other words, increasing the number of iterations can improve the accuracy of the AI ​​model. Note that series 1102 represents a smoothed curve of RMSE data points collected at each iteration.

[0180] Graph 1150 is shown to include series 1152. Series 1152 can illustrate how a loss associated with an AI model changes over time based on several iterations. In this case, the loss describes how inaccurate the AI ​​model's predictions are, with a loss of 0 indicating that a particular prediction is equal to the actual measurement. As is evident from series 1152 and series 1102, the accuracy of the AI ​​model for predicting oil properties increases based on the number of iterations.

[0181] 12A-12C, graphs are shown illustrating oil property predictions of an AI model trained in the exemplary experiment of FIGURES 11A-11B, according to some embodiments. In the exemplary experiment, the AI ​​model took inputs of compressor speed, ambient temperature, discharge temperature, suction pressure, discharge pressure, and gas temperature to generate predicted outputs of oil viscosity, oil volume in compressor, and oil volume in accumulator.

[0182] 12A, a graph 1200 is shown illustrating an AI model's prediction of the amount of oil in a compressor, according to some embodiments. The graph 1200 is shown to include a series 1202 illustrating the predicted amount of oil in the compressor over time. In some embodiments, a goal of operational decisions associated with the compressor may be to maintain a relatively constant value of the amount of oil in the compressor to ensure stable and reliable operation.

[0183] 12B, a graph 1210 is shown illustrating an AI model's prediction of the amount of oil in the accumulator, according to some embodiments. The graph 1210 is shown to include a series 1212 illustrating the predicted amount of oil in the accumulator over time. A change in series 1212 may result from, for example, a change in the operation of the compressor. Specifically, the compressor may provide an oil / refrigerant mixture in which the oil of the oil / refrigerant mixture is separated to be collected by the accumulator, which may result in a predicted increase in series 1212. Alternatively, the compressor may operate to withdraw oil from the accumulator, which may result in a decrease in series 1212.

[0184] 12C, a graph 1220 illustrating an AI model's prediction of oil viscosity, according to some embodiments, is shown. Oil viscosity can change depending on external temperature, how the compressor utilizes the oil, etc. Thus, the AI ​​model can predict how oil viscosity will change over time based on how the overall system that utilizes the oil changes.

[0185] Leveraging AI for VRF system management 13A-24, systems and methods for utilizing AI to generate predictions regarding characteristics of a VRF system are shown and described, according to some embodiments. In particular, the systems and methods described below illustrate how AI can be utilized to predict vibrations of a compressor in a VRF system, fault conditions in the VRF system, and efficiency of a motor in the compressor. Based on the AI ​​predictions, a determination can be made as to whether any characteristics of the VRF system and / or its components are operating outside of preferred operating limits. As described herein, corrective action can refer to any action taken to address some predicted characteristic of the VRF system, and / or one or more components therein, exceeding a predefined tolerance (e.g., a predefined threshold). For example, if an AI model predicts that a fault condition exists for the VRF system, a corrective action can be initiated to automatically schedule maintenance to correct the fault condition. As another example, if the AI ​​predicts that the vibration rate of the compressor is exceeding a predefined maximum threshold (e.g., maximum operating cycles per minute), a corrective action that may be initiated may be to reduce the input current provided to the compressor and / or to temporarily disable the compressor.

[0186] As described herein, the term "AI" may be used to refer to various different types of AI models used to generate predictions. In particular, the systems and methods described herein may utilize neural networks to generate predictions associated with a VRF system. For example, a recurrent neural network (RNN), such as a long short-term memory (LSTM) network, may be used to generate predictions. An exemplary structure of an LSTM is described in more detail below with reference to Figures 25A and 25B. Of course, it should be understood that other types of neural networks and / or different AI models may be utilized in generating predictions associated with a VRF system. For example, a convolutional neural network (CNN), a multi-layer perceptron, etc. may be utilized in generating the predictions. In some embodiments, multiple AI models may be trained to generate a particular prediction and tested for accuracy of the prediction against a known dataset. Based on the determined accuracy of each AI model, the AI ​​model that generates the closest prediction to the known dataset may be selected and utilized.

[0187] As described in more detail below, the AI ​​model utilized to generate the predictions can be trained based on a variety of different data sources. In some embodiments, the AI ​​model is trained based on real data collected from one or more operational VRF systems. In some embodiments, the AI ​​model is trained based on data generated based on a simulated environment, where the simulated environment may be configured to mimic an operational, actual VRF system. Advantageously, using a simulation to generate training data may generate a larger set of training data than could otherwise be generated by an operational, actual VRF system. In some embodiments, the AI ​​model is trained based on a mix of training data generated by simulation, collected from an actual VRF system, and / or from some other source of training data.

[0188] By utilizing AI models to generate predictions associated with the VRF system, issues within the VRF system can be quickly identified and addressed. Furthermore, AI models can help accurately identify possible issues within the VRF system before they become more serious issues (e.g., complete failure of a building device). In this way, AI models can:

[0189] Compressor vibration prediction 13A-18, systems and methods for managing vibrations of a compressor in a VRF system are shown and described, according to several embodiments. However, it should be understood that the following description is not necessarily limited to compressors in VRF systems. The systems and methods described below may be applied to a variety of different compressors in a variety of systems (e.g., different HVAC systems) and / or different building equipment.

[0190] Compressor vibration can result in rapid deterioration of the compressor. As the frequency and / or intensity of the vibration increases, the compressor can become increasingly susceptible to failures and other problems that may require maintenance and / or complete replacement of the compressor. However, complete elimination of vibration may not be practical since compressors will vibrate to some degree during operation. Thus, compressor operation should be balanced between reducing vibration while meeting the load and / or other necessary needs of the VRF system.

[0191] As referred to herein, "compressor" may be used generally to refer to a variety of different types of compressors that may be utilized in a VRF system. For example, a VRF system may include a single rotary compressor, a twin rotary compressor, a scroll compressor, etc. In a VRF system, the compressor may be operated based on an alternating current (AC) provided by an inverter. Specifically, the inverter may change the frequency of the AC provided to the compressor to change the speed at which the compressor's motor rotates and may change the amplitude of the AC to change the torque applied by the motor.

[0192] Because the AC provided by the inverter directly impacts the operation of the compressor, a correlation can be identified between the characteristics (e.g., frequency and / or amplitude) of the AC and compressor vibration. For example, the frequency of vibration may increase as the frequency and / or strength (i.e., amplitude) of the AC increases. Based on the identified relationship, an AI model can be trained to learn which values ​​of the current characteristics (e.g., amplitude) can result in reduced vibration while still satisfying the heating / cooling load and / or other requirements of the compressor. In some embodiments, as described in more detail below, the AI ​​model can be specifically trained to predict appropriate values ​​of AC amplitude to modify the torque applied by the motor to manage (e.g., reduce) vibration.

[0193] In some embodiments, the AI ​​model configured to predict the appropriate current value for the compressor utilizes noise as a proxy for vibration. In this manner, the AI ​​model can be trained to predict the appropriate amplitude of AC provided to the compressor that reduces the amount of noise generated by the compressor. Using noise as a proxy for vibration may be appropriate as a result of a known relationship between noise and vibration. In general, louder noise may indicate increased vibration, while quieter noise may indicate less vibration. More specifically, there may be a relationship between the frequency of the compressor's vibration in decibels (dB) and Hertz (Hz). For example, it may be determined that a 1 dB increase in noise correlates with a 1 Hz increase in vibration frequency. Advantageously, using noise as a proxy for vibration may reduce overall costs, as sensors for directly measuring vibration may be significantly more expensive compared to sensors for measuring sound (e.g., standard audio sensors).

[0194] 13A, a block diagram of a VRF system 1300 is shown, according to some embodiments. As will be appreciated, the VRF system 1300 shown in FIG. 13A illustrates a portion of a larger VRF system that utilizes a compressor to provide heating / cooling to a building. For example, the VRF system 1300 may be a subsystem of the VRF system 600 described with reference to FIGs. 6A and 6B.

[0195] The VRF system 1300 is shown to include a converter 1304 and an inverter 1306. The converter 1304 can receive alternating current (AC) power from an AC power source (not shown) and can convert the AC power to direct current (DC) power. The converter 1304 can provide the converted DC power to the inverter 1306. The inverter 1306 can be an electronic modulator that changes the frequency of a received DC signal and outputs an AC signal to the compressor 1302. In particular, the inverter 1306 can operate to manipulate the frequency and / or amplitude of the AC signal such that the output AC signal causes the motor 1308 of the compressor 1302 to operate at a particular speed and torque.

[0196] In some embodiments, the frequency of the AC signal provided by the inverter 1306 is related to a known amount of fluid passing through the compressor 1302 and / or a known amount of fluid expected to pass through the compressor 1302. The frequency of the AC signal can indicate how fast the motor 1308 should and / or is rotating to ensure an appropriate amount of fluid flows through the compressor 1302, which may be measured in revolutions per minute. The frequency of the AC signal can thereby be determined based on the heating / cooling load of the building and / or other object. For example, a higher cooling load required for the building may result in a higher frequency of the AC signal, thereby causing the motor 1308 to rotate faster to pass more fluid through the compressor 1302. The frequency of the AC signal can be determined by a controller, by the inverter 1306 itself, based on feedback from the compressor 1302, etc.

[0197] To determine what amplitude (also referred to herein as intensity) of AC signal to output, the inverter 1306 may receive an AC amplitude setpoint from a compressor vibration controller 1310. In some embodiments, the compressor vibration controller 1310 utilizes an AI model to predict what amplitude value (e.g., in volts) should be provided to the motor 1308 to ensure that the compressor 1302 does not degrade too quickly as a result of vibration. Based on the received AC amplitude setpoint, the inverter 1306 may modulate the DC signal received from the converter 1304 to output an AC signal to the compressor 1302 that matches (or nearly matches) the AC amplitude setpoint provided by the compressor vibration controller 1310, as well as the desired frequency as described above. The compressor vibration controller 1310 is described in more detail below with reference to FIG. 16.

[0198] The VRF system 1300 is also shown to include a compressor 1302. The compressor 1302 can be any type of compressor used to compress a fluid (e.g., single rotary compressor, twin rotary compressor, scroll compressor, etc.). The compressor 1302 can be configured to take in a fluid (e.g., gas) via a suction line, compress the fluid, and output the compressed fluid via a discharge line. As described above, the compressor 1302 can include a motor 1308 that drives components of the compressor 1302 to compress the fluid and move the fluid through the compressor 1302. The motor 1308 can operate based on an AC signal (current) received from an inverter 1306. In particular, the frequency of the AC signal can affect the operating speed of the motor 1308.

[0199] In conventional systems, the current provided by the inverter 1306 may be a predetermined stagnation value that does not take into account vibrations of the compressor 1302. These conventional systems may result in the compressor 1302 quickly deteriorating because the compressor 1302 is vibrating at an unsafe frequency and / or intensity. However, as described in more detail below with reference to FIG. 16, the compressor vibration controller 1310 may predict an appropriate value for the amplitude of the AC signal provided by the inverter 1306 so that the vibrations of the compressor 1302 may be managed (e.g., reduced). In this manner, the inverter 1306 may consistently provide the compressor 1302 with a current determined as a function of possible vibrations to increase the life of the compressor 1302.

[0200] 13B, a block diagram of a VRF system 1300 is shown in more detail, according to some embodiments. The motor 1308 can be any type of motor that operates based on an AC signal. For example, the motor 1308 can be a single-phase motor that operates based on a single-phase power source. As another example, and as shown in FIG. 13B, the motor 1308 can be a three-phase motor that operates based on a three-phase power source.

[0201] As discussed above with reference to Figure 13A, the inverter 1306 can be configured to provide an AC signal to the motor 1308 based on an AC amplitude setpoint provided by the compressor vibration controller 1310. As shown in Figure 13B, the signal provided to the motor 1308 can be a three-phase signal including phases A, B, and C. In this case, the compressor vibration controller 1310 can generate a prediction for the three-phase signal required to properly operate the motor 1308. Each of the phases A, B, and C can correspond to a particular axis, i.e., the A-axis, the B-axis, and the C-axis, respectively.

[0202] In some embodiments, the compressor 1302 may receive a DC signal as opposed to an AC signal. In this case, the inverter 1306 may or may not be a component of the VRF system 1300. For example, the converter 1304 may include some and / or all of the functionality of the inverter 1306 to modulate the signal it provides to the motor 1308. The strength of the DC signal may be determined by the compressor vibration controller 1310 and provided to the converter 1304 to affect the torque applied by the motor 1308. However, the speed at which the motor 1308 rotates may be determined based on some other input value other than the frequency of the electrical signal. For example, the rotational speed of the motor 1308 may be provided directly to the motor 1308 by the compressor vibration controller 1310 and / or some other computing device. However, the compressor 1302 / motor 1308 are described herein as receiving an AC signal from the inverter 1306 for ease of explanation and clarity.

[0203] FIG. 13B further illustrates the direct (d-axis) and quadrature (q-axis) axes of the motor 1308. The d-axis and q-axis may always be at a 90 degree angle with respect to each other. In some embodiments, the three-phase currents provided by the inverter 1306 may be separated with respect to the d-axis and the q-axis. In this case, and as described in more detail below with reference to FIG. 16, the compressor vibration controller 1310 may generate a first three-phase current prediction for the d-axis and a second three-phase current prediction for the q-axis. In this manner, the motor 1308 may operate appropriately to compress the received fluid. The axial current calculations performed by the compressor vibration controller 1310 are described in more detail below with reference to FIGS. 14A-14C.

[0204] The three-phase current signal provided by the inverter 1306 is shown to include phases A, B, and C. In some embodiments, the compressor vibration controller 1310 is configured to predict an appropriate amplitude value for each phase. In some embodiments, the compressor vibration controller 1310 is configured to predict a single amplitude value to be applied to phases A, B, and C. It should be appreciated that using a single amplitude value to be applied to each phase may be computationally less expensive compared to determining three separate amplitude values.

[0205] 14A-14C, a set of graphs illustrating current transformations are shown, according to some embodiments. FIGS. 14A-14C illustrate relationships between currents that can be used by the compressor vibration controller 1310 in generating a prediction of what current value to provide to a motor (e.g., motor 1308). The compressor vibration controller 1310 can be configured to perform both forward and reverse dq current transformations. The d-axis and q-axis of the motor are represented by current values ​​I dc and I qc In forward current conversion, I dc and I qc are the observed current values ​​I of the detected U and W phases. uc and I wc can be determined by converting the above into values ​​on the dcqc axes of the virtual rotating coordinate system.

[0206] The inverse current transform is used in the UVW current capture process to obtain the estimated current value.

number

number

number

[0207] 14A, a graph 1400 illustrating Kirchhoff's law is shown, according to some embodiments. The graph 1400 shows a current I emanating from point a. v , I u , and I w According to Kirchhoff's law, the sum of the currents at point a is zero. Therefore, Kirchhoff's law states: I v =I u -I w applies.

[0208] This relationship can be used in both forward and reverse current transformation, as will be described in more detail below with reference to Figures 14A and 14B.

[0209] 14B, a graph 1420 illustrating αβ conversion is shown, according to some embodiments. In a forward current conversion, the αβ conversion is performed by the compressor vibration controller 1310 by converting the current I αc and I βc The αβ transformation can be expressed by the following equation:

number

[0210] I v =-I u -I w is substituted into the above equation by the compressor vibration controller 1310, the following relationship can be identified:

number

[0211] 14C, a graph 1440 illustrating a dq transformation is shown, according to some embodiments. dc can represent the angle between the fixed coordinate α axis and the virtual rotating coordinate dc axis. In the forward current transformation, the dq transformation is I dc and I qc But, I αc and I βc This can be obtained by substituting

number

[0212] I dc and I qc The value of I can be filtered and used to control a compressor motor (e.g., motor 1308). dc and I qcTo filter the current, the filter time constant (e.g., milliseconds, seconds, etc.) of the filter can be switched according to the rotation speed of the motor. In some embodiments, the rotation speed is divided into three regions: a low speed region, a medium speed region, and a high speed region. For example, the low speed region can be defined as 0-99 revolutions per minute (RPM), the medium speed region can be defined as 100-200 RPM, and the high speed region can be defined as 200+ RPM. Based on the identified rotation speed, the filter time constant can be switched according to the observed current. In particular, the filter time constant can be maximum in the low speed region and minimum in the high speed region. In this way, as the rotation speed increases, the filter time constant can be decreased to obtain a more granular resolution of the current.

[0213] For the inverse current transformation, the dq inversion can be performed by the compressor vibration controller 1310 based on the following dq inversion equation:

number

[0214] In this way, I αc and I βc The compressor vibration controller 1310 further performs an αβ inverse transformation to obtain the value of I uc and I wc You can get the value of

number

[0215] 15, a graph 1500 relationship between crank angle and torque is shown for different types of compressors, according to some embodiments. Specifically, graph 1500 illustrates the relationship between crank angle and gas compression torque. As described in detail above with reference to FIGS. 13A and 13B, the torque applied by a motor (e.g., motor 1308) can affect vibrations of the compressor. Thus, the torque applied by the motor should be monitored and maintained such that excessive vibrations of the compressor are avoided.

[0216] Graph 1500 is shown to include series 1502, series 1504, and series 1506. Series 1502 is associated with torque measurements of a single rotary compressor, series 1504 is associated with torque measurements of a twin rotary compressor, and series 1506 is associated with torque measurements of a scroll compressor all relative to the crank angle of the motor. As should be understood based on graph 1500, each type of compressor may have a different relationship between crank angle and torque (and therefore vibration). For example, a single rotary compressor represented by series 1502 is shown to have a high peak torque versus crank angle, while a scroll compressor represented by series 1506 has a nearly constant relationship between crank angle and torque. During the AI ​​model training process, compressor vibration controller 1310 can leverage these relationships to generate different AI models for various types of compressors.

[0217] 16, a more detailed block diagram of the compressor vibration controller 1310 is shown, according to some embodiments. The compressor vibration controller 1310 may be configured to utilize an AI model to generate predictions for values ​​of current that may be used to operate the compressor 1302 (and / or some other compressor) to reduce degradation of the compressor 1302 due to vibration. The compressor vibration controller 1310 may be implemented in a variety of ways. In some embodiments, the compressor vibration controller 1310 is a local controller of a building system. For example, the compressor vibration controller 1310 may be implemented in a desktop computer, a mobile device, a thermostat, and / or some other computing device / system local to the building. In some embodiments, the compressor vibration controller 1310 may be implemented as a component of a VRF system. For example, the compressor vibration controller 1310 may be implemented as a component of the inverter 1306 described with reference to FIG. 13A. In this case, the inverter 1306 itself may be able to determine the frequency of the current to provide to the compressor 1302. In some embodiments, the compressor vibration controller 1310 is implemented via some other computing device / system, such as by a cloud computing system.

[0218] The compressor vibration controller 1310 is shown to include a communication interface 1608 and a processing circuit 1602. The communication interface 1608 may include a wired or wireless interface (e.g., a jack, an antenna, a transmitter, a receiver, a transceiver, a wired terminal, etc.) for data communication with various systems, devices, or networks. For example, the communication interface 1608 may include an Ethernet card and port for transmitting and receiving data over an Ethernet-based communication network, and / or a Wi-Fi transceiver for communicating over a wireless communication network. The communication interface 1608 may be configured to communicate over a local area network or a wide area network (e.g., the Internet, a building WAN, etc.) and may use various communication protocols (e.g., BACnet, IP, LON, etc.).

[0219] The communications interface 1608 may be a network interface configured to facilitate electronic data communication between the compressor vibration controller 1310 and various external systems or devices (e.g., the inverter 1306, the sensors 1620, the user devices 1622, etc.). For example, the compressor vibration controller 1310 may provide an AC amplitude setpoint to the inverter 1306 via the communications interface 1608.

[0220] The processing circuit 1602 is shown to include a processor 1604 and a memory 1606. The processor 1604 may be a general purpose or special purpose processor, an application specific integrated circuit (ASIC), one or more field programmable gate arrays (FPGAs), a group of processing components, or other suitable processing components. The processor 1604 may be configured to execute computer code or instructions stored in the memory 1606 or received from another computer readable medium (e.g., a CD-ROM, a network storage device, a remote server, etc.).

[0221] The memory 1606 may include one or more devices (e.g., memory units, memory devices, storage devices, etc.) for storing data and / or computer code for completing and / or facilitating various processes described in this disclosure. The memory 1606 may include random access memory (RAM), read only memory (ROM), hard drive storage, temporary storage, non-volatile memory, flash memory, optical memory, or any other suitable memory for storing software objects and / or computer instructions. The memory 1606 may include database components, object code components, script components, or any other type of information structure for supporting various activities and information structures described in this disclosure. The memory 1606 may be communicatively connected to the processor 1604 via the processing circuitry 1602 and may include computer code for executing (e.g., by the processor 1604) one or more processes described herein. In some embodiments, one or more components of the memory 1606 are part of a single component. However, each component of the memory 1606 is shown independently for ease of explanation.

[0222] The memory 1606 is shown to include a training data collector 1610. The training data collector 1610 may collect training data used to train the artificial intelligence model from one or more training data sources 1618. In particular, the training data collector 1610 may obtain training data associated with vibrations of the compressor 1302.

[0223] The training data collected by the training data collector 1610 may include any relevant data that can be used to train the AI ​​model to learn associations between particular inputs and vibrations of the compressor 1302. Alternatively or additionally, the training data may include information indicative of a relationship between particular inputs and noise generated by the compressor 1302 when noise is used as a proxy for vibration. The training data may include values ​​for inputs including, for example, axis error between the A-axis and d-axis of the motor 1308, detected q-axis current, frequency of the AC signal provided by the inverter 1306, measured noise in the vicinity of the compressor 1302, etc. The training data may also include values ​​of current characteristics (e.g., amplitude) so that the AI ​​model can learn relationships between the inputs and outputs.

[0224] To gather training data, the training data collector 1610 may transmit a query to the training data source 1618 to obtain the training data. In some embodiments, the training data collector 1610 may passively receive training data from the training data source 1618 without having to actively request the training data. The training data source 1618 may include any data source that can store and / or provide training data to the training data collector 1610. For example, the training data source 1618 may be or include a user device (e.g., a laptop, a desktop computer, a mobile device, a tablet, etc.) that can provide a stored training data set to the training data collector 1610. As another example, the training data source 1618 may be or include a database (e.g., a cloud database) that stores data collected during operation of an actual VRF system. In this manner, the AI ​​model can be trained based on data collected directly from an actual VRF device in operation.

[0225] In some embodiments, the training data collector 1610 utilizes one or more simulation models to generate some or all of the training data used by the model generator 1612 to generate the AI ​​model. The simulation model (also referred to herein as a "simulation" or "simulation framework") can simulate how an actual VRF system may operate under various conditions and limitations (e.g., weather conditions, heating / cooling loads, inherent device limitations, etc.). The simulation model can take into account the relationships between components of the VRF system and how the components may react to changing conditions. For example, the simulation model can model how the compressor 1302 may operate to achieve a particular required heating / cooling load, and how the compressor 1302 oscillates as a result.

[0226] By utilizing a simulation model, the training data collector 1610 may not need to retrieve training data from the training data source 1618, but instead may generate the training data within the compressor vibration controller 1310. In some embodiments, the simulation model is hosted by a third party controller / device / system (e.g., a cloud computing system) that can provide the compressor vibration controller 1310 with training data generated as a result of running the simulation model. In either case, the simulation model can be used / executed to generate different training data representing different operating conditions that can be used to train the AI ​​model.

[0227] Advantageously, the simulation model can generate a large amount of data in a shorter time compared to collecting training data from an actual VRF system in operation. In this way, the compressor 1302 can operate based on the AI ​​model's decisions at an earlier time and for a shorter period of time, which may increase the lifespan of the compressor 1302 as vibrations are managed sooner. Furthermore, the simulation model can be used to generate training data for scenarios that may not occur frequently during actual VRF system operation. For example, the simulation model may be able to generate training data representing operating the compressor 1302 under intense heating / cooling loads that may not occur frequently during standard operation of an actual VRF system. In this way, a large amount of training data representing a variety of cases can be generated and made available to the model generator 1612 to generate accurate and representative models.

[0228] Based on the acquired training data, the training data collector 1610 can combine the collected training data into a training data set and provide the training data set to the model generator 1612. Based on the training data set, the model generator 1612 can generate an AI model that models the relationship between a set of inputs describing the operating conditions of the compressor 1302 and corresponding outputs describing the amplitude of an AC signal that can be provided to the compressor 1302 to manage vibrations of the compressor 1302. In some embodiments, the model generator 1612 trains the AI ​​model to predict the AC signal amplitude that can be provided to the motor 1308 via the inverter 1306 to reduce noise generated by the compressor 1302 while still meeting the required heating / cooling load.

[0229] The AI ​​model generated by the model generator 1612 may be of any of a variety of AI model architectures. For example, the AI ​​model may be an RNN, such as an LSTM network.

[0230] Inputs to the AI ​​model generated by the model generator 1612 may include various inputs associated with the operation of the compressor 1302 and / or the motor 1308. For example, the inputs to the AI ​​model may include an axis error (also referred to as an axial difference) between the d-axis and A-axis of the motor 1308, a q-axis current sense value, an inverter frequency indicative of the rotational speed of the motor 1308, and an actual and / or desired noise level. In some embodiments, the rotational speed of the motor 1308 (e.g., revolutions per minute (RPM)) is used as an input instead of or in addition to the inverter frequency. An output of the AI ​​model may input a target amplitude value (e.g., in volts, amperes, etc.) of the AC signal provided to the compressor 1302 to manage the torque applied by the motor 1308. Exemplary illustrations of AI models that may be generated by the model generator 1612 are described in more detail below with reference to FIGS. 17A and 17B.

[0231] The model generator 1612 may utilize a variety of training techniques to generate the AI ​​model. For example, the model generator 1612 may utilize a stochastic gradient descent with momentum approach, an adaptive moment estimation approach, a root mean square propagation approach, etc. For the root mean square propagation approach, the model generator 1612 may utilize the root mean square error (RMSE) to measure how accurate the model predictions are against the training data provided by the training data collector 1610. To monitor RMSE over time, the model generator 1612 may utilize the following formula:

number

[0232] The model generator 1612 can provide the generated AI model to a prediction generator 1614. The prediction generator 1614 can use the AI ​​model to generate a prediction for values ​​of AC signal amplitude that reduce the amount of vibration and / or noise generated by the compressor 1302. To generate the prediction, the prediction generator 1614 can operate to obtain values ​​of inputs required by the AI ​​model from various sources. For example, the prediction generator 1614 can obtain values ​​of measured variables from sensors 1620, variables manually observed by a user via a user device 1622, and / or any other suitable source of input values. With respect to a particular input to the AI ​​model, the rotational speed of the motor 1308 can be observed by a user, can be commanded and calculated by the prediction generator 1614 based on a known frequency provided to the motor 1308, etc. In some embodiments, the axial difference between the a-axis and d-axis of the motor 1308 is estimated by an observer (e.g., a user) by providing a small current to the motor 1308 and analyzing feedback from the motor 1308 to estimate the axial difference. In some embodiments, the axial difference is measured by a sensor of the sensors 1620. However, due to the price associated with sensors configured to measure the axial difference, manual estimation may be preferred. In some embodiments, the q-axis feedback current associated with the compressor 1302 is measured directly via a current sensor of the sensors 1620. With respect to noise, the actual noise level may be measured via an audio sensor of the sensors 1620 and / or estimated by the user. If the AI ​​model utilizes a required noise level as an input, the required noise level may be provided by a user via the user device 1622, estimated / determined by the prediction generator 1614 (or another component of the compressor vibration controller 1310), etc. In this case, the required noise level describes a predetermined noise value (e.g., in dB) to be achieved by the compressor 1302.

[0233] As a result of passing the received values ​​of inputs to the AI ​​model generated by the model generator 1612, the prediction generator 1614 may determine an amplitude setpoint for the AC signal provided by the inverter 1306 to the compressor 1302. In some embodiments, the AC amplitude setpoint includes different amplitudes for the current provided to the d-axis and the current provided to the q-axis. In some embodiments, a single AC amplitude setpoint is determined as a result of passing the input values ​​to the AI ​​model. In this case, the single AC amplitude setpoint may be applied to both the d-axis and the q-axis currents.

[0234] In some embodiments, the prediction generator 1614 provides the AC amplitude set point directly to the inverter 1306 (e.g., via the communication interface 1608). In some embodiments, the prediction generator 1614 provides the AC amplitude set point to a corrective action generator 1616 in the memory 1606. The corrective action generator 1616 can be configured to generate and initiate one or more corrective actions based on the AC amplitude set point. In some embodiments, one corrective action is to provide the AC amplitude set point to the inverter 1306. In this case, the corrective action generated and initiated by the corrective action generator 1616 can be to operate the inverter 1306 based on a control signal indicative of the AC amplitude set point. In this manner, the torque applied by the motor 1308 can be adjusted. In some embodiments, another corrective action that can be generated and initiated by the corrective action generator 1616 is to provide a notification to the user device 1622 to notify the user about the AC amplitude set point. Such notifications may enable a user to understand how different AC amplitudes affect vibration and may enable a user to take preemptive measures to ensure that building equipment (e.g., compressor 1302) avoids rapid deterioration due to vibration. In some embodiments, corrective action generator 1616 generates and initiates other types of corrective actions, such as scheduling maintenance on compressor 1302, in response to determining that the AC amplitude setpoint required to avoid rapid deterioration is below a threshold, thereby jeopardizing the performance of the building's heating / cooling load.

[0235] 17A, an example of a neural network (NN) 1700 for predicting a value of a current to provide to a compressor motor is shown, according to some embodiments. The NN 1700 can illustrate an example structure of an AI model that can be generated by the model generator 1612 as described with reference to FIG. 16. The NN 1700 is shown to include input nodes in an input layer that correspond to a set of inputs. Specifically, the NN 1700 is shown to include input nodes for an axis error between the A-axis and the d-axis of the motor 1308, a q-axis current sensed value, an inverter frequency describing the frequency of the AC signal output by the inverter 1306, and an actual noise level. Of course, it should be understood that these inputs are provided merely for the sake of example. The model generator 1612 can generate an AI model that utilizes different inputs that can be used to estimate the vibration / noise of the compressor. With respect to the outputs, the NN 1700 is shown to include separate outputs for the d-axis current amplitude and the q-axis current amplitude. Separate outputs for the d-axis and the q-axis can be useful in a more accurate operating compressor 1302. In some embodiments, the NN 1700 outputs a single current amplitude value that can be applied to both the d-axis and q-axis currents. Outputting a single value may reduce the complexity of the NN 1700, thereby improving processing efficiency when generating predictions.

[0236] As discussed above with reference to FIG. 16, the axis error may be measured by a user and / or a sensor, the q-axis current sensed value may be measured by a current sensor, the inverter frequency may be measured, may be a known value, etc. With respect to the actual noise level, the actual noise level may be measured by an audio sensor with a predefined physical proximity of the compressor 1302 (e.g., physically attached to the compressor 1302, within 10 feet, within 1 foot, etc.). Based on the actual noise level, the NN 1700 may essentially predict the vibration of the compressor 1302 and output an amplitude setpoint depending on the predicted vibration. In this case, feedback regarding the operation of the compressor 1302 has been provided to the NN 1700. Specifically, since noise can be used as a proxy for vibration, the NN 1700 has effectively acquired knowledge indicative of the vibration of the compressor 1302. In this manner, the NN 1700 may utilize this information to predict AC signal amplitudes that avoid unsafe operating conditions. For example, if the noise level is relatively low (e.g., <50 dB, <20 dB, etc.), the NN 1700 may output a larger amplitude (e.g., in volts or amperes) compared to if the noise level is relatively high (e.g., >50 dB, >100 dB, >200 dB, etc.). In this manner, the NN 1700 can change the amplitude value over time to achieve appropriate operating conditions for the compressor 1302 that avoid unsafe operating conditions and vibrations.

[0237] 17B, ​​an illustration of a neural network (NN) 1750 for predicting a value of current to provide to a compressor motor is shown, according to some embodiments. In some embodiments, the NN 1750 is similar to and / or the same as the NN 1700, as described above with reference to FIG. 17A. Thus, the NN 1750 may be another example of an AI model that may be generated by the model generator 1612 described with reference to FIG. 16. As shown in FIG. 17B, the NN 1750 is shown to receive as an input a desired noise level to compare with an actual noise level of the NN 1700. The desired noise level may be a target value of the noise generated by the compressor 1302. In some embodiments, the desired noise level serves as a threshold such that the noise generated by the compressor 1302 falls below the threshold. In particular, the desired noise level may be a noise level (e.g., in dB) known to be associated with vibrations determined (e.g., by a user) to not place the compressor 1302 in an unsafe operating condition. For example, the required noise level may be 50 dB expected to sustain vibration of the compressor 1302 under a rate of 30 Hz.

[0238] To use the desired noise level as an input, the NN 1750 can be trained based on a known correlation between noise and vibration. In this way, the NN 1750 can adjust the output current amplitude based on other inputs (e.g., inverter frequency, which is known to be associated with the amount of fluid passing through the compressor 1302) against the known correlation. In some embodiments, the model generator 1612 generates an AI model that utilizes both the actual noise level and the desired noise level as inputs.

[0239] 18, a flow diagram of a process 1800 for predicting an AC signal amplitude to provide to a compressor using an AI model is shown, according to some embodiments. The AC signal amplitude may be an amplitude value (e.g., in volts, amperes, etc.) predicted to result in reduced vibration of the compressor while still meeting the required heating / cooling load and / or other requirements of the building. In some embodiments, the amplitude of the AC signal affects the torque applied by the compressor motor. Specifically, an increased amplitude may be associated with increasing the torque applied by the motor. In some embodiments, some and / or all of the steps of the process 1800 are performed by the compressor vibration controller 1310 described with reference to FIGS. 13 and 16.

[0240] Process 1800 is shown to include obtaining training data describing a relationship between conditions affecting compressor operation and noise generated by the compressor (step 1802). In this case, the noise can be used as a proxy for compressor vibration. The training data can include values ​​such as, for example, axis error between the A-axis and d-axis of the compressor motor, q-axis current sensed value, frequency of the AC signal provided by the inverter (which can represent the amount of fluid passing through the compressor and thus the required heating / cooling load), actual and / or required noise levels, noise measurements, etc. The training data can be obtained from one or more training data sources, such as, for example, a cloud database, a database local to the building, a user, sensors in the building, etc. In some embodiments, step 1802 is performed by the training data collector 1610.

[0241] Process 1800 is shown to include training an AI model to predict, based on training data, the amplitude of an AC signal that will affect the torque applied by the motor of the compressor (step 1804). In some embodiments, a larger AC signal amplitude corresponds to an increased torque applied by the motor. However, as the torque applied by the motor increases, the vibration of the compressor may increase, which may be measured based on the amount of noise output by the compressor. The AI ​​model trained in step 1804 can learn which AC signal amplitude corresponds to a particular noise level that depends on a set of input values. In some embodiments, the goal of the AI ​​model is to learn which AC signal amplitude reduces and / or minimizes the noise output by the compressor while still meeting the required heating / cooling load. Of course, due to operation, some noise / vibration may be unavoidable, but the AI ​​model can be trained to identify amplitudes that avoid excessive deterioration of the equipment while still meeting the needs of the building. In some embodiments, step 1804 is performed by the model generator 1612.

[0242] Process 1800 is shown to include obtaining data associated with the operation of the compressor (step 1806). In this case, the data may be associated with inputs to the AI ​​model trained in step 1804. For example, the data obtained in step 1806 may include the frequency of the AC signal provided and / or provided to the compressor, the ambient temperature, the actual noise level in the vicinity of the compressor, etc. The data may be obtained from various sources, such as, for example, a user, a sensor, as feedback from the device, etc. In some embodiments, step 1806 is performed by the prediction generator 1614.

[0243] Process 1800 is shown to include generating an AC amplitude setpoint based on the acquired data using an AI model (step 1808). The AI ​​model can generate an AC amplitude setpoint such that the required heating / cooling load (and / or some other requirement / need of the building) is met and the corresponding noise generated by the compressor is reduced. In this manner, the motor is operated to compress a constant amount of fluid while avoiding unnecessary degradation. Advantageously, using an AI model to generate the AC amplitude setpoint can reduce the frequency of situations in which the compressor experiences unnecessary degradation as a result of the motor applying an unnecessary amount of torque. In some embodiments, the AC amplitude setpoint includes separate amplitude setpoints for the d-axis and q-axis of the motor. In some embodiments, step 1808 is performed by the prediction generator 1614.

[0244] Process 1800 is shown to include providing an AC amplitude setpoint to an inverter to operate the motor (step 1810). In some embodiments, the inverter generates the actual AC signal that is provided to the compressor. In this manner, the inverter can be operated such that the AC signal output by the inverter has an amplitude equal (or approximately equal) to the AC amplitude setpoint. As a result, the motor can be operated to apply a torque according to the amplitude of the AC signal. In some embodiments, step 1810 is performed by prediction generator 1614 and / or corrective action generator 1616.

[0245] VRF system fault condition prediction Generally, referring to Figures 19-22, a system and method for predicting fault conditions of a VRF system, according to some embodiments, are shown and described. Fault conditions can include any type of fault that can increase costs and / or result in some other undesirable characteristic of the VRF system. For example, fault conditions can include refrigerant leaks, outdoor unit frosting, clogged indoor fan, dirty indoor filters, dirty heat exchangers, dirty outdoor fans, motor demagnetization, compressor oil leaks, etc. Fault conditions can result in imperfect efficiency of the compressor, which can impact costs (e.g., operational costs) associated with the VRF system.

[0246] As will be described in more detail through Figures 19-22, the AI ​​model may be used to predict fault conditions of components in a VRF system. In particular, the AI ​​model may be used to predict fault conditions associated with a compressor of a VRF system. For each different fault condition, a percentage may be used to represent the amount that the fault will impact the VRF system. Thus, there may be two models (e.g., two RNNs) for the detailed fault classification. The first model may be configured to output a fault classification describing what fault condition, if any, exists in the VRF system. Effectively, the first model may be configured to determine / predict the fault (if any) that the VRF system and / or its components are experiencing. The second model may be configured to determine / predict the severity of each fault condition using a representative percentage and / or some other metric. The fault severity may effectively indicate the extent to which the fault will impact the VRF system. For example, an output of a refrigerant leakage index of 0.5 (i.e., 50%) by the second model may indicate that 50% of the refrigerant is lost through leakage. As another example, an indoor fan clogging index output of 0.9 by the second model may indicate that the indoor fan is 90% clogged. In some embodiments, the first model and the second model are combined into a single model. In this case, the single model may output fault classifications indicative of fault conditions in the VRF system and corresponding metrics representing the severity associated with each condition. In some embodiments, the second model may not be utilized. In this case, the fault classifications generated by the first model are utilized in determining the corrective action to initiate. In other words, if the second model is not utilized, corrective action may be initiated based on the fault classification and regardless of the amount the fault impacts the VRF system.

[0247] As should be understood, the systems and methods shown and described below through Figures 19-22 may be applied to a variety of building systems and may not necessarily be limited to VRF systems, for example, the systems and methods may be applied to other HVAC systems in buildings as well.

[0248] 19, a block diagram of a VRF system 1900 is shown, according to some embodiments. The VRF system 1900 can illustrate an example VRF system that can apply an AI model to generate a prediction of a fault classification. In particular, the VRF system 1900 can illustrate sources of input values ​​that can be used in generating a fault classification via an AI model. In some embodiments, the VRF system 1900 is similar and / or the same as the VRF system 600 described with reference to FIGS. 6A and 6B.

[0249] The VRF system 1900 is shown to include an outdoor unit (ODU) 1902 and an indoor unit (IDU) 1904. The ODU 1902 is shown to include a compressor 1906 and a heat exchanger 1908. The compressor 1906 can receive a current I from an inverter via a compressor power line 1914. I inverter can be any type of power signal, for example, AC. inverter can power the compressor 1906. In some embodiments, the compressor 1906 can be a compressor such as an I inverter The motor is operated according to the frequency and amplitude of the input signal.

[0250] The compressor 1906 may also receive a fluid (e.g., refrigerant) from the heat exchanger 1908. To obtain the fluid, the compressor 1906 may draw fluid from the heat exchanger 1908 via a suction line 1922. The draw of the fluid is represented by a suction pressure P, which may be measured in a suitable pressure metric, such as Pascals. sIn some embodiments, P s is measured by a pressure sensor associated with the suction line 1922.

[0251] As a result of drawing in fluid through suction line 1922, compressor 1906 may compress the fluid and output the compressed fluid through discharge line 1916 to IDU 1904. The output of fluid through discharge line 1916 is represented by a discharge pressure P d and discharge temperature T d In some embodiments, P d and T d are measured by a pressure sensor and a temperature sensor, respectively, associated with the discharge line 1916.

[0252] The fan 1912 of the IDU 1904 supplies compressed fluid (e.g., compressed refrigerant) via discharge line 1916 and indoor fan current I via indoor fan power line 1918. fan,i Using the compressed fluid, the fan 1912 can operate to provide heating / cooling to the space. Specifically, heat released by the compressed fluid can be pushed into the space using the fan 1912. In some embodiments, the IDU 1904 includes multiple fans 1912 that provide heating / cooling to various locations within the space.

[0253] The IDU 1904 can recirculate the fluid back to the ODU 1902 via return piping 1920. Specifically, the fluid can be returned to the heat exchanger 1908. The outdoor fan current I fan,o A fan 1910 powered by a refrigerant coolant supply 1924 can provide air to the heat exchanger 1908. In this way, some heat can be dissipated from the fluid to the outside space. Finally, the fluid can be returned to the compressor 1906 via suction piping 1922.

[0254] The variables shown in the VRF system 1900 and described above (e.g., Ifan,i , P s , P d , T d 19, etc.) may be measured by respective sensors attached to, within, or otherwise associated with the piping 1914-1924. The measurements of the variables may be provided back to a controller that utilizes an AI model to identify faults associated with the VRF system 1900. For example, the variables may be provided to a VRF fault controller 2000, as described in more detail below with reference to FIG. 20. As should be understood, the variables and components shown in FIG. 19 are provided merely for example purposes. The VRF system 1900 may include different components than those shown in FIG. 19. Similarly, variables other than or in addition to those shown in FIG. 19 may be provided to the controller.

[0255] Referring now to FIG. 20, a block diagram of a VRF fault controller 2000 for predicting faults in a VRF system is shown, according to some embodiments. The VRF fault controller 2000 can be used to generate predictions regarding faults associated with a VRF system (e.g., VRF system 600 described with reference to FIG. 16, VRF system 1900 described with reference to FIG. 19, etc.). The VRF fault controller 2000 can utilize one or more AI models to generate fault classifications and corresponding severities of fault conditions identified by the fault classifications. In some embodiments, the VRF fault controller 2000 is integrated with other controllers described herein as a single controller. For example, the functionality of the VRF fault controller 2000 can be integrated with the oil management controller 800 described with reference to FIG. 8 and / or the compressor vibration controller 1310 described with reference to FIG. 16.

[0256] The VRF fault controller 2000 is shown to include a communication interface 2008 and a processing circuit 2002. The communication interface 2008 may include a wired or wireless interface (e.g., a jack, an antenna, a transmitter, a receiver, a transceiver, a wired terminal, etc.) for data communication with various systems, devices, or networks. For example, the communication interface 2008 may include an Ethernet card and port for transmitting and receiving data over an Ethernet-based communication network, and / or a Wi-Fi transceiver for communicating over a wireless communication network. The communication interface 2008 may be configured to communicate over a local area network or a wide area network (e.g., the Internet, a building WAN, etc.) and may use various communication protocols (e.g., BACnet, IP, LON, etc.).

[0257] The communications interface 2008 may be a network interface configured to facilitate electronic data communication between the VRF fault controller 2000 and various external systems or devices (e.g., training data sources 2018, sensors 2020, user devices 2024, etc.). For example, the VRF fault controller 2000 may provide a notification describing a fault classification of the apparatus 2022 to the user device 2024 via the communications interface 2008.

[0258] The processing circuit 2002 is shown to include a processor 2004 and a memory 2006. The processor 2004 may be a general purpose or special purpose processor, an application specific integrated circuit (ASIC), one or more field programmable gate arrays (FPGAs), a group of processing components, or other suitable processing components. The processor 2004 may be configured to execute computer code or instructions stored in the memory 2006 or received from another computer readable medium (e.g., CDROM, network storage, remote server, etc.).

[0259] The memory 2006 may include one or more devices (e.g., memory units, memory devices, storage devices, etc.) for storing data and / or computer code for completing and / or facilitating various processes described in this disclosure. The memory 2006 may include random access memory (RAM), read only memory (ROM), hard drive storage, temporary storage, non-volatile memory, flash memory, optical memory, or any other suitable memory for storing software objects and / or computer instructions. The memory 2006 may include database components, object code components, script components, or any other type of information structure for supporting various activities and information structures described in this disclosure. The memory 2006 may be communicatively coupled to the processor 2004 via the processing circuitry 2002 and may include computer code for executing (e.g., by the processor 2004) one or more processes described herein. In some embodiments, one or more components of the memory 2006 are part of a single component. However, each component of the memory 2006 is shown independently for ease of explanation.

[0260] The memory 2006 is shown to include a training data collector 2010. In some embodiments, the training data collector 2010 is similar to and / or the same as the training data collector 1610 described with reference to FIG. 16. The training data collector 2010 can be configured to obtain training data for use in training the AI ​​model from a training data source 2018. The training data source 2018 can include a variety of sources that can provide training data to the VRF fault controller 2000. For example, the training data source 2018 can include a cloud database, an on-site database of a building, a user device, a building appliance, etc. In some embodiments, the training data source 2018 is similar to and / or the same as the training data source 1618.

[0261] The training data collected by the training data collector 2010 may include any information relevant to train the AI ​​model to predict what faults may be present in the VRF system and, in some cases, the severity of the faults (e.g., percentage values ​​of the faults, with 0% representing no problem present and 100% representing a catastrophic fault). For example, the training data may include values ​​for compressor speed, ambient temperature, discharge temperature, suction pressure, discharge pressure, indoor fan mode and VRF mode (e.g., heating or cooling), outdoor fan step, etc. The training data may also include fault classifications associated with the values. In this case, the fault classifications may be entered into the training data by a user (e.g., maintenance personnel, building owner, etc.) depending on the values. Alternatively or additionally, the fault classifications may be entered into the training data automatically based on equipment feedback indicating when a particular device has failed and / or is otherwise in a fault state. In some embodiments, the fault classifications are entered into the training data via some other source of fault classifications.

[0262] In some embodiments, the training data collector 2010 utilizes a simulation framework to generate some and / or all of the training data required to train one or more AI models. In this case, the simulation framework can run simulations of test cases representing the operation of the VRF system by using a closed-loop Functional Mockup Unit (FMU) model of the VRF system. The simulation framework can model how devices degrade over time and what failures may occur as a result. The simulation framework can run on different equipment, for different loads, for different durations, etc. In this way, a comprehensive collection of training data can be obtained that can be used in training the AI ​​models to learn how to identify failures in the VRF system. Advantageously, the simulation framework can ensure that the AI ​​models have enough training data to accurately model the dynamics in the VRF system and, as a result, how building devices are affected. The simulation training data can be used instead of or in addition to training data collected from the training data source 2018.

[0263] The training data collector 2010 may provide the collected training data as a training data set to the model generator 2012. In some embodiments, the model generator 2012 is similar to and / or the same as the model generator 1612 described with reference to FIG. 16. The model generator 2012 may utilize the training data obtained from the training data collector 2010 to train one or more AI models for predicting characteristics of fault conditions. In particular, the model generator 2012 may train a first AI model to output a fault classification in response to a set of inputs describing the operation of the VRF system and train a second AI model to output a severity (e.g., a percentage value) of faults identified by the first AI model having on the VRF system. In some embodiments, the model generator 2012 generates a single AI model that outputs a fault classification and a corresponding severity of each fault condition identified in the fault classification. In some embodiments, the model generator 2012 generates a single AI model that outputs only the fault classification. An example of an AI model that outputs a fault classification is described in more detail below with reference to FIG. 21.

[0264] The fault classifications output by the one or more AI models may describe faults in the VRF system. For example, the fault classifications may include text strings such as "refrigerant leak," "indoor fan clogged," "outdoor unit frosted," "dirty indoor filter," "dirty heat exchanger," "dirty outdoor fan," "motor demagnetization," "compressor oil leak," "compressor inefficient," and the like. As another example, the fault classifications may indicate the presence of a particular fault condition using a binary variable (e.g., 0 or 1) that represents whether a particular fault condition exists. The fault classifications may be predicted by the one or more AI models based on inputs describing the VRF system. Specifically, the one or more AI models may learn correlations between values ​​of input variables such as compressor speed, ambient temperature, discharge temperature, suction pressure, discharge pressure, indoor fan mode, outdoor fan step, and the like, and corresponding to a fault classification. As a particular example, the AI ​​models may learn, based on training data, that there may be a refrigerant leak in the VRF system if the suction pressure exceeds a particular threshold (e.g., 20 pounds per square inch (PSI), 30 PSI, etc.). The AI ​​model may be further configured to predict a percentage value of how much refrigerant is leaking from the VRF system. In an example, the AI ​​model may predict that 50% of the refrigerant is being lost due to leakage based on inputs such as compressor speed, discharge temperature, etc.

[0265] The one or more AI models generated by the model generator 2012 may be any type of AI model. For example, the one or more AI models may be an RNN (e.g., an LSTM model), a convolutional neural network, a multi-layer perceptron, a feed-forward neural network, etc. In particular, the one or more AI models may be an RNN due to its high speed and reliability. More specifically, an LSTM may be utilized even with lags of unknown duration between important events in the time series due to the LSTM's ability to process an entire sequence of time series data and make predictions. As a specific example structure, one of the AI ​​models generated by the model generator 2012 may have one sequence input layer, one bidirectional LSTM layer, one fully connected layer, one softmax layer, and one classification layer. In this way, the AI ​​model may be used to generate fault classifications and may learn bidirectional long-range dependencies between time steps of the time series or sequence data. These dependencies are useful for the network to learn from the complete time series for each time step.

[0266] In some embodiments, the model generator 2012 generates an individual AI model for each fault condition. In this case, each AI model can be trained to predict whether a particular fault condition exists in the VRF system. Generating individual AI models can improve accuracy for predicting whether a particular fault condition exists. In particular, generating individual AI models can avoid duplicate data for multiple fault conditions that may affect the generation of the model. In some embodiments, if the model generator 2012 generates individual AI models for the fault conditions, the model generator 2012 can combine each individual model to generate a composite AI model that considers multiple fault conditions.

[0267] The model generator 2012 can provide the one or more generated AI models to a prediction generator 2014. In some embodiments, the prediction generator 2014 is similar to and / or the same as the prediction generator 1614. Using the one or more AI models, the prediction generator 2014 can generate a prediction of fault conditions, if any, affecting the VRF system, and in some embodiments, the severity of the fault conditions. To generate a prediction, the prediction generator 2014 can obtain value inputs required by the one or more AI models. In particular, the prediction generator 2014 can obtain measured variables from the sensors 2020 and device feedback from the devices 2022. The sensors 2020 can include one or more sensors configured to measure certain characteristics of the VRF system. For example, the sensors 2020 can include a P d , P s , T d , ambient temperature, and other values. The device 2022 may include any device of a VRF system. For example, the device 2022 may include an ODU, an IDU, and the like. The device feedback may include inputs required by one or more AI models, such as, for example, indoor fan mode, outdoor fan step, compressor speed, discharge temperature, suction pressure, and the like. In some embodiments, the sensor 2020 is a component of the device 2022 and / or is otherwise associated with the device 2022.

[0268] The prediction generator 2014 can pass the received values ​​of inputs to one or more AI models to generate a prediction associated with a fault characteristic of the VRF system. The prediction generator 2014 can provide the prediction to a corrective action generator 2016. The corrective action generator 2016 can be configured to initiate a variety of different corrective actions depending on the fault condition indicated by the fault classification included in the prediction and the severity of the fault condition. The corrective action initiated by the corrective action generator 2016 can include any suitable action to resolve one or more of the fault conditions identified in the prediction. For example, the corrective action can include notifying a user about the fault condition by providing a notification to the user device 2024, generating and providing a control signal to the apparatus 2022, temporarily disabling some VRF devices of the apparatus 2022, scheduling maintenance of the apparatus 2022, etc.

[0269] In some embodiments, the corrective action initiated by the corrective action generator 2016 is based on the fault classification included in the prediction and the severity of each fault. Some fault conditions may require corrective actions that are associated with a higher cost (e.g., $) and may require more time of a maintenance technician, etc., compared to other corrective actions. For example, scheduling maintenance on a particular building device (e.g., an IDU) may have a higher associated cost compared to providing a notification to a user device 2024. Thus, the corrective action generator 2016 may determine the impact that a particular fault condition may have on the VRF system and initiate a corrective action depending on the determined impact. For example, if the fault classification included in the prediction received from the prediction generator 2014 indicates that the indoor fan filter is 50% clogged, the corrective action generator 2016 may determine that the impact on the VRF system is low and may initiate a corrective action to notify a user via the user device 2024 regarding the clog. However, if the fault classification indicates that the VRF system's compressor is 90% inefficient, the corrective action generator 2016 may determine that the compressor inefficiency has a high impact on operational costs, thereby initiating corrective action to perform maintenance and / or replacement of the compressor.

[0270] To determine what corrective action to initiate, the corrective action generator 2016 may utilize a mapping between fault conditions with corresponding severity and specific corrective actions. In particular, if severity is given as a percentage, where 0% indicates no impact to the VRF system and 100% indicates severe impact to the VRF system, the corrective action generator 2016 may initiate a specific corrective action based on the severity boundaries. In some embodiments, multiple corrective actions are initiated for a single fault condition. An example of a decision tree that may be utilized to determine what corrective action to initiate is shown below: Table 2 to be provided. [Table 2]

[0271] If a corrective action is initiated, it may address a fault condition predicted to exist. By utilizing one or more AI models to predict a fault classification indicative of a particular fault condition and a corresponding severity, appropriate corrective action may be initiated to address a particular fault predicted by the one or more AI models. Advantageously, the corrective action initiated to address a particular fault may be responsive to the estimated severity of the fault. In this manner, expensive and / or time-consuming corrective actions may be avoided for faults that have a low impact on the VRF system.

[0272] 21, an illustration of a neural network (NN) 2100 for predicting a fault classification of a VRF system is shown, according to some embodiments. The NN 2100 may illustrate an example structure of an AI model that may be generated by the model generator 2012 described with reference to FIG. 20. Specifically, the NN 2100 may represent an RNN structure for fault classification prediction.

[0273] NN2100 is shown to include input nodes in an input layer that correspond to a set of inputs. NN1700 is shown to include input nodes for compressor speed, ambient temperature, discharge temperature, suction pressure, discharge pressure, indoor fan mode, and outdoor fan step. As should be understood, the inputs shown in NN2100 are provided for example purposes only. The inputs to NN2100 can be modified based on what measurements may be collected from the VRF system, what measurements are relevant to generating a fault classification, etc.

[0274] The output of the NN2100 may include fault classifications that identify specific faults that may be present in the VRF system. For example, the NN2100 may identify fault conditions such as refrigerant leaks, clogged indoor fans, frosted outdoor units, dirty indoor filters, dirty heat exchangers, dirty outdoor fans, motor demagnetization, compressor oil leaks, imperfect compressor efficiency, etc. As with the inputs, the fault classifications that may be output by the NN2100 may be modified. The fault classifications that may be output by the NN2100 may be adjusted depending on user preferences, devices in the system, what inputs are available to the NN2100, etc. Specifically, the fault classifications may be adjusted to only indicate whether a particular set of fault conditions are present in the VRF system.

[0275] 22, a flow diagram of a process 2200 for predicting a fault classification of a VRF system using an AI model is shown, in some embodiments. In some embodiments, some and / or all of the steps of the process 2200 are performed by the VRF fault controller 2000.

[0276] Process 2200 is shown to include obtaining training data associated with operational conditions of the VRF system and fault conditions of the VRF system (step 2202). The training data can be obtained from various sources, such as, for example, a cloud database, directly from a user, device feedback, etc. The operational conditions included in the training data can be, for example, T d , P d , P s, ambient temperature, compressor speed, outdoor fan step (i.e., outdoor fan speed), indoor fan speed, and VRF mode (e.g., heating or cooling). The fault conditions included in the training data may include any type of fault condition that may be experienced by the VRF system. For example, the fault conditions may include refrigerant leaks, clogged indoor fan, outdoor unit frosting, dirty indoor filters, dirty heat exchangers, dirty outdoor fans, motor demagnetization, compressor oil leaks, imperfect compressor efficiency, and the like. In some embodiments, each fault condition included in the training data also includes an estimated severity (e.g., a percentage value, a value on a 1-10 scale, and the like). The estimated severity may be estimated automatically or manually by a user based on the changing operating costs as indicated by the training data, and the like. In some embodiments, the training data is time series data so that relationships between changes in operating conditions and particular fault conditions can be identified. In some embodiments, step 2202 is performed by the training data collector 2010.

[0277] Process 2200 is shown to include training an AI model to predict a fault classification of the VRF system based on the training data (step 2204). The AI ​​model trained in step 2204 may be any type of AI model, such as an RNN. The fault classification predicted by the AI ​​model may indicate whether one or more fault conditions exist in the VRF system and the corresponding severity of any existing faults. In some embodiments, two AI models are trained in step 2204, such that a first AI model predicts whether a fault classification exists indicating what fault conditions, if any, exist based on the input data, and a second AI model predicts the severity of the fault conditions that exist. Examples of AI models that may be generated in step 2204 are described above with reference to FIG. 21. In some embodiments, step 2204 is performed by a model generator 2012.

[0278] Process 2200 is shown to include obtaining data associated with operation of the VRF system (step 2206). The data obtained in step 2206 may include values ​​of inputs required by one or more AI models generated in step 2204. For example, the data may include values ​​of inputs required by one or more AI models generated in step 2204. d , P d , P s , ambient temperature, compressor speed, outdoor fan step, indoor fan speed, and VRF mode values. The data can be obtained from various sources such as sensors, device feedback, users, etc. In some embodiments, step 2206 is performed by the prediction generator 2014.

[0279] Process 2200 is shown to include using the AI ​​model to predict a fault classification of the VRF system based on the acquired data (step 2208). The fault classification may indicate what fault conditions, if any, are identified as being present in the VRF system. The fault classification may indicate the existing fault conditions through any suitable manner, such as, for example, by including a text string of the existing fault condition, by including an array of binary variables indicating the presence or absence of a particular fault condition, etc. In some embodiments, step 2208 also includes predicting a severity of each existing fault condition based on the acquired data using the AI ​​model (or a second AI model). In some embodiments, step 2208 is performed by a prediction generator 2014.

[0280] Process 2200 is shown to include determining a corrective action based on the fault classification output by the AI ​​model (step 2210). In some embodiments, the corrective action is determined based on what fault conditions, if any, are identified by the fault classification. In some embodiments, the corrective action is further determined based on an indicated severity of each fault condition identified by the fault classification. The corrective action may include any action that addresses the fault condition indicated by the fault classification. For example, the corrective action may include notifying a user regarding the identified fault condition and corresponding severity, scheduling maintenance / replacement of the VRF device, generating and transmitting control signals to the VRF device, recording the fault condition and corresponding severity in a database, etc. In some embodiments, at step 2210, multiple corrective actions are determined. In some embodiments, step 2210 is performed by corrective action generator 2016.

[0281] Process 2200 is shown to include initiating a corrective action (step 2212). Step 2212 may include performing any steps / processes necessary to ensure that the corrective action is performed successfully. For example, if the corrective action is to notify a user, step 2412 may include generating a notification and communicating the notification using a communication interface. As another example, if the corrective action is to disable a VRF device, step 2412 may include generating a control signal to turn off the VRF device and providing the control signal to the VRF device. In some embodiments, step 2212 is performed by a corrective action generator 2016.

[0282] Motor Efficiency Prediction 23-24, systems and methods for predicting efficiency of a motor in a VRF system using an AI model are shown and described, according to some embodiments. In some embodiments, the systems and methods described below can be integrated with any and / or all of the systems and methods described above through FIGS.

[0283] The efficiency of a motor for a compressor can be defined based on the input power and the output power of the motor. Specifically, the motor efficiency η motor can be given by the following formula:

number

number

[0284] Changes in motor efficiency can be an important indicator for detecting faults and inferential reasons that lead to efficiency degradation. Detection of motor efficiency can enable a system (e.g., a VRF system) to identify abnormal information early (e.g., before a catastrophic failure). Based on the detected changes in motor efficiency, the operation of the motor can be adjusted accordingly. Specifically, the detected changes in motor efficiency can result in offset compensation to the motor drive and / or complete shutdown of the machine. Thus, prediction of motor efficiency can reduce overall hardware costs (e.g., operating costs, maintenance costs, etc.) and can improve the reliability of the motor drive.

[0285] As described in more detail below, an AI model can be utilized to predict motor efficiency, and optionally, inverter efficiency can be utilized to detect abnormal conditions in the motor drive system. The AI ​​model used to predict efficiency can be any type of AI model. In particular, the AI ​​model for efficiency prediction can be a linear RNN model due to the case-dependent nature of the data describing efficiency. If a linear RNN model is used, the predicted efficiency can be the instantaneous efficiency of the motor.

[0286] The predicted motor efficiency may indicate the likelihood of the motor failing. As the motor becomes more inefficient, the probability of the motor failing may increase. In this manner, the predicted motor efficiency may be used to predict when the motor is predicted to fail. In some embodiments, a threshold efficiency value may be used to define a motor failure. For example, a motor may be considered to have failed if the predicted efficiency is below 50%, 30%, 20%, etc. In this manner, trends in the change in motor efficiency may be identified and projected into the future to predict when the motor efficiency will be predicted to fall below the threshold efficiency and the motor will be considered to have failed. Similar predictions may be made for inverter efficiency and inverter failure.

[0287] Based on the predicted efficiency of the motor (and / or the predicted time the motor will fail), a determination can be made as to what corrective action, if any, to initiate to address the efficiency. For example, if the motor is predicted to be operating at perfect efficiency (i.e., 100% efficiency), the determined corrective action may be to record the predicted efficiency. As another example, if the motor is predicted to be operating at 50% efficiency, a corrective action may be scheduled to schedule maintenance on the motor to increase its efficiency.

[0288] 23, a block diagram of a motor efficiency controller 2300 is shown, according to some embodiments. The motor efficiency controller 2300 can be configured to predict the efficiency of a motor used in a building system. Specifically, the motor efficiency controller 2300 can be configured to predict the efficiency of a motor of a compressor in a VRF system. However, the prediction of motor efficiency can be applied to motors in other systems (e.g., other HVAC systems) as well. In some embodiments, the functionality of the motor efficiency controller 2300 is combined with one or more controllers described herein (e.g., the oil management controller 800, the compressor vibration controller 1310, and / or the VRF fault controller 2000).

[0289] The motor efficiency controller 2300 is shown to include a communication interface 2308 and a processing circuit 2302. The communication interface 2308 may include a wired or wireless interface (e.g., a jack, an antenna, a transmitter, a receiver, a transceiver, a wired terminal, etc.) for data communication with various systems, devices, or networks. For example, the communication interface 2308 may include an Ethernet card and port for transmitting and receiving data over an Ethernet-based communication network, and / or a Wi-Fi transceiver for communicating over a wireless communication network. The communication interface 2308 may be configured to communicate over a local area network or a wide area network (e.g., the Internet, a building WAN, etc.) and may use various communication protocols (e.g., BACnet, IP, LON, etc.).

[0290] The communications interface 2308 may be a network interface configured to facilitate electronic data communication between the motor efficiency controller 2300 and various external systems or devices (e.g., training data source 2318, sensors 2320, motor 2322, user device 2324, etc.). For example, the motor efficiency controller 2300 may provide notifications describing the motor efficiency predictions to the user device 2324 via the communications interface 2308.

[0291] The processing circuit 2302 is shown to include a processor 2304 and a memory 2306. The processor 2304 may be a general purpose or special purpose processor, an application specific integrated circuit (ASIC), one or more field programmable gate arrays (FPGAs), a group of processing components, or other suitable processing components. The processor 2304 may be configured to execute computer code or instructions stored in the memory 2306 or received from another computer readable medium (e.g., CDROM, network storage, remote server, etc.).

[0292] The memory 2306 may include one or more devices (e.g., memory units, memory devices, storage devices, etc.) for storing data and / or computer code for completing and / or facilitating various processes described in this disclosure. The memory 2306 may include random access memory (RAM), read only memory (ROM), hard drive storage, temporary storage, non-volatile memory, flash memory, optical memory, or any other suitable memory for storing software objects and / or computer instructions. The memory 2306 may include database components, object code components, script components, or any other type of information structure for supporting various activities and information structures described in this disclosure. The memory 2306 may be communicatively connected to the processor 2304 via the processing circuitry 2302 and may include computer code for executing (e.g., by the processor 2304) one or more processes described herein. In some embodiments, one or more components of the memory 2306 are part of a single component. However, each component of the memory 2306 is shown independently for ease of explanation.

[0293] The memory 2306 is shown to include a training data collector 2310. In some embodiments, the training data collector 2310 is similar to and / or the same as the training data collector 1610 described with reference to FIG. 16 and / or the training data collector 2010 described with reference to FIG. 20. The training data collector 2310 can obtain training data associated with predicting motor efficiency from a training data source 2318. The training data source 2318 can include any suitable source of training data, such as a cloud database, local storage, user input, a mobile device, equipment feedback, etc. The training data can include training data describing the operation of the motor 2322 under various conditions. For example, the training data can include values ​​for the rotational speed of the motor 2322, the power supplied to the motor 2322 by the inverter 2328, the required heating / cooling load, etc. Similarly, the training data can include an estimated efficiency value of the motor. The efficiency value may be estimated automatically and / or via some other process based on detected changes to the cost associated with the motor 2322 (e.g., changes in operating costs where higher costs indicate lower efficiency).

[0294] In some embodiments, the training data collector 2310 utilizes a simulation framework to generate some and / or all of the training data. The simulation framework may model how the motor 2322 operates based on various operating conditions (e.g., temperature, heating / cooling loads, etc.). Similarly, the simulation framework may model how the efficiency of the motor 2322 degrades over time as a result of operation. In this manner, training data representative of the operation and degradation of the motor 2322 may be generated and utilized to train the AI ​​model.

[0295] The training data collector 2310 can provide the collected training data as a training data set to the model generator 2312. In some embodiments, the model generator 2312 is similar to and / or the same as the model generator 1612 and / or the model generator 2012. Using the training data, the model generator 2312 can generate an AI model (e.g., a linear RNN) that can be used to predict efficiency values ​​of a motor in a VRF system (e.g., VRF system 600, VRF system 1900, etc.). In particular, the AI ​​model can output a percentage that represents the percentage of maximum efficiency at which the motor 2322 is operating. In this case, maximum efficiency is when the input power to the motor 2322 is equal to the output power of the motor 2322 (i.e., φ out,inverter =φ in,inverter ) situation. In some embodiments, the model generator 2312 also generates the AI ​​model such that the AI ​​model also outputs a prediction of inverter efficiency. In this case, the inverter efficiency can be modeled similarly to the motor efficiency and can be based on the operating conditions of the inverter 2328 (e.g., required heating / cooling load, ambient temperature, etc.). In some embodiments, the model generator 2312 generates separate AI models for predicting motor efficiency and inverter efficiency.

[0296] With respect to an AI model configured to predict efficiency of the motor 2322, the AI ​​model may include inputs such as the rotational speed of the motor 2322, the power provided to the motor 2322 (e.g., by the inverter 2328), the ambient temperature, the required heating / cooling load, etc. Based on such inputs, the AI ​​model may output a prediction of the motor efficiency (e.g., as a percentage of maximum efficiency).

[0297] The model generator 2312 can provide the AI ​​model to a prediction generator 2314. In some embodiments, the prediction generator 2314 is similar to and / or the same as the prediction generator 1614 and / or the prediction generator 2014. The prediction generator 2314 can utilize the AI ​​model and the received values ​​of the input variables to generate a prediction of the motor efficiency. With respect to the input variables, the prediction generator 2314 can receive values ​​of the measured variables from the sensors 2320. The sensors 2320 can include any type of sensor (e.g., temperature sensor, pressure sensor, etc.) capable of measuring the value of the input required by the AI ​​model. For example, the sensors 2320 can include a temperature sensor that provides an ambient temperature value to the prediction generator 2314. In some embodiments, the sensors 2320 include a third party source that provides the required input value. For example, the sensors 2320 can include a weather service that provides an ambient temperature value.

[0298] The prediction generator 2314 can also obtain values ​​of input variables based on device feedback from devices 2326, including a motor 2322 and an inverter 2328. The device feedback can include values ​​of variables such as the rotational speed of the motor 2322, the power supply current, voltage, and the input power provided to the motor 2322 by the inverter 2328, for example.

[0299] Using the obtained values ​​of the input variables, the prediction generator 2314 can pass the obtained input variables to an AI model to generate a prediction of the motor efficiency. The prediction of the motor efficiency can be provided to a corrective action generator 2316. In some embodiments, the corrective action generator 2316 is similar to and / or the same as the corrective action generator 1616 and / or the corrective action generator 2016.

[0300] The corrective action generator 2316 can be configured to determine what corrective action, if any, to initiate based on the prediction of the motor efficiency. In particular, the corrective action generator 2316 can determine whether action should be taken to address the decrease in motor efficiency. Corrective actions that can be initiated by the corrective action generator 2316 can include, for example, alerting a user via a user device 2324 of the predicted motor efficiency, adjusting operation of the motor 2322 via a control signal, disabling the motor 2322, scheduling maintenance / replacement of the motor 2322, recording the predicted value of motor efficiency in a database, etc.

[0301] To determine what corrective action to initiate, the corrective action generator 2316 may correspond a particular efficiency range to a particular corrective action. For example, the corrective action generator 2316 may associate an efficiency value between 75% and 100% with a corrective action to notify a user of the predicted efficiency, an efficiency value between 50% and 75% with a corrective action to adjust the operation of the motor 2322 via a control signal, and an efficiency value between 0% and 50% with a corrective action to schedule maintenance and / or replacement of the motor 2322. In some embodiments, the corrective action generator 2316 utilizes several other determinations to determine what corrective action to initiate based on the predicted efficiency value.

[0302] As a result of initiating corrective action, the inefficiency of the motor 2322 can be addressed. In this manner, the frequency of situations in which the motor 2322 operates under high inefficiency can be avoided. This can increase the reliability of the VRF system, reduce the overall cost, and simplify general maintenance of the VRF system.

[0303] 24, a flow diagram of a process 2400 for predicting the efficiency of a motor in a VRF system using an AI model is shown, according to some embodiments. Although the process 2400 is described with reference to a motor in a VRF system, it should be understood that the process 2400 is similarly applicable to generating an efficiency prediction associated with an inverter in a VRF system. In some embodiments, some and / or all of the steps of the process 2400 are performed by the motor efficiency controller 2300.

[0304] Process 2400 is shown to include obtaining training data describing operating conditions of a motor of the VRF system and an associated efficiency value of the motor (step 2402). The training data may be obtained from various training data sources, such as, for example, a cloud database, local storage of the building, based on user input, via feedback from equipment, etc. The training data may include any information relevant to training an AI model to predict an efficiency value of the motor. For example, the training data may include values ​​such as the rotational speed of the motor during operation, the ambient temperature, heating / cooling load, the current provided to the motor, the voltage provided to the motor, the power provided to the motor, an estimated efficiency value of the motor, etc. In some embodiments, step 2402 is performed by the training data collector 2310.

[0305] Process 2400 is shown to include training an artificial intelligence (AI) model for predicting efficiency of the motor based on the training data (step 2404). The AI ​​model may be trained to learn correlations between a set of variables that affect the operation of the motor and corresponding efficiency values ​​of the motor. The AI ​​model trained in step 2404 may be any of a variety of different AI models. For example, the AI ​​model may be a linear RNN for assessing efficiency of the motor. In some embodiments, step 2404 is performed by the model generator 2312.

[0306] Process 2400 is shown to include obtaining data associated with the operation of the motor (step 2406). The data obtained at step 2406 may be based on the actual operating motor if a prediction of motor efficiency is desired. In particular, the data obtained at step 2406 may include values ​​of input variables to the AI ​​model generated at step 2404. Thus, the data may include values ​​of the rotational speed of the motor, the ambient temperature, the input power to the motor, etc. In some embodiments, the data may be obtained from some and / or all of the same sources used to obtain the training data at step 2402. In some embodiments, step 2406 is performed by the prediction generator 2314.

[0307] Process 2400 is shown to include predicting efficiency of the motor based on the acquired data using an AI model (step 2408). In step 2408, the data acquired in step 2406 may be passed as input to the AI ​​model to generate a prediction regarding the efficiency of the motor. In some embodiments, the predicted efficiency of the motor is provided as a percentage of maximum efficiency. In some embodiments, the predicted efficiency is provided as another metric (e.g., a rating of "good," "average," or "poor"). In some embodiments, step 2408 is performed by prediction generator 2314.

[0308] The process 2400 is shown to include determining a corrective action based on the predicted motor efficiency output by the AI ​​model (step 2410). The corrective action may be determined based on an estimated degradation state of the motor. For example, a high efficiency value (e.g., >80% efficiency, >90% efficiency, etc.) may indicate that the motor is not significantly degraded. Thus, the corrective action determined in step 2410 may be a corrective action associated with a low cost (e.g., monetary cost, time cost, etc.), such as dispatching a notification to a user device or recording the estimated efficiency value in a database. However, if the estimated efficiency is low (e.g., <50% efficiency, <40% efficiency, etc.), the corrective action determined in step 2410 may be a corrective action associated with a higher cost, such as scheduling maintenance on the motor, disabling the motor, etc. In this manner, the corrective action determined in step 2410 may be dynamically adjusted based on the estimated efficiency of the motor. In some embodiments, step 2410 is performed by a corrective action generator 2316.

[0309] Process 2400 is shown to include initiating a corrective action (step 2412). Based on the corrective action determined in step 2410, step 2412 may include determining how to initiate the corrective action and performing the steps necessary for the corrective action to be completed. For example, if the corrective action is to notify a user, step 2412 may include generating a notification and communicating the notification using a communication interface. As another example, if the corrective action is to disable a motor, step 2412 may include generating a control signal to turn the motor off and providing the control signal to the motor. In some embodiments, step 2412 is performed by corrective action generator 2316.

[0310] Neural network implementation examples 25A and 25B, examples associated with AI model structures that may be for generating predictions, according to some embodiments, are shown. The AI ​​model structures described below may be utilized in any of the various controllers and / or methods described herein. For example, the AI ​​model structures provided in FIGS. 25A-B may be utilized by the compressor vibration controller 1310, the VRF fault controller 2000, etc. It should be understood that the AI ​​model structures provided in FIGS. 25A and 25B are provided merely for example and are not intended to be limiting of AI model structures that may be utilized to generate predictions.

[0311] Referring now to FIG. 25A, an illustration of an RNN structure 2500 is shown, according to some embodiments. In some embodiments, the RNN structure 2500 is similar and / or the same as the RNN structure 900 described with reference to FIG. 9A. In some embodiments, the RNN structure 2500 may illustrate a high-level view of an LSTM model. The LSTM model is a specific artificial RNN architecture that can be used in the field of deep learning. The LSTM model can classify and process entire sequences of time series data and make predictions. Advantageously, the LSTM model can generate predictions even with lags of unknown duration between significant events in the time series data. The LSTM model can include various layers, such as, for example, a sequence input layer, one or more dropout layers, one or more fully connected layers, one or more LSTM layers, and an output layer.

[0312] As shown in Figure 25A, the RNN structure 2500 can be represented by both a condensed model structure and an "expanded" model structure. The expanded model structure can illustrate in more detail how the RNN structure 2500 preserves information over time that may affect the output, so that information is not lost entirely between time steps. In Figure 25A, x can represent the input to the RNN, U can represent the input parameters applied to x, o can represent the output of the RNN, W can represent the output parameters applied to the output of h that produces o, h can represent the primary block of the RNN including the weights and activation functions of the RNN, and v can represent information communicated between time steps.

[0313] Referring now to FIG. 25B, an illustration of an LSTM model structure 2550 is shown, according to some embodiments. The LSTM model structure 2550 can illustrate how information is preserved between time steps in an RNN. The LSTM model structure 2550 is shown to include functions f, g, i, and o that can be used to generate output to the blocks shown in FIG. 25B. The LSTM model structure 2550 is further shown to include a forget gate, an update gate, and an output gate. The forget gate can be configured to eliminate non-relevant data from being considered and stored for future time steps in the time sequence. The update gate can combine the input information and apply some operation to account for changes in the data. Finally, the output gate can determine which information is passed as output to the next time step. The LSTM model structure 2550 can include multiple blocks that pass information associated with a particular time step in the time sequence to the next time step. Advantageously, this structure allows information to be preserved and not lost between time steps, thereby increasing the accuracy of predictions for time series data.

[0314] Configuration of an Exemplary Embodiment The configurations and arrangements of the systems and methods shown in the various exemplary embodiments are merely exemplary. Although only a few embodiments have been described in detail in this disclosure, many modifications are possible (e.g., changes in size, dimensions, structure, shape, and proportions of various elements, parameter values, mounting arrangements, use of materials, colors, orientations, etc.). For example, the positions of elements can be reversed or otherwise changed, and the nature or number or positions of separate elements can be modified or changed. As such, all such modifications are intended to be included within the scope of the present disclosure. The order or sequence of any process or method steps can be changed or reordered according to alternative embodiments. Other substitutions, modifications, changes, and omissions can be made in the design, operating conditions, and arrangements of the exemplary embodiments without departing from the scope of the present disclosure.

[0315] The present disclosure contemplates methods, systems, and program products on any machine-readable medium for accomplishing various operations. The embodiments of the present disclosure can be implemented using existing computer processors or by dedicated computer processors in suitable systems for this or another purpose, or incorporated by wired systems. An embodiment within the scope of the present disclosure includes a program product comprising a machine-readable medium for carrying or having machine-executable instructions or data structures stored thereon. Such machine-readable media can be any available medium that can be accessed by a general-purpose or special-purpose computer or other machine with a processor. By way of example, such machine-readable media can include RAM, ROM, EPROM, EEPROM, CD-ROM, or other optical disk storage, magnetic disk storage or other magnetic storage, or any other medium that can be used to carry or store desired program code in the form of machine-executable instructions or data structures and that can be accessed by a general-purpose or special-purpose computer or other machine with a processor. Combinations of the above are also included within the scope of machine-readable media. Machine-executable instructions include, for example, instructions and data that cause a general-purpose computer, special-purpose computer, or special-purpose processing machine to perform a certain function or group of functions.

[0316] Although the diagram shows a particular order of method steps, the order of steps may differ from that depicted. Also, two or more steps may be performed simultaneously or with partial concurrence. Such variations depend on the software and hardware systems selected, and designer choice. All such variations are within the scope of this disclosure. Similarly, software implementations may be accomplished by standard programming techniques with rule-based logic and other logic to accomplish the various connection, processing, comparison, and decision steps.

Claims

1. 1. A controller for predicting faults in a heating, ventilation, or air conditioning (HVAC) system, comprising: inputting operational data associated with the HVAC system into a machine learning model to predict a fault classification for identifying a fault condition affecting the HVAC system; identifying an HVAC device of the HVAC system associated with the fault condition; In response to identifying the HVAC device and the fault condition, automatically initiating corrective action to address the fault condition. A controller comprising a processing circuit configured to:

2. the fault classification includes a severity metric associated with the fault condition; the severity metric indicating an impact of the fault condition on the HVAC system; The controller of claim 1 , wherein the corrective action is determined based on both the fault condition and the severity metric associated with the fault condition.

3. The processing circuitry includes: automatically initiating a first corrective action in response to the value of the severity metric being below a severity threshold; automatically initiating a second corrective action in response to the value of the severity metric exceeding the severity threshold; The controller of claim 2 configured to:

4. the first corrective action includes providing a notification to a user device; The controller of claim 3 , wherein the second corrective action includes scheduling maintenance on the HVAC system or replacement of the HVAC device associated with the fault condition.

5. The controller of claim 2 , wherein the corrective action comprises taking no action in response to the value of the severity metric being below a severity threshold.

6. the machine learning model is a recurrent neural network (RNN) model; Analyzing the operational data includes: providing a time series of values ​​of said motion data as input to said RNN model; obtaining a prediction of the fault classification as an output of the RNN model; The controller of claim 1 .

7. The controller of claim 1 , wherein the processing circuitry is further configured to generate the machine learning model using a set of simulated training data obtained from a simulation model of the HVAC system.

8. the fault classification identifies a plurality of fault conditions affecting the HVAC system; The controller of claim 1 , wherein the plurality of fault conditions are associated with a plurality of HVAC devices of the HVAC system.

9. The fault condition is: Refrigerant leaks, Frost on outdoor unit, Indoor fan clogged, Clogged indoor filter Heat exchanger clogging, Outdoor fan clogged, Motor demagnetization, or Oil leak from compressor The controller of claim 1 , wherein the at least one of

10. the machine learning model is a first machine learning model; The processing circuitry includes: predicting the fault classification of the HVAC system using the first machine learning model; using a second machine learning model to predict a severity of the fault condition identified by the fault classification; and The controller of claim 1 configured to:

11. 1. A method for predicting faults in a heating, ventilation, or air conditioning (HVAC) system, comprising: inputting operational data associated with the HVAC system into a machine learning model to predict a fault classification for identifying a fault condition affecting the HVAC system; identifying an HVAC device of the HVAC system associated with the fault condition; In response to identifying the HVAC device and the fault condition, automatically initiating corrective action to address the fault condition. A method comprising:

12. the fault classification includes a severity metric associated with the fault condition; the severity metric indicating the extent to which the fault condition is affecting the HVAC system; The method of claim 11 , wherein the corrective action is determined based on both the fault condition and the severity metric associated with the fault condition.

13. automatically initiating a first corrective action in response to the value of the severity metric being below a severity threshold; automatically initiating a second corrective action in response to the value of the severity metric exceeding the severity threshold; 13. The method of claim 12, comprising:

14. the first corrective action includes providing a notification to a user device; 14. The method of claim 13, wherein the second corrective action includes scheduling maintenance on the HVAC system or replacement of the HVAC device associated with the fault condition.

15. The method of claim 12 , wherein the corrective action comprises taking no action in response to the value of the severity metric being below a severity threshold.

16. the machine learning model is a recurrent neural network (RNN) model; Analyzing the operational data includes: providing a time series of values ​​of said motion data as input to said RNN model; obtaining a prediction of the fault classification as an output of the RNN model; The method of claim 11 , comprising:

17. The method of claim 11 , comprising generating the machine learning model using a set of simulated training data obtained from a simulation model of the HVAC system.

18. the fault classification identifies a plurality of fault conditions affecting the HVAC system; The method of claim 11 , wherein the plurality of fault conditions are associated with a plurality of HVAC devices of the HVAC system.

19. The fault condition is: Refrigerant leaks, Frost on outdoor unit, Indoor fan clogged, Clogged indoor filter Heat exchanger clogging, Outdoor fan clogged, Motor demagnetization, or Oil leak from compressor The method of claim 11 , comprising at least one of:

20. the machine learning model is a first machine learning model; The method comprises: predicting the fault classification of the HVAC system using the first machine learning model; using a second machine learning model to predict a severity of the fault condition identified by the fault classification; and The method of claim 11 , comprising:

Images