Heat exchanger system with machine learning-based optimization

Abstract

In one aspect, a heat exchanger system is provided, comprising a cooling system and a sensor configured to detect a variable of the cooling system. The heat exchanger system includes a processor circuit configured to provide the cooling system variables and a plurality of potential operating parameters to a representative machine learning model of the cooling system to estimate at least one of energy consumption, water usage, and chemical usage for the potential operating parameters. The processor circuit is further configured to determine optimal operating parameters of the cooling system that satisfy a target optimization metric based, at least in part, on at least one of energy consumption, water usage, and chemical usage estimated for the potential operating parameters.

Description

[Technical Field]

[0001] Cross-reference of related applications

[0001] This application claims the benefit of U.S. Provisional Patent Application No. 62 / 946,778, filed December 11, 2019, which is incorporated herein by reference in its entirety.

[0002]

[0002] This disclosure relates to a heat exchanger system, and more particularly to a control system for a heat exchanger system. [Background technology]

[0003]

[0003] Heat exchanger systems, such as cooling systems, can be used in a variety of applications, such as cooling process fluids from industrial processes or cooling process fluids that absorb heat from inside a building. For example, buildings utilize heating, ventilation, or air conditioning (HVAC) systems to provide desired air properties (e.g., temperature) within the building. HVAC systems may include cooling systems that remove heat from inside the building and release it into the surrounding environment. Cooling systems may also include refrigeration systems, such as those used in supermarkets, refrigeration facilities, and ice skating rinks.

[0004]

[0004] Cooling systems often utilize cooling towers to transfer heat from a hotter process fluid to cooler ambient air. Some conventional cooling towers are equipped with variable-speed fans that can be controlled to adjust the airflow across the heat exchangers within the cooling tower in order to coordinate heat transfer between the process fluid and the air. Furthermore, some cooling towers are configured to operate in different modes to meet the cooling demands of an HVAC system. For example, a cooling tower may switch between dry mode and wet mode operation to meet the demands of an HVAC system. For example, a wet mode may include a cooling tower that disperses water in the indirect heat exchanger of the cooling tower, as it utilizes evaporative cooling to cool the process fluid. A cooling tower operating in wet mode may consume water but cools the process fluid more efficiently, while a dry mode is less efficient but does not consume water. Some cooling systems include multiple cooling towers operating in series or parallel to each other to accommodate the cooling load of a building with one or more chillers. When multiple cooling towers are used, the cooling towers may be configured to switch between different operating modes, such as dry, wet, and adiabatic modes. Some cooling systems may include ice thermal storage systems or chilled water storage systems to store or accumulate cooling capacity when cooling demand and / or energy rates are low, and to utilize the stored cooling capacity when cooling demand and / or energy rates are high.

[0005]

[0005] One problem that occurs with some of these conventional cooling systems is that the cooling tower operates according to fixed rules or programming so that the cooling tower operates in a specific mode and / or the cooling tower's fan operates at a specific speed when certain conditions are present. Because these cooling systems operate according to fixed rules, they often do not operate efficiently and consume more energy and / or water than necessary. [Overview of the Initiative]

[0006]

[0006] Some cooling systems are known to use artificial intelligence to control the operation of the cooling system. However, these conventional systems tend to overlook how the components of the cooling system interact with each other and affect the overall operation of the cooling system. Furthermore, some of these cooling systems are configured to optimize energy consumption but do not take into account the consumption of water and chemicals used in the system's cooling towers. Because these systems do not take into account all the inputs used by the cooling system, they may not be able to accurately optimize the operating cost of the cooling system. [Brief explanation of the drawing]

[0007] [Figure 1A]

[0007] This is a conceptual diagram of a cooling system comprising a cooling subsystem having a cooling tower, pumps, and chillers, and Figure 1A further shows a cooling subsystem controller in communication with a master controller and a server computer. [Figure 1B]

[0008] This is a conceptual diagram of an alternative cooling tower equipped with an indirect heat exchanger having a series of serpentine tube runs. [Figure 2]

[0009] This is a flowchart of a method that includes determining one or more optimal setting points for the cooling subsystem shown in Figure 1, and the operating mode of the cooling tower using a machine learning model representing the cooling subsystem. [Figure 3A]

[0010] Figure 2 is a flowchart illustrating an embodiment of the method, which involves aggregating sensor data and using a machine learning model representing the cooling subsystem to determine one or more optimal setpoints and operating modes. [Figure 3B]

[0010] This is a flowchart of an embodiment of the method shown in Figure 2, which involves aggregating sensor data and using a machine learning model representing the cooling subsystem to determine one or more optimal setpoints and operating modes. [Figure 4]

[0011] This is a flowchart illustrating an example method for calculating the lowest temperature or pressure of the process fluid after it has left the cooling tower, used in conjunction with the methods shown in Figures 3A and 3B. [Figure 5]

[0012] This is a flowchart of a method used in conjunction with the methods in Figures 3A and 3B to calculate the minimum process fluid flow rate of a cooling subsystem, as an example. [Figure 6]

[0013] This is a flowchart of a method for calculating the maximum temperature or pressure of the process fluid after it has left the cooling tower, which is used in conjunction with the methods shown in Figures 3A and 3B. [Figure 7]

[0014] This is a flowchart of an example method for calculating the maximum process fluid flow rate of a cooling subsystem, used in conjunction with the methods shown in Figures 3A and 3B. [Figure 8]

[0015] This is a graph representation of weighted k-nearest neighbor regression as an example. [Figure 9]

[0016] This is a graphical representation of neural network regression as an example. [Figure 10]

[0017] This is a scatter plot of estimated energy consumption by a cooling subsystem as an example, predicted by machine learning models using weighted k-nearest neighbor regression and neural network regression, for a range of setpoints for the cooling tower's decoupling water temperature. [Figure 11]

[0018] Figure 10 shows a scatter plot of estimated water usage by the cooling subsystem, predicted by a machine learning model using weighted k-nearest neighbor regression and neural network regression, for a range of cooling tower departure water temperature setpoints. [Figure 12]

[0019] Figure 10 shows a scatter plot of the estimated operating costs of the cooling subsystem, predicted by machine learning models using weighted k-nearest neighbor regression and neural network regression, for a range of setpoints for the cooling tower's decoupling water temperature. [Figure 13]

[0020] A graph of the recommended leaving water temperature setpoint as an example of a cooling subsystem cooling tower to minimize energy consumption, water use, or operational cost estimated by a machine learning model using weighted k-nearest neighbor regression. [Figure 14]

[0021] A graph showing the recommended leaving temperature setpoint as an example of a cooling subsystem cooling tower to minimize energy consumption, water use, or operational cost estimated by a machine learning model using neural network regression. [Figure 15]

[0022] A flowchart of the recommended setpoints and operating modes of a cooling subsystem cooling tower not restricted by the previous setpoint and previous operating mode. [Figure 16]

[0023] A flowchart of the recommended setpoints and operating modes of a cooling subsystem cooling tower restricted by the previous setpoint and previous operating mode. [Figure 17]

[0024] A flowchart of a method including providing recommended setpoints and cooling tower operating modes of a cooling subsystem based on prediction of future states.

DETAILED DESCRIPTION

[0008]

[0025] According to one aspect of the present disclosure, a heat exchanger system with machine learning-based optimization is provided. In one embodiment, the heat exchanger system includes a cooling system with a heat generating device such as a chiller or a water source heat pump configured to transfer heat to a process fluid. The heat generating device may generate heat by removing heat from another fluid. The cooling system further includes a heat blocking device such as a cooling tower configured to remove heat from the process fluid, and a sensor configured to detect variables of the cooling system. In some embodiments, the heat blocking device includes, in addition to or instead of a cooling tower, a thermal energy storage system. For example, the thermal energy storage system may include an ice storage system or a chilled water storage system.

[0009]

[0026] The cooling system device further includes a processor circuit configured to provide variables of the cooling subsystem and several potential parameters to a machine learning model in order to estimate at least one of the energy consumption, water usage, and chemical consumption of the cooling system for potential operating parameters. The water usage of the cooling system may include, for example, the volume of makeup water added to the system, the flow rate of water circulating within the system, and / or the speed of the cooling system's water pump.

[0010]

[0027] The processor circuit is further configured, at least in part, to determine the optimal operating parameters of the cooling system that satisfy the target optimization index, based on at least one of the estimated energy consumption, water usage, and chemical consumption for the potential operating parameters. Thus, the processor circuit may select the optimal operating parameters that best satisfy the target optimization index after using multiple potential operating parameters in a representative machine learning model of the cooling system to predict how the cooling system will respond to various operating parameters.

[0011]

[0028] The target optimization indicator may be, for example, minimizing energy consumption, water use, chemical consumption, or cost. The target optimization indicator may include achieving a specific threshold or combination of thresholds. A target optimization indicator may include multiple indicators. For example, the target optimization indicator may include achieving a threshold water reduction, a threshold energy reduction, and / or a threshold cost reduction. The target optimization indicator may be used to achieve overall performance or optimization for a particular system. As another example, the target optimization indicator may be defined by limit values, such as upper limits on energy consumption, water use, and / or cost.

[0012]

[0029] The processor circuit uses predictive dynamic optimization based on historical, live, and / or future data to predict the optimization operating parameters that will achieve the target optimization metrics. In one embodiment, the optimal operating parameters include at least one of the following: the optimal operating mode of the thermal shutoff device, the optimal temperature of the process fluid that releases the thermal shutoff device, the optimal pressure of the process fluid that releases the thermal shutoff device, and the optimal flow rate of the process fluid.

[0013]

[0030] In one embodiment, the cooling system includes a pump capable of pumping process fluid from a heat generator to a thermal shutoff device. The interaction between the heat generator, the pump, and the thermal shutoff device is typically characterized by three main metrics: energy consumption, water usage, and system operating cost. A processor circuit may use a machine learning model to predict these metrics for operating conditions and parameters such as process fluid flow rate, departure process fluid temperature, and operating mode. Based on the predictions, the processor circuit can recommend optimal operating parameters for a given optimization metric, namely minimizing energy consumption, minimizing water usage, or minimizing operating cost. The processor circuit may also utilize machine learning models that take into account water and chemical usage to more accurately estimate the actual operating and maintenance costs of the cooling subsystem. Cost estimates may include chemical processing consumption, water treatment, water pollution costs, and / or associated water maintenance costs. Cost estimates may also additionally or substitutually include other maintenance costs, such as expected component wear and breakage and component replacement according to the usage schedule. The recommended optimal operating parameters may also involve optimizing the operating mode of the thermal shielding device, such as operating the cooling tower in wet, dry, hybrid, or adiabatic mode, in order to achieve the desired optimization metrics.

[0014]

[0031] In one embodiment, the processor circuit utilizes a thermal shutoff device-driven approach, in which optimization is driven by the operation and performance of the thermal shutoff device rather than by the chiller or water source heat pump. Furthermore, the processor circuit performs optimization in real time using current, historical, and / or predicted future data. In one embodiment, the processor circuit includes a memory configured to contain performance models of the heat generator, pump, and / or thermal shutoff device in order to provide factory pre-configurations for machine learning models that the processor circuit may utilize when historical data is insufficient.

[0015]

[0032] In one embodiment, the processor circuit is configured to provide a machine learning model with a plurality of potential parameters of the cooling system in order to estimate power consumption and water usage for the potential operating parameters. The plurality of potential operating parameters may include a range of operating parameters in which the cooling system can operate at and / or within the limits of the cooling system (e.g., maximum and minimum values). The processor circuit is further configured, at least in part, to determine the optimal operating parameters of the cooling system that satisfy a target optimization metric, based on the estimated power consumption and water usage. As an example, the operating parameters may be the fan speed and / or the amount of water distributed by the water distribution system of the cooling system.

[0016]

[0033] In one embodiment, the sensor includes one or more sensors configured to detect malfunctions in the cooling system. The processor circuit is configured, at least in part, to determine a plurality of potential operating parameters to provide at least one representative machine learning model of the cooling system based on the detected malfunction. The processor circuit then, at least in part, determines the optimal operating parameters of the cooling system that satisfy a target optimization metric based on the detected malfunction. For example, if a variable-speed fan in a cooling tower is malfunctioning and remains "on" at a fixed speed, the potential operating parameters provided to the machine learning model may include the fixed speed, and if the fan is fully operational, the potential operating parameters may include a range of potential fan speeds.

[0017]

[0034] The cooling system may include various components that perform heat exchange. In one embodiment, the heat-generating device includes a chiller, and the heat-insulating device includes a cooling tower. In another embodiment, the cooling system further includes an air handling unit operably connected to the chiller, and a pump configured to pump process fluid between the chiller and the cooling tower.

[0018]

[0035] In one embodiment, the thermal shutdown device includes a heat storage device such as an ice or chilled water storage system. The processor is configured to store or accumulate energy in the heat storage device and release the stored energy to cool the process fluid. The heat storage device may store energy when energy costs are low, and release and use energy when energy costs are high (during peak energy usage time) to cool the process fluid. The processor circuit may determine one or more optimal operating parameters for the operation of the heat storage device as part of determining the optimal operating parameters of the cooling system that satisfy target optimization metrics.

[0019]

[0036] This disclosure also provides a thermal shutoff device for a cooling system. The thermal shutoff device comprises a cooling tower having an evaporative heat exchanger operable to cool a process fluid. The thermal shutoff device comprises a sensor configured to detect variables of the cooling tower and a controller operablely coupled to the sensor. The controller is configured to implement optimal operating parameters of the cooling tower to satisfy target optimization metrics. Optimal operating parameters may include, as some examples, the operating mode of the cooling tower, fan speed, departure process fluid temperature, departure process fluid pressure, and / or evaporative liquid distribution rate.

[0020]

[0037] The optimal operating parameters are determined by providing the cooling tower's variables detected by sensors and several potential operating parameters to a representative machine learning model of the cooling tower, at least in part, in order to estimate power consumption and water usage for the potential operating parameters. Thus, since the optimal operating parameters implemented by the controller are determined by estimating power consumption and water usage for multiple potential operating parameters, rather than by the following fixed rules, the cooling tower can be made more efficient during operation.

[0021]

[0038] This disclosure also provides a method for operating a cooling system. This method includes receiving cooling system variables detected by sensors in the cooling system in a processor associated with the cooling system. This method includes providing the cooling system variables and a plurality of potential operating parameters to a representative machine learning model of the cooling system in order to estimate at least one of energy consumption, water use, and chemical use for the potential operating parameters. This method includes, at least in part, determining the optimal operating parameters of the cooling system that satisfy a target optimization metric, based on at least one of the estimated energy consumption, water use, and chemical use for the potential operating parameters. This method further includes enabling the cooling system to utilize the optimal operating parameters. In one embodiment, the processor is a component of a master controller for a building's HVAC or industrial system. In another embodiment, the processor is a component of a cloud-based computing system, and enabling the cooling system to utilize the optimal operating parameters includes communicating the optimal operating parameters to the cooling system over a network such as the Internet.

[0022]

[0039] With respect to Figure 1A, a cooling system 10 is provided, which is part of the building's HVAC system. The cooling system 10 comprises one or more air handling units 12 located within the building, and at least one cooling subsystem 14, which includes a cooling tower 16, a chiller 18, and a pump 20 configured to circulate process fluid between the cooling tower 16 and the chiller 18, in order to block heat from entering the environment. Although the cooling subsystem 14 is a cooling system in itself, it is referred to as a subsystem for the purposes of the description in Figure 1A because it is a component of the overall cooling system 10.

[0023]

[0040] In one embodiment, the cooling system 10 further comprises a pump 23 capable of pumping a process fluid such as water or water / glycol between the chiller 18 and the air handling unit 12. The cooling system 10 may have various configurations, including one or more bypass valves, a water source heat pump in place of the chiller 18, and various types of condensers or fluid cooling devices. The cooling system 10 may also comprise other devices such as intermediate heat exchangers between the chiller and the cooling tower, between the chiller and the air handling unit, and / or between the cooling tower and the air handling unit. As a further example, in a refrigeration system, the process fluid may be ammonia, and the cooling tower may be an air-cooled, adiabatic, hybrid, or evaporative condenser that condenses ammonia from gas to liquid. The process fluid is flowed or pumped to cool the process or building cooling in which the process fluid evaporates, prior to being directed towards the chiller 18 and the tower.

[0024]

[0041] In one embodiment, the cooling tower 16 includes an airflow generator such as a fan assembly 21 containing a fan 22, a motor 24, a fluid distribution system 26, and one or more heat exchange elements 28 such as direct or indirect heat exchangers. For example, the cooling tower 16 may use water as the thermal barrier liquid, and the evaporative fluid distribution system 26 injects water directly onto the heat exchanger, which usually contains a filler, and the airflow generated by the fan 22 cools the water, and the cooled water is collected in a wastewater tank 30.

[0025]

[0042] In another embodiment, the cooling tower 16 may utilize a process fluid moving through a coil of an indirect heat exchanger or heat exchange element 28, and the fluid distribution system 26 indirectly cools the process fluid in the coil by injecting a heat-insulating liquid, such as water, onto the coil. The injected water is collected in a sump 30 and pumped back to the fluid distribution system 26. The cooling subsystem 14 includes, as an example, a makeup water supply that provides makeup water into the sump 30 to replenish water lost by evaporation.

[0026]

[0043] Referring to Figure 1B, for example, the cooling tower 10A may operate humidly in humid mode or evaporative mode, partially humidly in hybrid mode, or in dry mode with the injection pump 12A turned off when ambient conditions or lower loads allow. In some embodiments, the cooling tower may additionally or alternatively operate in adiabatic mode, in which case the air is cooled adiabatically in the process of evaporating water, changing the air from dry-bulb temperature to a value closer to wet-bulb temperature while the heat exchanger itself operates without evaporation.

[0027]

[0044] The dry, wet, hybrid, and adiabatic modes of operation of a cooling tower reflect the operating characteristics of the cooling tower. In dry mode, the cooling tower may have indirect heat exchangers with only reasonable heat transfer to the air, without water injection to the indirect heat exchanger surface facing the air. In wet mode, the cooling tower may have fully wet direct or indirect heat exchangers with direct water-air potential and reasonable thermal insulation to the external direct / indirect heat exchanger surface. In hybrid mode, the cooling tower may have a combination of wet and dry heat exchangers in a single package (e.g., in series and / or parallel) to allow for better control of the cooling tower's water and energy consumption. In adiabatic mode, the cooling tower may have two heat exchangers in series, which are typically direct heat exchangers with water-air contact to pre-cool the air before it enters the dry heat exchanger section. In adiabatic mode, the cooling tower may have the ability to control energy and water use by turning the water supply on and off.

[0028]

[0045] The injection pump 12A receives the coldest cooled evaporative injection fluid, usually water, from the chilled wastewater tank 11A and pumps it up to the primary injection water header 19A, where the water exits through a nozzle or orifice 17A and distributes to the indirect heat exchanger 14A. The injection water header 19A and nozzle 17A function to distribute the water uniformly to the top of the indirect heat exchanger 14A. Once the coldest water has been distributed to the top of the indirect heat exchanger 14A, the motor 21A of the fan assembly 22B causes the fan 22A of the fan assembly 22B to swirl, which guides or draws in ambient air through the inlet louvers 13A through the indirect heat exchanger 14A, and then through a flow deflector 20A that functions to prevent the flow deflection from leaving the unit, after which the warmed air is blown into the environment. The air usually flows in the opposite direction to the falling injection water. Figure 1B shows an axial fan 22 guiding or drawing air through a unit, and the fan system may be any style of fan system that moves air through a unit, which typically includes, but is not limited to, traction guiding or forced into a reverse, cross, or parallel flow relative to the injection. The motor 21A may be a variable speed motor capable of rotating the fan 22 at a variable speed. Furthermore, the motor 21A may be belt-driven, gear-driven, or directly connected to the fan, as shown. The indirect heat exchanger 14A is shown with an inlet connecting pipe 15A connected to an inlet header 24A and an outlet connecting pipe 16A connected to an outlet header 25A. The inlet connecting pipe 15A may receive a process fluid such as water from a chiller such as a chiller 18, and the outlet connecting pipe 16A may direct water to a pump such as a pump 20 (see Figure 1A). The relative positions of the inlet header 24A and the outlet header 25A may be swapped or otherwise configured according to a particular process fluid and a particular installation. The inlet header 24A connects to the inlets of the multiple serpentine tube circuits of the indirect heat exchanger 14A, and the outlet header 25A connects to the outlets of the multiple serpentine tube circuits. The serpentine tube run 18B connects to the return bend section 18A.The return bend section 18A may be formed continuously with the serpentine tube run 18B of the circuit, or it may be welded between the runs 18B.

[0029]

[0046] With respect to Figure 1A, pump 20 directs cooled water from cooling tower 16 along cooling process line 32 to water-cooled condenser 34 of chiller 18, where the water receives heat from chiller 18. The water then moves along hot process fluid line 36 and returns to cooling tower 16, such as fluid distribution system 26. In one embodiment, chiller 18 includes evaporator 40, compressor 42, and expansion valve 45 which works with condenser 34 to remove heat from chilled water supply 44 from heat exchanger 46 of air handling unit 12. Pump 23 pumps water from chiller 18 along chilled water fluid return line 48 which proceeds to heat exchanger 46 of air handling unit 12.

[0030]

[0047] The cooling system 10 may be part of a building's HVAC system controlled by a master controller 50. The master controller 50 may be connected to, or part of, a building automation system, a building management system, or other building or process system or industrial process. The master controller 50 may control the operation of the cooling system 10 and the heating system. The cooling system 10 includes a cooling subsystem controller 52 operably connected to a cooling subsystem 14 and configured to control the operation of at least one of the pumps 20, cooling towers 16, and chillers 18. The cooling subsystem controller 52 may be operably connected to the master controller 50 and operate the cooling subsystem 14 according to instructions from the master controller 50. The cooling subsystem controller 52 includes memory 60, a processor 62, and a communication circuit 64. The communication circuit 64 may communicate with the master controller 50, a server computer 54, and / or a user device 58 via wired and / or wireless approaches. The cooling subsystem controller 52 may communicate with a master controller, a server computer 54, and / or a user device 58 via one or more networks 56. The networks 56 may be interconnected or separate, for example. Exemplary networks 56 include, for example, a local Wi-Fi network, a cellular network, and the internet. The user device 58 may be, for example, a smartphone, a smartwatch, a personal computer, a laptop computer, or an in-car display. The user device 58 includes a user interface 59 that allows the user to monitor and / or adjust the operation of the cooling subsystem 14. The user interface 59 may include, for example, at least one of a screen, a touchscreen, a microphone, a speaker, a haptic feedback generator, a hologram, or an augmented reality display.

[0031]

[0048] In some embodiments, the cooling system 10 may comprise a plurality of pumps 20, cooling towers 16, and / or chillers 18. For example, the cooling system 10 may comprise two or more cooling towers 16 operating in parallel, so that each cooling tower 16 receives process fluid from the chiller 18 and returns the process fluid to the chiller 18. In another example, the cooling towers 16 operate in series so that after a first cooling tower 16 receives process fluid from the chiller 18, the process fluid flows to at least one other cooling tower 16 before leaving the first cooling tower 16 and returning to the chiller 18. In some embodiments, the cooling system 10 comprises a plurality of cooling towers 16 operating both in series and in parallel with each other. These cooling towers 16 may be dry cooling towers, wet cooling towers configured to switch between wet mode operation and dry mode operation, or a combination of a plurality of types of cooling towers 16. The cooling system 10 may include a plurality of chillers 18 within the building that supply process fluid to one or more cooling towers 16 of the cooling system 10 for cooling. A master controller 50, a cooling subsystem controller 52, and / or a server computer 54 may be configured to control the operation of the cooling system 10 and its components.

[0032]

[0049] In some embodiments, the cooling system 10 further comprises a thermal storage system. The thermal storage system may comprise a cooled storage medium for storing energy to be used later by the cooling system 10. Examples of thermal storage systems include ice thermal storage systems and chilled water thermal storage systems. For example, an ice thermal storage system may create ice to store energy and then melt the ice to help the cooling system 10 cool later. For example, an ice thermal storage system may help the cooling tower 16 cool the process fluid from the chiller 18 so that the cooling tower 16 reduces its energy consumption. The thermal storage system may be operated in a partial thermal storage mode, in which case the thermal storage system assists the cooling tower 16 when cooling the process fluid from the chiller 18, or in a full thermal storage mode when the cooling tower 16 is not operating and the thermal storage system provides all the cooling. The thermal storage system may operate to store energy (e.g., by making ice) when energy costs are low or during off-peak hours and release energy (e.g., by melting ice) when energy costs are high or during peak hours. Therefore, the running costs of the cooling system 10 may be further minimized by using a heat storage system.

[0033]

[0050] With respect to Figure 1A, the cooling subsystem controller 52 may communicate information about the cooling subsystem 14 with the master controller 50. As will be discussed in more detail below, the cooling subsystem controller 52 may analyze the current environment and operating conditions of the cooling subsystem 14 and / or predict future environment and operating conditions of the cooling subsystem 14 in order to provide one or more recommended parameters to the master controller 50. The master controller 50 may instruct the cooling subsystem controller 52 to control the cooling subsystem 14, at least in part, based on the recommended parameters from the cooling subsystem controller 52. In another embodiment, the cooling subsystem controller 52 implements the recommended parameters independently of the master controller 50.

[0034]

[0051] The server computer 54 comprises a processor 70, a communication circuit 72, and electronic storage or memory 74. The server computer 54 includes hardware, software, and / or firmware that operate to provide the operability described herein. The processor 70 may include at least one of a digital processor, a digital circuit designed to process information, and software. The processor 70 may include a single processor or multiple processors. The processors may be located in the same or different computers, such as a cloud of the server computer. The memory 74 may include, for example, optical storage, a magnetically readable storage medium, random access memory, and / or other electronic storage media.

[0035]

[0052] As an example, the cooling subsystem controller 52 communicates data from one or more sensors of the cooling subsystem 14 to the server computer 54, and the processor 70 deploys one or more machine learning models 151 (see Figure 3B) that represent the relationships between the environment and variables of the cooling subsystem 14. The machine learning models are input with potential operating parameters of the cooling subsystem 14 to obtain energy, chemical, and / or water usage estimated by the cooling subsystem 14. The server computer 54 may receive data from the cooling subsystem 14 at different facilities to generate more accurate machine learning algorithms. The machine learning models may be stored in memory 74 and / or memory 60 and may be utilized by processor 70 and / or processor 62. As another example, the cooling subsystem controller 52 and / or master controller 50 may deploy one or more machine learning models 151.

[0036]

[0053] In one embodiment, the processor 70 modifies one or more machine learning models 151 over time, utilizing reinforcement learning and self-tuning to make the machine learning models 151 more accurate as more historical data is collected from the cooling subsystem 14 and other cooling subsystems of the installation. The processor 70 may tune several factors that can be tuned, such as data collection rate, optimization frequency, and / or model hyperparameters. The processor 70's ability to modify one or more machine learning models 151 over time improves the autonomy and agnostic capability of the machine learning models 151. Reinforcement learning may include comparing predicted variables with measured variables and making reward and / or action decisions based on the difference between the predicted and measured variables. The processor 70 may automatically self-tune one or more machine learning models 151 at fixed or variable intervals, such as hourly, daily, or weekly. The processor 70 may self-tune one or more machine learning models 151 in response to events such as user requests or measured parameters exceeding thresholds, as some examples. Self-adjustment may include determining, for example, which coefficients should be used as input to one or more machine learning models 151, and / or which sensor data should be used. The processor 70 may also determine which machine learning models 151 to use, such as using several machine learning models 151 initially, and then using only one machine learning model 151 once a sufficient amount of historical data has been accumulated for the cooling subsystem 14. As another example, the cooling subsystem controller 52 and / or the master controller 50 may self-adjust one or more machine learning models 151.

[0037]

[0054] With respect to Figure 2, in one embodiment, the master controller 50, the cooling subsystem controller 52, and / or the server computer 54 perform a method 80 which includes determining one or more optimal control settings or operating parameters, such as one or more setpoints and cooling tower operating modes, in order to achieve specific target optimization indicators for the cooling subsystem 14. Target optimization indicators for the cooling subsystem 14 may include, for example, minimizing energy consumption, minimizing water consumption, minimizing chemical water treatment, and / or minimizing the operating costs and maintenance of the cooling subsystem 14. Another target optimization indicator is minimizing CO2 / greenhouse gas emissions, which may depend on energy consumption and energy sources (e.g., natural gas, hydroelectric power, wind power, etc.). Method 80 provides recommended optimal control actions, which are based on the current state of the cooling subsystem 14, and may perform recommended optimal control actions to optimize the operation of the cooling subsystem 14.

[0038]

[0055] Method 80 recognizes that the operation of each component of the cooling subsystem 14 affects the other components. Method 80 provides a holistic approach to deliver the desired operation of the cooling subsystem 14 by deploying a machine learning model of the cooling subsystem 14 that recognizes the interdependence of the components of the cooling subsystem 14. In one approach, the machine learning model utilizes input variables that have been determined to be important for accurately estimating the operation of the cooling subsystem 14, such as the variables shown in Figure 3A.

[0039]

[0056] With respect to Figure 2, Method 80 includes collecting data from sensors of the cooling subsystem 14 82 and providing the potential operating parameters to at least one machine learning algorithm to estimate energy and water consumption based on a plurality of potential parameters of the provided cooling subsystem 14 84. Method 80 further includes, at least in part, determining recommended or optimal operating parameters for the cooling system 10 based on the estimated energy and water consumption 86. Optimal parameters may include one or more optimal setpoints and / or optimal operating modes for one or more components of the cooling system 10. Optimal parameters may include turning one or more components of the cooling system 10 on or off. Optimal parameters for the cooling system may be parameters that achieve a target optimization metric for the cooling subsystem 14, such as minimizing energy consumption, minimizing water consumption, minimizing water treatment chemical consumption, or minimizing operating costs.

[0040]

[0057] The aggregation of data from sensors in the cooling subsystem 14 82 includes aggregating variables of the cooling subsystem 14, such as collecting sensor data and setpoints for the cooling subsystem 14 90. The sensor data and setpoints may include one or more representative variables of, for example, the cooling load (building load, etc.), chillers, water source heat pumps (WSHP), compressors, pumps, and thermal shutoff equipment. The sensor data may also include one or more malfunctions detected by one or more sensors of the cooling subsystem 14. Aggregation 82 further includes collecting sensor data for one or more environment variables 92. The environment variables may include, for example, the dry-bulb air temperature, relative humidity, wet-bulb temperature, date, time, utility costs (e.g., electricity and water), and / or the cost of water treatment chemicals used within the cooling subsystem 14.

[0041]

[0058] Providing the operation 84 may include providing cooling subsystem variables and environment variables to one or more machine learning models of the cooling subsystem 14 94. The one or more machine learning models may include, for example, machine learning models utilizing weighted k-nearest neighbor regression (wk-NN), decision tree regression (DT), and / or neural network regression (NN). The machine learning models may be updated in real time. The update frequency, data accumulation period, and optimization frequency may be fixed or variable.

[0042]

[0059] The aggregation 82 and / or provision 84 may include processing the aggregated data for use in one or more machine learning models. Processing may include cleaning and normalizing the data, such as addressing outliers, missing data, and resolving timestamp issues. Processing may make the aggregated data functional or actionable. For example, the sensor sampling rate may be 1 second, the data aggregation operation may have a duration of 15 minutes, and processing may include averaging data collected over a period of more than 15 minutes.

[0043]

[0060] The cooling subsystem controller 52 may have one or more default machine learning models for the cooling subsystem 14 preloaded, to be selected by the installer during the installation of the cooling subsystem 14. The preloaded machine learning models provide a rough model of the cooling subsystem 14. For example, the installer may provide the cooling subsystem controller 52 with creations and models for the cooling tower 16, pumps 20, and chillers 18, and the processor 62 retrieves machine learning models for energy and water consumption from memory 60 for the identified cooling subsystem 14. One or more preloaded machine learning models may be refined over time using measured environment variables and operating variables and the corresponding behavior of the cooling subsystem 14, including energy and water consumption. In another approach, once a model individualized for the cooling subsystem 14 is deployed using historical data that meets an accurate threshold, the preloaded machine learning models are no longer used.

[0044]

[0061] The cooling subsystem controller 52 may be configured to detect actual and / or estimated anomalies during the operation of the cooling subsystem 14. The cooling subsystem controller 52 may compare actual and / or estimated operational data with historical data. When the cooling subsystem controller 52 detects an anomaly, it may send a warning to the master controller 50, the server computer 54, and / or the user device 58. The cooling subsystem controller 52 may send a warning if the magnitude of the anomaly, such as the temperature of a fluid or component, exceeds a maximum threshold. The maximum threshold may be set, for example, by the user, by the manufacturer, or based on the output of at least one other machine learning model, such as a clustering algorithm. As another example, the cooling subsystem controller 52 may send a warning if a large number of anomalies occur within a given period. Some examples of warnings may include email, application notifications, service phone calls, and / or SMS messages. Alternatively or additionally, the cooling subsystem controller 52 may adjust one or more components of the cooling subsystem 14 to address the anomaly.

[0045]

[0062] Providing 84 includes utilizing one or more machine learning models 96 to estimate the operation of the cooling subsystem 14, such as energy and / or water consumption, based on a number of potential parameters for the cooling subsystem 14. Each of the potential parameters provided to the one or more machine learning models includes minimum and maximum values ​​corresponding to the actual minimum and maximum values ​​allowed by the cooling subsystem 14. Thus, the one or more machine learning models are limited to providing potential parameters that the cooling subsystem 14 can act upon, or within the operational limitations of the cooling subsystem 14 and / or the cooling system 10.

[0046]

[0063] Determining 86 may include providing at least one optimal operating parameter for the cooling subsystem 14, such as setpoints and / or operating modes for one or more components of the cooling subsystem 14, according to the target optimization index 98. In one form, determining 86 includes selecting at least one optimal operating parameter for the cooling subsystem 14 from one or more optimal operating parameters for the cooling subsystem 14 predicted by a machine learning model to satisfy the cooling requirements of the cooling subsystem 14. Selecting may include selecting at least one optimal operating parameter based on the target optimization index. The cooling subsystem controller 52 then implements the optimal parameter 99. Implementing 99 may include adjusting one or more components of the cooling subsystem 14 to operate according to the provided optimal parameter.

[0047]

[0064] Further details of method 80 are shown with respect to Figures 3A and 3B. In one embodiment, aggregation 82 includes collecting sensor data 100 that indicates one or more environment variables and one or more operating variables of the cooling subsystem 14. Environment variables may include, for example, air dry-bulb (DB), atmospheric pressure, and relative humidity (RH) variables 110 collected by the temperature sensor 77 (see Figure 1A) and humidity sensor 79 of the cooling tower 16. Aggregation may include identifying at least one time-related variable, such as time, date, month, and season.

[0048]

[0065] Sensor data relating to the operating variables of the cooling subsystem 14 may include an entering process fluid temperature (EPFT) variable 104, which can be collected by one or more sensors on the hot process fluid line 36, such as one or more thermistors 36A. The sensor data may further include energy consumption variables 106 for each of the individual components of the cooling subsystem 14, such as chillers, water source heat pumps, compressors, condenser water pumps, and cooling towers. For cooling towers, the energy consumption variable 160 may include the energy consumption of one or more fans of the cooling tower, and in some embodiments, the energy consumption of the injection water pumps. The energy consumption variable 106 may be measured directly from each component by one or more sensors that measure the current and / or voltage used by the component. The energy consumption variable 106 may be measured in units of kilowatts (kW), for example. The sensor data may further include departure process fluid variables 108, such as temperature and / or pressure, collected using one or more thermistors 32A and / or sensors such as pressure sensors 32B on the cooling process fluid line 32. The sensor data may also include a makeup water flow variable 112, which may be detected using a flow meter that monitors the flow rate of makeup water piped into the wastewater tank 30. In some examples, the flow rate may be an instantaneous flow rate measurement or the total water consumption output by the meter over time. The makeup water flow variable 112 may include a blowdown flow, which can be measured or calculated in some applications.

[0049]

[0066] Sensor data for the operating variables of the cooling subsystem 14 may include process fluid pump variables 114 such as the flow rate of the process fluid generated by the pump 20 (e.g., gallons / minute (GPM)) and the speed of the pump 20. In one embodiment, the pump 20 is flow-adjustable. For example, the pump 20 may be driven at a variable frequency, and the speed of the pump 20 may be determined by measuring the power frequency of the pump 20. In another application, the speed of the pump 20 may be fixed, and the process fluid flow rate may be a constant value. In another example, the cooling subsystem 14 does not include a pump 20, and the process fluid pump variables 114 are not utilized. Instead, refrigerant mass flow rate measured or calculated from compressor speed, compressor energy consumption, condensation temperature, and / or condensation pressure may be utilized.

[0050]

[0067] The sensor data may further include fluid distribution system variables 116 such as the status (on, off, speed, and / or pressure) and / or flow rate of the injection pump of the fluid distribution system 26. The sensor data may further include chemical consumption variables 118 such as the amount of chemicals added per hour to the makeup water supplied to the cooling subsystem 14. Chemicals commonly used in the cooling system include corrosion inhibitors (e.g., bicarbonates) to neutralize acidity and protect metal components, and algaecides and biocides (e.g., bromine, chlorine, ozone, hydrogen peroxide, bleach) to reduce the growth of microorganisms and biofilms. In addition, scale inhibitors (e.g., phosphoric acid) may be added to prevent contamination from forming scale deposits. The chemical consumption variables 118 may be determined, for example, by a scale weighing a container containing chemicals added to the circulating process fluid of the cooling subsystem 14. The aggregation 82 may also include receiving pricing data 119 for energy, water, and / or chemicals. Pricing data 119 may be fluctuating current “live” pricing data, or it may be a set value if current pricing data is not available.

[0051]

[0068] The sensor data may also include fault variables for components, such as the status of one or more components of the cooling subsystem 14 (fully operational, limited capacity, or malfunction). For example, one or more sensors in system 10 may detect whether one or more components of the cooling subsystem 14 are no longer functioning or are in an error state that prevents them from being operated by the cooling subsystem controller 52. Components may cease to function if they are missing or require service or repair. Components may also be considered non-functional when they are in an error state. Components may enter an error state if they detect the existence of specific circumstances that prevent them from operating. For example, if the fan 22 of the cooling tower 16 exceeds a predetermined temperature, the sensor may determine that the fan 22 is in an error state and cannot operate until the fan 22's temperature falls below the predetermined temperature. Components may be determined to have limited capacity when certain circumstances exist. For example, if the fan 22 of the cooling tower 16 is approaching a certain temperature, the fan 22 may be configured not to operate above a certain speed. As another example, components may have limitations on their operating time. For example, fan 22 may be configured not to operate at a set speed for more than 10 hours a day in order to extend the lifespan of fan 22. When fan 22 is approaching or has reached its operating time limit, sensor data may indicate that the fan 22's capabilities are limited, i.e., it cannot operate at a set speed. As will be discussed below, component failure variables may be used to determine potential operating parameters on which the cooling subsystem 14 operates to provide to a machine learning model.

[0052]

[0069] As described above, aggregation 82 includes deriving variables 102 from one or more of the sensor data collected in operation 100. For example, the cooling load 120 may be derived from the entry process fluid temperature variable 104, the exit process fluid variable 108, and the process fluid pump variable 114. The derived 102 parameters may further include a system energy consumption variable 122, which may be the sum of the energy consumption variables 106 for all components of the cooling subsystem 14. The derived 102 variables may further include the wet-bulb (WB) temperature 124, which may be directly measured or derived from the dry-bulb, atmospheric pressure, and relative humidity variables 110. The relative humidity variable may be replaced by a direct wet-bulb measurement. The derived 102 variables may further include an approach variable 126, which is the difference between the exit process fluid and the entry wet-bulb temperature.

[0053]

[0070] Referring to Figure 3B, in one embodiment, one or more representative machine learning models 151 of the cooling subsystem 14 include machine learning models for system water consumption 150 and system energy consumption 152. Providing 84 includes providing the machine learning models for system water consumption 150 and system energy consumption 152 with a number of potential parameters, such as a range of potential parameters. In one embodiment, providing 84 includes circulating values ​​for departure process fluid temperature (LPFT) and / or pressure and process fluid flow rate through potential parameters, including the operating mode of the cooling tower 16 (wet, dry, hybrid, or adiabatic), in order to calculate the system energy, water consumption, and operating cost for all possible combinations of potential parameters 160, such as all possible combinations of potential parameters. If the cooling system 10 comprises multiple cooling towers 16, providing 84 may include circulating values ​​through potential parameters, including the operating status or mode (e.g., on, off, wet, dry, adiabatic, etc.) for each of the cooling towers 16 and / or potential configurations (e.g., series, parallel, or a combination thereof). If the cooling system 10 includes a heat storage system, providing 84 may include circulating 154 through the potential operating modes of the heat storage system.

[0054]

[0071] The potential parameters used in the circulating (154) operation reflect the capabilities of the cooling subsystem 14. For example, the possible operating modes of a cooling tower 16 may be limited by the operating modes permitted by the cooling tower 16. As another example, there may be a cooling tower 16 that is only capable of dry operation, while other cooling towers 16 are capable of dry or wet operation, and yet another cooling tower 16 is capable of dry, wet, hybrid, or adiabatic operation. Furthermore, the process fluid temperature leaving the cooling tower 16 may be limited to the highest or lowest return temperature permitted by the chiller 18, and the process fluid flow rate may be limited by the lowest and highest flow rates of the cooling tower 16. As yet another example, the potential parameters may be limited to the components of the cooling subsystem 14 that are currently operational and not malfunctioning. For example, if the fan 22 of the cooling tower 16 is malfunctioning, the potential parameters will reflect the fact that the fan 22 cannot operate. Method 80 may include determining the maximum and minimum values ​​of the potential parameters that can be used in the circulating (154) operation, as will be discussed in more detail below with reference to Figures 4 to 7.

[0055]

[0072] In one embodiment, one or more machine learning models 151 also include a chemical consumption machine learning model 156. The chemical consumption machine learning model 156 may directly estimate the use of chemicals by means of sensors associated with the chemicals, such as a digital scale. In an alternative approach, the chemical consumption machine learning model 156 indirectly estimates chemical consumption by utilizing water consumption predicted by a machine learning model for system water consumption 150 and the estimated chemical consumption rate (e.g., kilograms per gallon of water).

[0056]

[0073] The machine learning models for system water consumption 150, system energy consumption 152, and chemical consumption 156 may each include one or more machine learning models that utilize different types of modeling algorithms. For example, the machine learning models for system water consumption 150, system energy consumption 152, and chemical consumption 156 may each utilize weighted k-nearest neighbor regression (wk-NN) as shown in Figure 8, and / or neural network regression (NN) as shown in Figure 9, which will be discussed in more detail below. When historical data is limited, such as immediately after the installation or repair of components of the cooling subsystem 14, the cooling subsystem controller 52 may utilize manufacturer default data for one or more components of the cooling subsystem 14 to provide a rough guide for the machine learning model 151 when estimating the operation of the cooling subsystem 14. If the reliability of the model prediction is low, or to ensure that the system operates as expected by the manufacturer, manufacturer default data may also be used in conjunction with the machine learning model 151. Machine learning models for system water consumption 150, system energy consumption 152, and chemical consumption 156 could be, for example, cooling subsystem-level models and / or specific models for each piece of equipment in the cooling subsystem 14.

[0057]

[0074] Providing 84 includes calculating the system energy and water consumption and operating costs for a number of potential cooling subsystem parameters, provided for a machine learning model for system water consumption 150, system energy consumption 152, and system chemical consumption 156. The calculated water consumption may not include blowdown, and concentrated cycle (CoC) calculations may be used for water usage estimation.

[0058]

[0075] With respect to Figure 3B, determination 86 may include searching 170 for the optimal operating mode of the cooling tower 16 and the optimal setpoints for the temperature of the process fluid leaving the cooling tower 16, the pressure of the process fluid leaving the cooling tower 16, and / or the flow rate of the process fluid. Searching 170 may further include searching for the optimal combination of cooling towers 16 to be turned on / off or operated in series / parallel / combined configurations, in which case the cooling system 10 includes more than one number of cooling towers 16. Searching 170 adjusts the search based on desired or target optimization metrics for the cooling subsystem 14, such as minimizing energy consumption, minimizing water consumption, minimizing water treatment chemicals, or minimizing operating costs. Different optimization metrics may yield different results for a given operating condition of the cooling subsystem 14. For example, in geographically scarce locations, minimizing the operating costs for the cooling subsystem 14 may include reducing water consumption for a given environmental and building load, while in geographically richer areas with the same environmental and building load, more water may be used. As another example, minimizing the operating costs for the cooling subsystem 14 may result in higher energy consumption of the components of the cooling subsystem 14 during the earlier part of the day when energy is cheaper, and lower energy consumption later in the day when energy is more expensive.

[0059]

[0076] As another example, the master controller 50, the cooling subsystem controller 52, and / or the server computer 54 may select optimization metrics in response to user input from a user device 58, etc., or events such as requests for energy consumption and water consumption. Examples in this regard include adjusting energy consumption to accommodate the available supply of a renewable energy source (e.g., solar power) and adjusting water consumption during a drought. In one embodiment, the master controller 50 may receive communications from a utility provider indicating available power and / or water. In response to these communications, the master controller 50 may temporarily disable optimization metrics for the cooling subsystem controller 52 provided by the user or the master controller 50.

[0060]

[0077] As another example, the target optimization indicator may be scheduled for a specific time and may vary based on the time of day, day of the week, or month. For example, the target optimization indicator may be scheduled to minimize energy consumption during peak energy usage times, but may switch to minimizing water usage at night.

[0061]

[0078] Another example of an event-dependent target optimization index is when the master controller 50, cooling subsystem controller 52, and / or server computer 54 change their target optimization index from minimizing water consumption to minimizing energy consumption when the cooling subsystem 14 consumes its daily allocated water. Once the daily allocated water is consumed, the cooling towers of the cooling subsystem may need to operate in dry mode. The target optimization index may remain focused on minimizing energy consumption until the next day when the target optimization index is reset to minimize water consumption. As another example, the event triggering the change in the target optimization index may be a determination by a resource conservation algorithm that the target optimization index must change in order to conserve limited resources (e.g., water from a renewable energy source, and / or electricity). The resource conservation algorithm may use a rank-based voting method to determine how to utilize limited resources based on past, present, and projected future environmental and load conditions.

[0062]

[0079] Another example of how target optimization metrics change depending on the event is when the cooling subsystem 14 is configured to minimize CO2 or greenhouse gas emissions. The master controller 50, the cooling subsystem controller 52, and / or the server computer 54 may receive data on the current amount of CO2 / kWh of electricity on the grid. The amount of CO2 / kWh may fluctuate daily based on the energy source that powers the grid from which the cooling subsystem 14 draws power. If the amount of CO2 / kWh falls below a predetermined threshold, the system may be configured to switch from minimizing energy consumption to minimizing cost or water consumption. Alternatively, if the amount of CO2 / kWh exceeds a certain threshold, the system may be configured to switch to minimizing energy consumption in order to reduce the amount of CO2 / greenhouse gases that the system effectively emits.

[0063]

[0080] Another example involves switching between different target optimization metrics based on the real-time, i.e., current cost of each resource used by the cooling system 10. For example, the master controller 50, the cooling subsystem controller 52, and / or the server computer 54 may receive data providing real-time, scheduled, and / or predicted costs of water and / or energy. The system may also take into account the incentive for peak load reduction provided by utilities with reduced real-time costs / kW. The cooling subsystem 14 may be configured to minimize water consumption as long as the cost of energy does not exceed a certain predetermined threshold. If it is determined that the current cost of energy is above the predetermined threshold, the system may switch to minimizing energy consumption. If it is determined that the price of energy is below the predetermined threshold, the system may switch to minimizing water consumption. Similarly, if it is determined that the cost of water is above a certain predetermined threshold, the system may switch to minimizing water consumption.

[0064]

[0081] Another example involves switching the target optimization index based on boundary parameters set for the cooling system 10 equipment. If the target optimization index requires one or more components of the cooling system 10 to operate outside the boundary parameters to meet the cooling load, the target optimization index may be switched to operate within the boundary parameters set for the cooling system 10 to meet the cooling load. For example, in some applications, a limit may be set on the operating speed of the chiller 18. As an example, the chiller 18 may be set to operate within a preferred operating range (e.g., a speed between 40% and 85%). If the recommended operating parameters for the cooling system 10 require the chiller 18 to operate outside the preferred operating range, the target optimization index may be changed to allow the chiller 18 to operate within the preferred operating range. As another example, the pumps or fans of the cooling system 10 may have a limited operating time, or may be set not to exceed a certain operating time at a speed above a certain limit. Therefore, if the recommended operating parameters for a particular target optimization index require the equipment to operate outside the operating time limit, the target optimization index may be changed to comply with the operating time limit.

[0065]

[0082] As yet another example, the master controller 50, the cooling subsystem controller 52, and / or the server computer 54 may be configured to switch target optimization indicators if the chiller 18 cannot meet its setpoint due to the cooling tower 16 operating to meet a specific target optimization indicator. For example, if the cooling tower 16 is configured to minimize water consumption, and the chiller 18 cannot meet its cooling water temperature setpoint while the cooling tower 16 is operating to minimize water consumption, the target optimization indicator may be switched to minimizing energy consumption or cost so that the chiller 18 can meet its setpoint.

[0066]

[0083] Even when the objectives are to minimize energy consumption and water consumption, different results can be obtained. For example, minimizing energy consumption may result in the cooling subsystem controller 52 providing a higher optimal parameter for a given process fluid flow rate for a given environmental and building load than the process fluid flow rate provided when the objective of minimizing water consumption is used (98). Specifically, if the target minimization index is minimizing water consumption rather than minimizing energy consumption, the cooling subsystem controller 52 may provide a lower optimal parameter for the process fluid flow rate but a faster speed for the fan 22 of the cooling tower 16. It will be recognized that different system operating temperatures, air temperatures, humidity, and system designs can lead to different optimal parameters.

[0067]

[0084] Determining 86 may further include providing or returning one or more optimal parameters for the cooling subsystem 14 to achieve a target optimization metric, such as minimized energy consumption, minimized water consumption, or minimized operating cost. One or more optimal parameters may include the optimal operating mode of the cooling tower 16, the temperature of the process fluid leaving the cooling tower 16, the pressure of the process fluid leaving the cooling tower 16, and / or the process fluid flow rate. As an example, returning 172 may include returning the wet operation of the cooling tower 16 and a specific frequency, or speed, or flow rate for the variable frequency drive of the pump 20.

[0068]

[0085] With respect to Figures 2 and 3B, action 99 may include adjusting one or more components of the cooling subsystem 14 173. For example, if the detachment process fluid temperature of the cooling tower 16 is higher or lower than the optimal parameter currently detected, adjustment 173 may include increasing the speed of fan 22 to lower the detachment process fluid temperature, or decreasing the speed of fan 22 to raise the detachment process fluid temperature. As another example, if the detachment process fluid pressure of the cooling tower 16 is higher or lower than the optimal parameter currently detected, the detachment process fluid pressure would be lowered by increasing the speed of fan 22, or raised by decreasing the speed of fan 22. Alternatively or additionally, adjustment 173 may include changing the operating mode of the cooling tower 16 to achieve stepwise changes in the detachment process fluid temperature and detachment water pressure of the cooling tower 16. More specifically, if the cooling tower 16 is operating in dry mode at 50% fan speed, switching the cooling tower 16 to wet mode while maintaining 50% fan speed will significantly reduce the departure process fluid temperature and / or departure process fluid pressure. The departure process fluid temperature and / or departure process fluid pressure may be further adjusted by increasing or decreasing the fan speed in the new operating mode of the cooling tower 16. As yet another example, given a fan speed and ingress process fluid temperature in the cooling tower 16, increasing the speed of the pump 20 to increase the water flow rate will increase the departure process fluid temperature, and decreasing the speed of the pump 20 to decrease the water flow rate will decrease the departure water temperature. As yet another example, if the cooling system 10 includes a heat storage system, the heat storage system may be switched to a full or partial heat storage release mode to adjust the assistance that the heat storage system provides to the cooling tower 16 when cooling the process fluid.

[0069]

[0086] With respect to Figure 4, providing multiple latent parameters to the machine learning model 151 84 includes providing maximum and minimum values ​​for each latent parameter corresponding to the cooling subsystem 14. The latent parameters are limited in each case by limitations of the cooling subsystem 14, such as maximum and minimum return water temperatures.

[0070]

[0087] Determining the minimum values ​​of potential parameters that can be provided to the machine learning model 151 (84) may include a method 200 for calculating the lowest temperature and / or pressure of the process fluid that will leave the thermal cutoff device. Method 200 includes collecting relevant sensor data and derived parameters 202, as discussed above with reference to operations 100, 102 and Figure 3A. Method 200 includes determining the lowest acceptable and achievable process fluid temperature and / or pressure for the heat receiving device based on the expected heat capacity 204. Determining 204 may include a user input or calculation of the lowest acceptable chiller or water source heat pump return process fluid temperature. Alternatively, determining 204 may include the lowest condensation temperature. Determining 204 results in a lowest temperature A represented by reference numeral 206.

[0071]

[0088] Method 200 further includes determining the minimum process temperature and / or pressure for a thermal shutoff device 208. For example, determining 208 may include calculating the possible minimum cooling tower or fluid cooler detachment process fluid temperature or pressure. Determining 208 results in a variable B represented by reference numeral 210. Method 200 further includes comparing variables A and B 212. If variable A is greater than variable B, Method 200 then includes setting a minimum detachment process fluid temperature and / or pressure for variable A 214. If variable A is less than or equal to variable B, Method 200 then includes setting a minimum detachment process fluid temperature and / or pressure for variable B 216.

[0072]

[0089] With respect to Figure 5, determining the minimum values ​​of potential parameters that can be provided to the machine learning model 151 (84) may include a method 250 for calculating a desired minimum process fluid flow rate for the cooling subsystem 14. Method 250 includes collecting relevant sensor data variables and derived variables 252 and determining the minimum process fluid flow rate for the heat receiving device 254. For example, determining 254 may include a user input or calculation of an acceptable minimum process fluid flow rate for a chiller or water source heat pump. Method 250 may further include determining 256 the minimum process fluid flow rate for a thermal shutoff device. Determining 256 may include calculating an acceptable minimum cooling tower or fluid cooler process fluid flow rate. Examples of fluid cooling include, for example, the PF series, FXV series, HXV, and TCFC series fluid coolers from Baltimore Aircoil, Inc. of Jessup, Maryland. Determining 256 may result in different minimum process fluid flow rates, for example, depending on whether the cooling tower 16 can operate in dry, wet, hybrid, and / or adiabatic modes. Method 250 includes determining 257 a minimum process fluid pump flow rate that may be set according to data supplied by the pump manufacturer.

[0073]

[0090] As a result of determining 254, 256, variables A258, B260, and C261 are obtained. Method 250 includes setting the minimum process fluid flow rate 262 to be equal to one of variables A258, B260, and C261. Setting 262 includes setting the minimum process fluid flow rate to variable A258 if variable A258 is greater than variable C261, setting it to variable C261 if variable A258 is less than or equal to variable C261, setting it to variable B260 if variable B260 is greater than variable C261, and setting it to variable C261 if variable B260 is less than or equal to variable C261.

[0074]

[0091] With respect to Figure 6, determining the maximum value of potential parameters that can be provided to the machine learning model 151 (84) may include a method 300 for calculating the maximum temperature and / or pressure of the process fluid after leaving the thermal cutoff device. Method 300 includes collecting relevant sensor data variables and derived variables 302 and determining the maximum process fluid temperature and / or pressure of the heat receiving device 304. Determining 304 may include user input or calculation of the maximum acceptable process fluid temperature and / or pressure for the chiller, water source heat pump, or condenser. Method 300 may further include determining 306 the maximum process fluid temperature and / or pressure of the thermal cutoff device. Determining 306 may further include user input or calculation of the maximum acceptable process fluid temperature and / or pressure for the cooling tower or fluid cooler. As another example, determining 306 may include using a constant offset from the inlet water temperature (if the range is kept constant) or from the inlet air wet-bulb temperature (if the approach is kept constant). Determining 304, 306 results in variables A308 and B310. Method 300 includes setting the maximum detachment process fluid temperature and / or pressure equal to variable B if variable A308 is greater than variable B 312. Method 300 includes setting the maximum detachment process fluid temperature and / or pressure equal to variable A308 if variable A308 is less than or equal to variable B310 314.

[0075]

[0092] With respect to Figure 7, determining the maximum value of potential parameters that can be provided to a machine learning model may include a method 350 for calculating the maximum process fluid flow rate. Method 350 includes collecting relevant sensor data variables and derived variables 352 and determining the maximum process fluid flow rate for a heat receiving device 354. Determining 354 may include a user input or calculation of the maximum acceptable process fluid flow rate for a chiller or water source heat pump. Method 350 further includes determining the maximum process fluid flow rate for a heat shutoff device 356. Determining 356 may include a user input or calculation of the maximum acceptable process fluid flow rate for a cooling tower or fluid cooler. Method 350 includes determining the maximum process flow rate 357, which may be set according to data supplied by the pump manufacturer. Determining 354, 356, and 357 results in variables A358, B360, and C361.

[0076]

[0093] Method 350 includes setting the maximum process fluid flow rate to be equal to variable C361 if variable B360 is greater than variable C361, setting it to variable B360 if variable B360 is less than or equal to variable 361, setting it to variable C361 if variable C361 is less than or equal to variable A358, and setting it to variable A358 if variable C361 is greater than variable 358 (362).

[0077]

[0094] With respect to Figures 8 and 9, the machine learning models 151 for system water consumption 150, system energy consumption 152, and chemical consumption 156 may each encompass one or more machine learning models. For example, the machine learning models for water consumption 150, energy consumption 152, and chemical consumption 156 each include multiple machine learning models, including a first machine learning model using weighted k-nearest neighbor regression (wk-NN) 400 shown in Figure 8 and a second machine learning model using neural network regression (NN) 450 shown in Figure 9.

[0078]

[0095] Regarding Figure 8, the wk-NN regression 400 is shown to be trained with values ​​that correlate the building load 402 on the x-axis and the energy consumption 404 of the cooling subsystem 14 on the y-axis. Figure 8 is an example; in the application, refer to Figure 3A and consider one or more of the parameters described above. Given input x1...x n In Model 400, the k-nearest neighbors (e.g., k=4) are found. Next, the wk-NN regression 400 uses the input x1...x n Output values ​​y1···y n To predict this, a weighted mean is calculated based on the k-nearest neighbor distance. The historical data used to train the wk-NN regression 400 may include current data along with data from previously collected sensor data. Thus, for a given building load value 402, a machine learning model using the wk-NN regression 400 can provide an estimated energy consumption 404 for the cooling subsystem 14. Water consumption may be estimated using a similar approach.

[0079]

[0096] With respect to Figure 9, the neural network (NN) regression 450 generates a neural network of relationships between one or more inputs 452 and outputs 454. To model the relationship between inputs 452 and outputs 454, the NN regression 450 utilizes historical data from the cooling subsystem 14, and includes a hijacking layer 454, h1(1)···h x (n) and output layer 456, f 1 ···f n The following is expanded. Output 454 could be, for example, the energy consumption of the cooling subsystem 14. In this example, the load on the cooling subsystem 14, the dry-bulb air temperature, the wet-bulb air temperature, and the temperature of the water leaving the cooling tower 16 are provided as input 452, and the system energy consumption is provided as output 454. Therefore, a machine learning model using neural network (NN) regression 450 on a given load, dry-bulb air temperature, wet-bulb air temperature, and water leaving temperature can provide an estimated energy consumption output 454 for the cooling subsystem 14. A similar approach may be used to model / predict water consumption.

[0080]

[0097] As an example, referring to Figures 3B and 10 to 12, calculation 160 includes calculating the energy consumption (Figure 10), water consumption (Figure 11), and operating cost (Figure 12) of the cooling subsystem 14 for possible combinations of the operating mode of the cooling tower, the temperature and pressure of the process fluid leaving the cooling tower 16, and the flow rate of the process fluid, using machine learning models with wk-NN regression 400 and NN regression 450. Possible combinations may be all or fewer of the possible combinations of the possible parameters of the operating mode, the temperature and pressure of the leaving process fluid, and the flow rate of the process fluid. As discussed above with reference to Figures 4 to 7, each potential parameter has minimum and maximum values ​​that reflect the components of the cooling subsystem 14.

[0081]

[0098] With respect to Figures 10 to 12, the cooling subsystem controller 52 provides the water consumption machine learning model 150 and the system energy machine learning model 152 with a temperature range of process fluid leaving the cooling tower 16, for example, a possible departure water temperature setpoint 502, in order to estimate the energy consumption 500, water consumption 501, and operating cost 503 of the cooling subsystem 14 for a range of departure water temperature setpoints 502. The scatter plots in Figures 10 to 12 graphically represent the estimated energy consumption 500, water consumption 501, and operating cost 503 for the possible departure water setpoints 502, as predicted by either wk-NN regression 400 or NN regression 450 for each of the machine learning models 150 and 152. The estimates in the scatter plots may be generated, for example, every hour, to determine whether to adjust the cooling system 14 according to the current conditions. The model used to determine the operating cost 503 may utilize the estimated energy consumption 500, the estimated water consumption 501, and the energy and water costs included in the pricing data 119.

[0082]

[0099] With respect to Figures 3B and 10 to 12, determination 86 includes searching for optimal operating parameters for the cooling subsystem 14 by providing machine learning models 150 and 152 with a dewatering temperature in the range of 69°F to 84°F. Searching 170 may include searching for estimated energy consumption 500, estimated water consumption 501, and estimated cost 503 for the minimum value, and determining the dewatering temperature setpoint 502 corresponding to the minimum value. The estimated energy consumption 500, estimated water consumption 501, and estimated cost 503 can be determined based on the estimated energy consumption and estimated water consumption used by the cooling subsystem 14 when implementing the operating parameters using a representative machine learning model of the cooling subsystem 14. For example, in Figure 10, the minimum energy consumption 504 predicted by the energy consumption machine learning model 152 using wk-NN regression 400 occurs at a dewatering temperature setpoint of 75°F. The minimum energy consumption predicted by the energy consumption machine learning model 152 using NN regression 450 occurs at a set point of 74°F for the detached water temperature 506.

[0083]

[0100] Next, the cooling subsystem controller 52 may adjust, for example, the operating mode of the cooling tower 16, the status of the pumps of the fluid distribution system 26, the speed of the fan 22, and / or the speed of the pump 20, so that the cooling subsystem 14 has a desired departure water temperature setpoint of 75°F in order to achieve the minimum energy consumption predicted by the energy consumption machine learning model 152 using wk-NN regression 400. In this example, the energy consumption machine learning model 152 using wk-NN regression 400 may have higher confidence than the energy consumption machine learning model 152 using NN regression 450. Alternatively, if the energy consumption machine learning model 152 using NN regression 450 has higher confidence, the cooling subsystem controller 52 may adjust the components of the cooling subsystem 14 to achieve a desired departure water temperature setpoint of 74°F. As yet another example, the cooling subsystem controller 52 may operate the components of the cooling subsystem 14 to achieve a setpoint of detached water temperature determined by a weighted average of detached water temperatures of 74°F and 75°F, such that weights are assigned to temperature based on the confidence interval of the associated machine learning model 152.

[0084]

[0101] With respect to Figure 11, the water consumption machine learning model 150 is used to estimate the water consumption of the cooling subsystem 14 for a dewatering temperature range of 69°F to 84°F. The water consumption learning model 150 using wk-NN regression 400 estimates a minimum water consumption of 552 at a dewatering temperature of 76°F, while the water consumption learning model 150 using NN regression 450 estimates a minimum water consumption of 550 at a dewatering temperature of 75°F. To achieve the objective of optimizing to minimize the water consumption of the cooling subsystem 14, the cooling subsystem controller 52 may adjust, for example, the operating mode of the cooling tower 16, the status of the pumps of the fluid distribution system 26, the speed of the fan 22, and / or the speed of the pump 20 to cause the cooling subsystem 14 to achieve a dewatering temperature setpoint of 76°F. The cooling subsystem controller 52 may similarly adjust the components of the cooling subsystem 14 to achieve a dewatering temperature setpoint of 75°F if the water consumption machine learning model 150 using NN regression 450 has higher confidence. As an alternative, the optimal departure water temperature setpoint may be calculated as a weighted average of the values ​​75°F and 76°F.

[0085]

[0102] With respect to Figure 12, the operating cost 503 is calculated using the estimated water consumption 500, estimated water consumption 501, and the energy and water costs for the detachment water temperature range of 69°F to 84°F. The operating cost 503 estimated by machine learning models 150 and 152 using wk-NN regression 400 estimates a minimum operating cost of 582 at a liquid water temperature setpoint of 75°F. The operating cost 503 estimated by machine learning models 150 and 152 using NN regression 450 estimates a minimum operating cost of 580 at a temperature of 74°F. The cooling subsystem controller 52 may adjust the control settings of the components of the cooling subsystem 14 to achieve a desired detachment water temperature setpoint of 75°F based on wk-NN regression 400, a desired detachment water temperature setpoint of 74°F based on NN regression 450, or setpoints derived from 75°F and 74°F, in order to achieve a target optimization index that minimizes the operating cost of the cooling subsystem 14.

[0086]

[0103] Comparing Figures 10, 11, and 12, it is clear that the water and energy consumption machine learning models 150 and 152, utilizing wk-NN regression 400 and NN regression 450, can provide different recommended dewatering water temperature setpoints depending on whether the target optimization metric is minimizing water consumption, energy consumption, or operating costs. Thus, Method 80 optimizes the operation of the cooling subsystem 14 to suit the desired optimization objective.

[0087]

[0104] The cooling subsystem controller 52 may perform method 80 continuously or periodically. As some examples, all or part of method 80 may be performed seasonally, weekly, monthly, daily, every 12 hours, every 4 hours, every hour, every 15 minutes, and / or every 30 seconds. The sampling rate and optimization frequency may vary over time and may be parameters adjusted to achieve an optimization metric. For example, the optimization frequency may be adjusted between every hour and every two hours to determine the optimization frequency that best achieves a desired optimization metric. The optimization frequency may be adjusted, for example, by the user, by predefined rules, and / or autonomously.

[0088]

[0105] In one embodiment, the cooling subsystem controller 52 collects data for a period of 15 minutes (82), provides the determined optimal parameters (84), makes a determination (86), and implements them (99), and repeats this process every hour. The cooling subsystem controller 52 may perform method 80 according to a schedule. Alternatively or additionally, the cooling subsystem controller 52 may perform method 80 in response to events such as the ambient temperature or building temperature exceeding or falling below a threshold, or falling outside a predetermined temperature range.

[0089]

[0106] The cooling subsystem controller 52 continuously determines optimal parameters based on the changing environment and operating conditions of the cooling subsystem 14 (86). Referring to Figure 13, the test was conducted using an example cooling subsystem controller 52 to analyze data for a 60,000 square foot building in North America over a 24-hour period, using a 200-ton cooling tower with a 5 hp fan motor, two 7.5 hp pumps operating simultaneously, and a 200-ton chiller with a 100 hp motor. Figure 13 is graph 600 of the dewatering water temperature setpoint recommendations 607 by the cooling subsystem controller 52 over time (608), as determined by water and energy consumption machine learning models 150, 152 using wk-NN regression 400. Graph 600 shows the variation in the dewatering water setpoint recommendations 607 over a 24-hour period. Graph 600 was generated using one day's worth of data from the building's cooling subsystem.

[0090]

[0107] The different lines in Graph 600 show estimates of recommended de-water temperature setpoints 607 for achieving target optimization metrics such as minimizing energy consumption 602, minimizing water consumption 604, or minimizing operating costs 606. Graph 600 includes a fixed approach 612 calculated using a standard improved rule-based controller, where the minimum de-water temperature setpoint is limited to the chiller's capacity.

[0091]

[0108] Prior to 601, the cooling subsystem 14 is turned off, and since the fixed approach 612 is always calculable and not based on operating conditions, all setpoints other than the fixed approach 612 are turned off so that their values ​​are the same.

[0092]

[0109] Once the cooling subsystem 14 is turned on at 601, the water and energy consumption models 150 and 152 can now begin receiving data and start making recommendations every hour using a 15-minute sampling rate. Recommendations 602, 604, and 606 will initially be close to each other and differentiate from each other after the cooling subsystem 14 is turned on at 601, and will stop changing when the cooling subsystem 14 is turned off at 601A.

[0093]

[0110] In the test reflected in Figure 13, the actual decoupling water temperature 607 was kept constant at 77°F for illustrative purposes while the cooling system controller 52 calculated the recommended decoupling water temperature setpoint 607. In other words, the cooling subsystem controller 52 calculated the recommended decoupling water temperature setpoint 607 but did not adjust the components of the cooling subsystem 14. This was done to provide a baseline from which the optimization recommendations 602, 604, and 606 could be observed.

[0094]

[0111] In Graph 600, the optimization recommendations 602, 604, and 606 change very frequently, highlighting the need for dynamic optimization. Specifically, the system cooling load and ambient conditions fluctuate frequently, and energy and water costs can also fluctuate dynamically. The large fluctuations in the optimization estimates 602, 604, and 606 highlight the responsiveness of the Method 80 model to rapid changes such as sunrise and solar load spikes on the building. The first rapid change 614 in the optimization recommendations 602, 604, and 606 is due to sunrise and people arriving at the building (approximately 7:00 AM to 9:00 AM). The rest of the morning is relatively stable because the sun shines from one side of the building and the ambient temperature is stable. The second rapid change 616 occurs in the early afternoon. People are returning from lunch, the sun is fully up and shining mostly on the building where there is the least amount of shade due to windows. These factors increase the load on the cooling subsystem 14. The building is at its maximum capacity, the afternoon load is highest because the sun is up for a long time, heating the building, and the ambient air temperature is at its highest. The third sharp change 618 is linked to people leaving the building at the end of the workday and the setting sun.

[0095]

[0112] In Graph 600, optimization recommendations 602 and 606 are quite close to each other because the cost of water at the test site was much lower than the cost of energy. Contrary to common industrial practice, the performance curves of the components of the cooling subsystem 14 are highly nonlinear, so minimizing water consumption, energy consumption, or operating costs does not necessarily lead to minimizing energy use. In Graph 600, the minimum energy optimization recommendation 602 and the minimum cost optimization recommendation 606 are quite close because water is quite inexpensive at the test site. However, Figures 10 and 11 show that the rate of increase or decrease in water and energy consumption with increases or decreases in the dewatering water setpoint is very different. This indicates that, based on the relative costs of energy and water, the optimal dewatering water temperature setpoint can be distorted toward the minimum energy line if energy is more expensive than water (and chemicals), distorted toward the minimum flow rate if the opposite is true, or distorted somewhere in between.

[0096]

[0113] With respect to Figure 14, Graph 650 is provided, which shows example departure temperature setpoint recommendations 651 over time 653, determined by water and energy consumption machine learning models 150, 152 using NN regression 450. Graph 650 is based on the same test data as Graph 600, but different regression approaches used in different figures result in different departure temperatures being recommended. The different lines in Graph 650 show the recommendations for departure temperature setpoints 651 to achieve the target optimization metrics of minimizing energy consumption 652, minimizing water consumption 654, or minimizing operating costs 656. Graph 650 includes a constant departure temperature setpoint 658 and a fixed approach 660 calculated using a standard improved rule-based controller where the minimum departure temperature is limited to the chiller's capacity.

[0097]

[0114] The recommended spikes and values ​​for 652, 654, and 656 differ from the optimized estimates for 602, 604, and 606 because, although the modeling approaches differ, the overall trends are similar.

[0098]

[0115] Comparing FIGS. 13 and 14, it is shown that the recommended outlet water temperature of the water and energy consumption machine learning models 150, 152 varies depending on whether w-k-NN regression 400 or NN regression 450 is used.

[0099]

[0116] With respect to FIGS. 3B and 15, calculating 160, searching 170, and returning 172 the optimal parameters may not be restricted by previous parameters. For example, method 700 shows the optimal parameters 702 recommended at a time t-1 Calculating 160, searching 170, and returning 172 provide the current recommended optimal parameters 704 at a time, including the process fluid flow rate 706, the outlet process fluid temperature 708, and the operating mode 710. The process fluid flow rate 706 is between the minimum process fluid flow rate and the maximum process fluid flow rate of the cooling subsystem 14 regardless of the process fluid flow rate 706A at a time t Similarly, the outlet process fluid temperature 708 is between the lowest outlet process fluid temperature and the highest outlet process fluid temperature of the cooling subsystem 14 regardless of the outlet process fluid temperature 708A. Further, the optimal operating mode 710 is determined without being restricted by the operating mode 710A. t-1 With reference to FIG. 16, in another embodiment, one or more of calculating 160, searching 170, and returning 172 the optimal parameters may be restricted by past parameters. For example, method 750 includes providing 752 optimal parameters including the process fluid flow rate 754, the outlet process fluid temperature 756, and the operating mode 758 at a time. Method 750 includes the process fluid flow rate 762, the outlet process fluid temperature 764, and the operating mode 766 at a time

[0100]

[0117] With reference to FIG. 16, in another embodiment, one or more of calculating 160, searching 170, and returning 172 the optimal parameters may be restricted by past parameters. For example, method 750 includes providing 752 optimal parameters including the process fluid flow rate 754, the outlet process fluid temperature 756, and the operating mode 758 at a time. Method 750 includes the process fluid flow rate 762, the outlet process fluid temperature 764, and the operating mode 766 at a time t-1 Calculating 160, searching 170, and returning 172 provide the current recommended optimal parameters 704 at a time, including the process fluid flow rate 706, the outlet process fluid temperature 708, and the operating mode 710. The process fluid flow rate 706 is between the minimum process fluid flow rate and the maximum process fluid flow rate of the cooling subsystem 14 regardless of the process fluid flow rate 706A at a time tThis includes calculating the optimal parameters in 160, searching for them 170, and returning them 172. However, the change in flow rate between 762 and 754 is limited to a predetermined ΔPF flow rate 768 to avoid instability. Furthermore, the difference between the detachment process fluid temperature 764 and the detachment process temperature 756 is limited to a predetermined ΔLPFT 770 to avoid instability. Thus, the maximum and minimum process fluid flow rates and detachment process fluid temperatures are constrained to a range 774 determined by the aforementioned operating values ​​752. Furthermore, the operating mode 766 is limited by the operating mode 758 to a predetermined frequency of operating mode changes (772) to avoid instability. Constraints on changes from past optimal parameters can be set by user-inputted limits or limits learned based on past data of the cooling subsystem 14. In some embodiments, the process fluid flow rate 762 and the detachment process fluid temperature 764 may be replaced with detachment refrigerant temperature or pressure for condenser applications.

[0101]

[0118] Referring to Figure 17, a method 800 is provided that is similar in many respects to method 80 discussed above, as the differences are highlighted. Method 800 provides one or more estimated optimal parameters for the cooling subsystem 14, based on predictions of the future state of the cooling subsystem 14 rather than the current state.

[0102]

[0119] Method 800 includes collecting variables of the cooling system 14 804 and including collecting environment variables 806, which includes variable aggregation 802. Aggregation 802 further includes collecting weather forecast data such as dry-bulb temperature, wet-bulb temperature, rainfall, and solar radiation forecast 808. Aggregation 802 may also include identifying at least one time-related variable such as hour of day, day, month, and season. Method 800 includes estimating future operating conditions of the cooling subsystem 14 810. Estimation 810 includes utilizing a machine learning model for building load and energy cost forecasting, which may be similar to the machine learning model described above for estimating energy and water consumption 812. One potential difference may be the input parameters. In the case of load forecasting, the input parameters may include at least one of the following: hour of day / week / year, weather data (current and forecast), and live occupancy data. For energy cost forecasting, the input parameters may include at least one of the following: time of day / week / year, and weather data (current and forecast). Estimation 810 further includes defining the future operating state of the system 814, such as estimating the operating variables of the cooling subsystem 14 at a specific date and time in the future. Defining 814 may be similar to the approach discussed above with reference to Figure 15, but instead of proceeding to (t-1) to (t), method 800 includes using data at (t) or (t-1) to predict state (t+n), such as using load forecasts, weather forecasts, and recommended setpoints / modes. In practice, method 800 may anticipate changes in operating conditions and make proactive changes to avoid operating suboptimally in order to compensate for sudden changes that may occur in future operating conditions.

[0103]

[0120] Using this approach, the model takes into account the predicted future operation of the cooling system, and consequently, the operation of the cooling system over a longer period, which can further improve the achievement of the target optimization index. The cooling system does not only consider which setting will, as a result, achieve the target optimization index at a particular moment, but also uses the predicted future operation of the cooling system to inform how the cooling system should operate now. For example, if the current weather conditions are hot and sunny, but the weather forecast includes sudden rainfall at ambient temperatures and rain over several hours, the cooling system may reduce the cooling provided by predicting the cooler ambient temperature and rain in the future, for example, to conserve energy and water use. As another example, the cooling system may operate in a dry region where water use is limited by government regulations. The cooling system may be allocated a specific number of gallons of water to be used throughout the day. By predicting the future operating conditions of the cooling system, method 800 may include determining, based on the predicted cooling load of the cooling system, when the cooling system should use the limited water supply throughout the day. Determining when water will be used may be based not only on current and / or past conditions, but also in part on the target optimization indicator and how best to achieve that indicator throughout the day. Thus, method 800 may predict future operating conditions of the cooling system and update the currently implemented control settings accordingly.

[0104]

[0121] As another example, the cooling system 10 may use predicted or forecasted energy and / or water cost data to guide its operation. For example, knowing that energy costs or water costs will increase in the future, the cooling system 10 may be made to optimize its operation based on past, present, and forecasted operating parameters and conditions. For example, if the cooling system 10 includes thermal energy storage such as an ice thermal storage system, the cooling system 10 may be configured to consume energy to make ice while keeping energy costs low, and to release or use the energy stored in the ice to cool in a way that reduces energy consumption from the grid when energy costs are high. By using predicted or forecasted energy costs, the cooling system 10 can predict future changes and update its current operating parameters.

[0105]

[0122] Method 800 further includes providing a plurality of potential operating parameters to one or more machine learning models, such as water and energy use machine learning models similar to Model 151 discussed above 820. Providing 820 may include providing a water and energy use machine learning model for the cooling subsystem 14 822. The water and energy use machine learning model may utilize environment variables and cooling subsystem variables for future operating states defined in operation 814. Each potential parameter may be within the range of minimum and maximum values ​​for a potential parameter corresponding to a defined future state of the cooling subsystem 14.

[0106]

[0123] Providing 820 further includes performing a search 824 for one or more optimal operating parameters of the cooling subsystem 14 based on a defined future state of the cooling subsystem 14, using a method similar to that of the search 170 described above. For example, the search 824 may include searching for the minimum values ​​of energy consumption, water consumption, and operating costs estimated by a water and energy consumption machine learning model.

[0107]

[0124] Method 800 further includes determining one or more optimal operating parameters for the cooling subsystem 14 based on target optimization indicators such as minimizing water consumption, minimizing energy consumption, or minimizing operating costs (830). The cooling subsystem controller 52 may implement one or more recommended optimal parameters so that the cooling subsystem 14 operates to achieve the target optimization indicators at a given date and time in a specified future state (832). In one embodiment, Method 800 may include anticipating changes in operating conditions and making proactive changes to the cooling subsystem 14 to avoid suboptimal operation in order to compensate for potential abrupt changes in future operating conditions. For example, Method 800 may include pre-cooling the relevant building several hours before people enter the building in the morning, and / or preemptively reducing system capacity in anticipation of lunch breaks and / or the end of the workday. The decision to pre-cool the building may be partly driven by rising energy costs later in the day. Alternatively or additionally, Method 800 may include storing thermal energy in the thermal energy storage system when the system load is low and releasing thermal energy in the thermal energy storage system when the load on the cooling system is high, such as when many people are entering and leaving the building. As another example, the building may be set to a first temperature (e.g., 70°F) for a specific time of day (e.g., 8 a.m. to 5 p.m.) and a second temperature (e.g., 75°F) for the rest of the day (5 p.m. to 8 a.m.). By anticipating changes in the building's temperature point, the cooling subsystem controller 52 may implement changes to achieve the target optimization indicator over an extended period rather than at that moment. For example, continuing with the above example, if, as a result, Method 800 is used and a machine learning model predicts that the building temperature will be within an acceptable range from the first temperature setpoint by 5 p.m., then the cooling by the cooling system may be reduced from 4:30 p.m. onward, anticipating the change in the building's temperature setpoint at 5 p.m.

[0108]

[0125] The use of singular terms such as "a" and "an" is intended to cover both singular and plural forms unless otherwise stated herein or it is clear from the context. Terms such as "equip," "have," "include," and "contain" should be understood as open-ended terms. The phrase "at least one of" as used herein is intended to be interpreted disjunctively. For example, the phrase "at least one of A and B" is intended to cover A, B, or both A and B.

[0109]

[0126] While specific embodiments of the present invention have been illustrated and described, those skilled in the art will understand that numerous changes and modifications are possible, and the present invention is intended to cover all such changes and modifications within the scope of the appended claims.

Claims

1. A system equipped with a cooling subsystem, The cooling subsystem is A heating device configured to transfer heat to a process fluid, A thermal shielding device configured to remove heat from the process fluid, A sensor configured to detect variables of the cooling subsystem, The system comprises a processor circuit operably connected to the sensor and configured to provide the variable and a plurality of potential operating parameters on which the cooling subsystem can operate to a representative machine learning model of the cooling subsystem, in order to estimate at least one of the energy consumption, water consumption, and chemical consumption of the cooling subsystem for the potential operating parameters, The processor circuit is configured, at least in part, to determine the optimal operating parameters of the cooling subsystem to satisfy a target optimization index, based on at least one of the estimated energy consumption, water consumption, and chemical consumption for the potential operating parameters. The processor circuit is configured to allow the cooling subsystem to utilize the optimal operating parameters. The processor circuit is configured to provide the machine learning model with the plurality of potential operating parameters of the cooling subsystem in order to estimate chemical consumption based on the potential operating parameters. The system comprises a processor circuit configured, at least in part, to determine the optimal operating parameters of the cooling subsystem that satisfy the target optimization index, based on the consumption of the chemical substance.

2. A system equipped with a cooling subsystem, The cooling subsystem is A heating device configured to transfer heat to a process fluid, A thermal shielding device configured to remove heat from the process fluid, A sensor configured to detect variables of the cooling subsystem, The system comprises a processor circuit operably connected to the sensor and configured to provide the variable and a plurality of potential operating parameters on which the cooling subsystem can operate to a representative machine learning model of the cooling subsystem, in order to estimate at least one of the energy consumption, water consumption, and chemical consumption of the cooling subsystem for the potential operating parameters, The processor circuit is configured, at least in part, to determine the optimal operating parameters of the cooling subsystem to satisfy a target optimization index, based on at least one of the estimated energy consumption, water consumption, and chemical consumption for the potential operating parameters. The processor circuit is configured to allow the cooling subsystem to utilize the optimal operating parameters. The processor circuit is configured to identify a malfunction in the cooling subsystem. The system comprises a processor circuit configured, at least in part, to determine the plurality of potential operating parameters to be provided to the at least one representative machine learning model based on the identified defects.

3. A system equipped with a cooling subsystem, The cooling subsystem is A heating device configured to transfer heat to a process fluid, A thermal shielding device configured to remove heat from the process fluid, A sensor configured to detect variables of the cooling subsystem, The system comprises a processor circuit operably connected to the sensor and configured to provide the variable and a plurality of potential operating parameters on which the cooling subsystem can operate to a representative machine learning model of the cooling subsystem, in order to estimate at least one of the energy consumption, water consumption, and chemical consumption of the cooling subsystem for the potential operating parameters, The processor circuit is configured, at least in part, to determine the optimal operating parameters of the cooling subsystem to satisfy a target optimization index, based on at least one of the estimated energy consumption, water consumption, and chemical consumption for the potential operating parameters. The processor circuit is configured to allow the cooling subsystem to utilize the optimal operating parameters. The thermal shielding device includes a plurality of cooling towers, A system in which at least one of the cooling towers is capable of operating in a dry mode, and at least one of the cooling towers is capable of operating in a wet mode or an adiabatic mode.

4. A system equipped with a cooling subsystem, The cooling subsystem is A heating device configured to transfer heat to a process fluid, A thermal shielding device configured to remove heat from the process fluid, A sensor configured to detect variables of the cooling subsystem, The system comprises a processor circuit operably connected to the sensor and configured to provide the variable and a plurality of potential operating parameters on which the cooling subsystem can operate to a representative machine learning model of the cooling subsystem, in order to estimate at least one of the energy consumption, water consumption, and chemical consumption of the cooling subsystem for the potential operating parameters, The processor circuit is configured, at least in part, to determine the optimal operating parameters of the cooling subsystem to satisfy a target optimization index, based on at least one of the estimated energy consumption, water consumption, and chemical consumption for the potential operating parameters. The processor circuit is configured to allow the cooling subsystem to utilize the optimal operating parameters. The thermal shielding device includes a plurality of cooling towers, The system includes parameters indicating whether the cooling towers operate in a series or parallel configuration, as the aforementioned optimal operating parameters.

5. A system equipped with a cooling subsystem, The cooling subsystem is A heating device configured to transfer heat to a process fluid, A thermal shielding device configured to remove heat from the process fluid, A sensor configured to detect variables of the cooling subsystem, The system comprises a processor circuit operably connected to the sensor and configured to provide the variable and a plurality of potential operating parameters on which the cooling subsystem can operate to a representative machine learning model of the cooling subsystem, in order to estimate at least one of the energy consumption, water consumption, and chemical consumption of the cooling subsystem for the potential operating parameters, The processor circuit is configured, at least in part, to determine the optimal operating parameters of the cooling subsystem to satisfy a target optimization index, based on at least one of the estimated energy consumption, water consumption, and chemical consumption for the potential operating parameters. The processor circuit is configured to allow the cooling subsystem to utilize the optimal operating parameters. The thermal insulation device includes a thermal energy storage system. The system wherein the aforementioned optimal operating parameters include parameters indicating whether the thermal energy storage system is storing or releasing thermal energy.

6. A thermal shielding device for a cooling system, A cooling tower equipped with an evaporative heat exchanger capable of operating to cool process fluids, A sensor configured to detect variables of the cooling tower, The system comprises a controller operably connected to the cooling tower and the sensor, configured to implement optimal operating parameters for the cooling tower that satisfy target optimization indicators, The optimal operating parameters are determined, at least in part, by providing the variables and the potential operating parameters to a representative machine learning model of the cooling tower in order to estimate the power consumption and water consumption for a plurality of potential operating parameters on which the cooling tower can operate. The evaporative heat exchanger is a thermal shutoff device comprising an indirect heat exchanger configured to receive the process fluid and an evaporative liquid distribution system configured to distribute the evaporative liquid on the indirect heat exchanger.

7. The sensor variables of the cooling tower include variables of the process fluid leaving the cooling tower. The thermal shutoff device according to claim 6, wherein the cooling tower is operated such that the variable of the process fluid detaching from the cooling tower corresponds to the optimal variable of the process fluid, based on the optimal operating parameters.

8. The thermal insulation device according to claim 6, wherein the target optimization indicator is to minimize energy consumption, to minimize water consumption, or to minimize cost.

9. The thermal shutoff device according to claim 6, wherein the optimal operating parameter includes at least one of the operating mode of the cooling tower, the temperature of the process fluid leaving the cooling tower, the pressure of the process fluid leaving the cooling tower, and the flow rate of the process fluid.

10. The controller is configured to provide the variables of the cooling tower and the plurality of potential operating parameters to the machine learning model in order to estimate energy consumption and water consumption. The thermal insulator according to claim 6, wherein the controller is configured, at least in part, to determine the optimal operating parameters of the cooling system that satisfy the target optimization index based on the estimated energy consumption and water consumption.

11. The thermal shielding device according to claim 10, wherein the controller is configured to perform a first target optimization index and a different second target optimization index in response to a predetermined event.

12. The controller is configured to estimate future operating conditions. The controller is configured to provide the machine learning model with a plurality of future potential operating parameters of the cooling tower, based on the future operating conditions on which the cooling tower can operate, in order to estimate future energy consumption and future water consumption. The controller is configured to determine the optimal operating parameters of the cooling tower that satisfy the target optimization index, based on at least one of the following: The thermal insulation device according to claim 10, wherein the following are energy consumption, water consumption, future energy consumption, and future water consumption.

13. The plurality of potential operating parameters include a first plurality of potential operating parameters and a second plurality of potential operating parameters. The machine learning model includes a first machine learning model for estimating the energy consumption of the cooling tower and a second machine learning model for estimating water consumption. The thermal shutoff device according to claim 10, wherein the controller is configured to provide a first plurality of potential operating parameters to a first machine learning model in order to estimate the energy consumption of the cooling tower, and to provide a second plurality of potential operating parameters to a second machine learning model in order to estimate the water consumption of the cooling tower.

14. The thermal shutoff device according to claim 6, wherein the plurality of potential operating parameters provided to the machine learning model include a range of operating parameters within the operating limits of the cooling tower.

15. The thermal shutoff device according to claim 6, wherein the plurality of potential operating parameters include the fan speed and water consumption variables of the cooling tower.

16. The sensor includes a plurality of sensors configured to detect variables including the following: The following are at least one of the following: the water consumption variable of the cooling tower, the energy consumption variable of the cooling tower, and the temperature variable and pressure variable of the cooling tower's detachment process fluid. The thermal insulator according to claim 6, wherein the controller is configured to provide the variables of the cooling tower and the plurality of potential operating parameters to the machine learning model in order to estimate energy consumption and water consumption.

17. A thermal shielding device for a cooling system, A cooling tower equipped with an evaporative heat exchanger capable of operating to cool process fluids, A sensor configured to detect variables of the cooling tower, The system comprises a controller operably connected to the cooling tower and the sensor, configured to implement optimal operating parameters for the cooling tower that satisfy target optimization indicators, The optimal operating parameters are determined, at least in part, by providing the variables and the potential operating parameters to a representative machine learning model of the cooling tower in order to estimate the power consumption and water consumption for a plurality of potential operating parameters on which the cooling tower can operate. The controller is configured to identify malfunctions within the cooling tower. A thermal shutoff device in which the plurality of potential operating parameters of the cooling tower are determined, at least in part, based on the identified malfunctions.

18. The thermal shutoff device according to claim 6, wherein the optimal operating parameters are determined at least in part by providing the variables of the cooling tower and the plurality of potential operating parameters to a representative machine learning model of the cooling tower in order to estimate power consumption, water consumption and chemical consumption for the potential operating parameters.

19. A method for operating a cooling system, In a processor associated with a cooling system, The system receives the variables of the cooling system detected by the sensors of the cooling system, To estimate at least one of the energy consumption, water consumption, and chemical consumption of the cooling system for a plurality of potential operating parameters on which the cooling system can operate, the variables and the potential operating parameters are provided to a representative machine learning model of the cooling system. At least in part, the optimal operating parameters of the cooling system that satisfy the target optimization index are determined based on at least one of the energy consumption, water consumption, and chemical consumption for the estimated potential operating parameters, This includes enabling the use of the optimal operating parameters by the cooling system, Providing the plurality of potential operating parameters of the cooling system to the machine learning model includes providing the plurality of potential operating parameters to the machine learning model in order to estimate chemical consumption based on the potential operating parameters. A method for determining the optimal operating parameters that satisfy the target optimization indicators, comprising determining the optimal operating parameters based at least in part on the estimated chemical consumption.

20. A method for operating a cooling system, In a processor associated with a cooling system, The system receives the variables of the cooling system detected by the sensors of the cooling system, To estimate at least one of the energy consumption, water consumption, and chemical consumption of the cooling system for a plurality of potential operating parameters on which the cooling system can operate, the variables and the potential operating parameters are provided to a representative machine learning model of the cooling system. At least in part, the optimal operating parameters of the cooling system that satisfy the target optimization index are determined based on at least one of the energy consumption, water consumption, and chemical consumption for the estimated potential operating parameters, This includes enabling the use of the optimal operating parameters by the cooling system, A method in which the plurality of potential operating parameters of the cooling system are determined, at least in part, based on a malfunction detected within the cooling system.

Images