Air conditioner operation state identification data processing method

Abstract

The invention provides an air conditioner operation state recognition data processing method which is applied to the technical field of smart home. The method comprises the steps that edge equipment sends pre-acquired multi-dimensional feature data of the air conditioner to cloud equipment, the cloud equipment carries out multi-modal fusion processing on the multi-dimensional feature data to obtain multi-modal fusion feature data, and the multi-modal fusion feature data is processed based on a pre-trained state recognition model to obtain a state recognition result of the air conditioner; and the target operation state of the air conditioner is obtained, and then the operation parameters of the air conditioner are dynamically adjusted based on the target operation state and user preference information obtained in advance. By means of the method, the accuracy and efficiency of air conditioner operation state recognition are improved, energy consumption management is optimized, the adaptability and reliability of the system are enhanced, and meanwhile more personalized and intelligent user experience is provided.

Description

Technical Field

[0001] The present application relates to the field of smart home technology, and in particular to a method for processing data for identifying the operating status of an air conditioner. Background Art

[0002] As people's living standards improve, they have higher and higher requirements for the comfort of their living environment. Air conditioning is a device that regulates the ambient temperature and achieves the cooling effect of indoor air by recycling refrigerants. The refrigerant undergoes a cycle of compression, condensation, expansion, evaporation, etc. in the system, thereby absorbing and releasing heat to achieve the purpose of regulating the air temperature. With the advancement of technology, modern air conditioners not only have refrigeration functions, but also integrate multiple functions such as dehumidification, air purification, and heating. Through the intelligent control system, users can freely adjust the temperature and humidity of the indoor environment according to their needs.

[0003] In the prior art, the operating status of air conditioning systems is mainly identified by simple sensor data, such as temperature and humidity, which are analyzed by traditional algorithms to determine the switch and operating mode of the air conditioner.

[0004] However, the air conditioner adjusts the air conditioner status and indoor ambient temperature through the above method, the data is relatively single, lacks the ability to adapt to complex scenes, cannot achieve personalized settings, and is inconvenient to maintain. Summary of the invention

[0005] The present application provides a method for processing air conditioner operation status identification data, which is used to solve the problems of single processing data, inaccurate status identification and poor flexibility.

[0006] In a first aspect, the present application provides a method for processing air conditioner operation status identification data, which is applied to an intelligent management system, wherein the intelligent management system includes an edge device for data processing, a cloud device, a terminal device, and a server for model training, wherein the edge device, the cloud device, and the server are communicatively connected, and the cloud device and the terminal device are communicatively connected; the method includes:

[0007] The edge device sends the pre-acquired multi-dimensional feature data of the air conditioner to the cloud device, wherein the multi-dimensional feature data includes feature data of at least two dimensions;

[0008] The cloud device performs multimodal fusion processing on the multi-dimensional feature data to obtain multimodal fusion feature data;

[0009] The cloud device processes the multimodal fusion feature data based on a pre-trained state recognition model to obtain a target operating state of the air conditioner.

[0010] The cloud device dynamically adjusts the operating parameters of the air conditioner based on the target operating state and pre-acquired user preference information.

[0011] In combination with the first aspect, in some embodiments, the method further includes:

[0012] The cloud device performs fault prediction on the multimodal fusion feature data based on a pre-trained fault prediction model to obtain fault information;

[0013] Accordingly, the cloud device dynamically adjusts the operating parameters of the air conditioner based on the target operating state and the pre-acquired user preference information, including:

[0014] The cloud device dynamically adjusts the operating parameters based on the target operating state, the fault information and pre-acquired user preference information.

[0015] In combination with the first aspect, in some embodiments, the cloud device dynamically adjusts the operating parameters based on the target operating state, the fault information, and the pre-acquired user preference information, including:

[0016] The cloud device inputs the target operating state, the fault information and the user preference information into a preset parameter adjustment model to generate target parameters;

[0017] The cloud device dynamically adjusts the operating parameters based on the target parameters.

[0018] In combination with the first aspect, in some embodiments, the method further includes:

[0019] The server constructs an initial state recognition model;

[0020] The server performs iterative forward propagation on the initial state recognition model based on the pre-acquired first training data to calculate a first target loss function value;

[0021] The server updates the parameters of the initial state recognition model based on the first objective loss function value until a preset number of iterations is reached to obtain the state recognition model.

[0022] In combination with the first aspect, in some embodiments, the method further includes:

[0023] The server constructs an initial fault prediction model;

[0024] The server performs iterative forward propagation on the initial fault prediction model based on the pre-acquired second training data to calculate a second objective loss function value;

[0025] The server updates the parameters of the initial fault prediction model based on the second objective loss function value until the second objective loss function converges to obtain the state recognition model.

[0026] In combination with the first aspect, in some embodiments, the method further includes:

[0027] The server constructs an initial parameter adjustment model;

[0028] The server randomly extracts experience data from a pre-acquired experience pool and inputs the data into the initial parameter adjustment model to calculate a reward value;

[0029] The server updates the parameters of the initial parameter adjustment model through back propagation, and continues to extract experience data from the experience pool to iteratively calculate the initial parameter adjustment model until the accumulated reward value reaches a preset threshold, thereby obtaining the parameter adjustment model.

[0030] In combination with the first aspect, in some embodiments, before the cloud device dynamically adjusts the operating parameters of the air conditioner based on the target operating state and the pre-acquired user preference information, the method further includes:

[0031] Based on the pre-acquired user behavior data and user attribute data, the user preference information is obtained through processing by a pre-set user preference recognition model.

[0032] In combination with the first aspect, in some embodiments, the cloud device performs multimodal fusion processing on the multi-dimensional feature data to obtain multimodal fusion feature data, including:

[0033] For the feature data of each dimension, the cloud device performs weight calculation on the feature data to obtain an initial weight of the feature data;

[0034] The cloud device dynamically adjusts the initial weight of the feature data based on the pre-acquired environmental parameters to obtain the target weight of the feature data;

[0035] Based on the target weight of the feature data of each dimension, the multi-dimensional feature data is weightedly fused to obtain the multimodal fused feature data.

[0036] In combination with the first aspect, in some embodiments, before the edge device sends the pre-acquired multi-dimensional feature data of the air conditioner to the cloud device, the method further includes:

[0037] The edge device acquires multi-dimensional parameter data;

[0038] The edge device preprocesses and extracts features from the multi-dimensional parameter data to obtain the multi-dimensional feature data.

[0039] In a second aspect, the present application provides an intelligent management system, including: an edge device for data processing, a cloud device, a terminal device, and a server for model training;

[0040] The edge device, the cloud device and the server are communicatively connected, and the cloud device and the terminal device are communicatively connected, and are used to execute the air conditioning operation status identification data processing method described in the first aspect.

[0041] In a third aspect, the present application provides a cloud device, comprising: a processor, and a memory communicatively connected to the processor;

[0042] The memory stores computer-executable instructions;

[0043] The processor executes the computer-executable instructions stored in the memory to implement any one of the first aspects.

[0044] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, they are used to implement the air conditioning operation status identification data processing method described in any of the aforementioned aspects.

[0045] In a fifth aspect, the present application provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the air conditioning operation status identification data processing method described in any of the aforementioned aspects.

[0046] The air conditioner operation status identification data processing method provided in the present application is that the edge device sends the pre-acquired multi-dimensional feature data of the air conditioner to the cloud device, and the cloud device performs multi-modal fusion processing on the multi-dimensional feature data to obtain multi-modal fusion feature data, and processes the multi-modal fusion feature data based on the pre-trained state recognition model to obtain the target operation status of the air conditioner, and then dynamically adjusts the operation parameters of the air conditioner based on the target operation status and the pre-acquired user preference information. The above method not only improves the accuracy and efficiency of air conditioner operation status identification, but also optimizes energy consumption management, enhances the adaptability and reliability of the system, and provides a more personalized and intelligent user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0048] Figure 1A flowchart of a first embodiment of the method for processing air conditioner operation status identification data provided in the embodiment of the present application;

[0049] Figure 2 A flowchart of a second embodiment of the method for processing air conditioner operation status identification data provided in the embodiment of the present application;

[0050] Figure 3 A flowchart of a third embodiment of the method for processing air conditioner operation status identification data provided in the embodiment of the present application;

[0051] Figure 4 A flowchart of a fourth embodiment of the method for processing air conditioner operation status identification data provided in the embodiment of the present application;

[0052] Figure 5 A flowchart of a fifth embodiment of the method for processing air conditioner operation status identification data provided in the embodiment of the present application;

[0053] Figure 6 A flowchart of a sixth embodiment of the method for processing air conditioner operation status identification data provided in the embodiment of the present application;

[0054] Figure 7 A flowchart of Embodiment 7 of the method for processing air conditioner operation status identification data provided in the embodiment of the present application;

[0055] Figure 8 A schematic diagram of the structure of the intelligent management system provided in the embodiment of the present application;

[0056] Fig. 9 A schematic diagram of the structure of the cloud device provided in an embodiment of the present application.

[0057] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0058] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0059] As people's living standards improve, their requirements for the comfort of their living environment are getting higher and higher. Air conditioning is a device that regulates the ambient temperature and achieves the cooling effect of indoor air by recycling refrigerants. The refrigerant undergoes a cycle of compression, condensation, expansion, evaporation, etc. in the system, thereby absorbing and releasing heat to achieve the purpose of regulating the air temperature. With the advancement of science and technology, modern air conditioners not only have refrigeration functions, but also integrate multiple functions such as dehumidification, air purification, and heating. Through intelligent control systems, users can freely adjust the temperature and humidity of the indoor environment according to their needs. In the prior art, the operating status identification of the air conditioning system mainly relies on simple sensor data, such as temperature and humidity. These data are analyzed by traditional algorithms to determine the switch and operating mode of the air conditioner. However, the air conditioner adjusts the air conditioning status and indoor ambient temperature through the above method, the data is relatively single, lacks the ability to adapt to complex scenes, cannot achieve personalized settings, and is inconvenient to maintain.

[0060] In response to the above problems, the present application provides an air-conditioning operating status identification data processing method, which realizes the comprehensive collection and fusion of multimodal data, provides more accurate air-conditioning status identification, significantly improves the energy efficiency and intelligence level of the air-conditioning system, improves the maintainability of the system and reduces the operation and maintenance costs. Specifically, the prior art mainly relies on simple sensors and control logic to determine the switch and operation mode of the air conditioner, with a low degree of intelligence, insufficient energy efficiency management, and poor user experience. In view of these problems, the inventor studied whether it is possible to accurately identify the operating status of the air conditioner based on multi-dimensional information, and perform fault detection, and dynamically adjust the air-conditioning operating parameters based on the accurately identified operating status, fault information and user preference information, so as to more accurately identify the air-conditioning status, improve the maintainability of the system and reduce the operation and maintenance costs. Based on this, the technical solution of the present application is proposed.

[0061] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0062] Figure 1 The flowchart of the first embodiment of the method for processing the air conditioner operation status identification data provided in the embodiment of the present application is as follows: Figure 1 As shown, the method is applied to an intelligent management system, which includes an edge device for data processing, a cloud device, a terminal device and a server for model training, wherein the edge device, the cloud device and the server are communicated and connected, and the cloud device and the terminal device are communicated and connected. The method specifically includes:

[0063] S101: The edge device sends the pre-acquired multi-dimensional feature data of the air conditioner to the cloud device.

[0064] In this step, in order to reduce the cloud computing load and communication delay, thereby improving data processing efficiency, the sensor data collected by the air conditioner can be processed by the edge device, and then the processed multi-dimensional feature data can be sent to the cloud device.

[0065] Optionally, edge computing nodes generally refer to computing and storage devices deployed at the edge of the network, close to the data source. Edge devices can be devices specifically set up to process data from air conditioning systems, or they can be more general devices with this function. It should be noted that edge computing nodes are located near the location where the data is generated, such as in the room or building where the air conditioner is located.

[0066] S102: The cloud device performs multimodal fusion processing on the multi-dimensional feature data to obtain multimodal fusion feature data.

[0067] In this step, after the cloud device receives the multi-dimensional feature data uploaded by the edge device, since the multi-dimensional feature data is data of different modalities, in order to improve the accuracy of subsequent state recognition and fault prediction, multi-modal fusion processing can be performed on the multi-dimensional feature data to obtain multi-modal fusion feature data.

[0068] Specifically, for the feature data of each dimension, the cloud device calculates the weight of the feature data to obtain the initial weight of the feature data, and then dynamically adjusts the initial weight of the feature data based on the pre-acquired environmental parameters to obtain the target weight of the feature data. Finally, based on the target weight of the feature data of each dimension, the multi-dimensional feature data is weightedly fused to obtain multimodal fused feature data.

[0069] S103: The cloud device processes the multimodal fusion feature data based on a pre-trained state recognition model to obtain a target operating state of the air conditioner.

[0070] In this step, after the modal fusion feature data is obtained, the target operating state of the air conditioner is accurately identified based on the modal fusion feature data.

[0071] Specifically, the multimodal fusion feature data is input into the state recognition model to identify the target operating state.

[0072] S104: The cloud device dynamically adjusts the operating parameters of the air conditioner based on the target operating state and pre-acquired user preference information.

[0073] In this step, after the cloud device accurately identifies the target operating state of the air conditioner, in order to improve the intelligence of the air conditioner and enhance the user experience, the operating parameters are adjusted based on the target operating state and pre-acquired user preference information.

[0074] Specifically, the target operating state and user preference information are input into the parameter adjustment model for processing to generate target parameters, and then the cloud device adjusts the operating parameters according to the target parameters.

[0075] Optionally, fault information may also be predicted based on multimodal fusion feature data, thereby inputting the target operating state, fault information, and user preference information into a parameter adjustment model for processing to obtain target parameters.

[0076] In a possible implementation, before step S104, it is also necessary to obtain user preference information based on pre-acquired user behavior data and user attribute data through a pre-set user preference recognition model.

[0077] In order to meet the personalized settings of users and improve user satisfaction and experience intelligently according to different users, the cloud device obtains the user behavior data and user attribute data of the user through the terminal device.

[0078] Among them, user behavior data includes the user's usage history data of the air conditioner, for example, temperature setting history: the user's past record of setting the air conditioner temperature at different time points and situations. Usage time: the daily or weekly usage time of the air conditioner, and the distribution of usage time (such as whether it is mainly used during the day or at night). Operation mode selection: whether the user prefers to use automatic mode, cooling mode, heating mode or dehumidification mode, etc. Wind speed and wind direction preference: the user's preference settings for air conditioner wind speed and wind direction. Energy-saving habits: whether the user consciously adjusts the air conditioner settings to save energy, such as setting a higher indoor temperature to reduce energy consumption. Interactive behavior: the way and frequency of user interaction with the air conditioning system (such as through a mobile phone APP, remote control or voice control).

[0079] User attribute data includes basic information about the user, such as age: people of different ages have different comfort requirements for environmental conditions such as temperature and humidity. Gender: Gender affects an individual's perception and preference for temperature. Living environment: The climate characteristics of the residence, room size, orientation, number of windows, etc., will also affect the user's demand for air conditioning. Weight and health status: People with heavier weight or certain health conditions (such as cardiovascular disease, respiratory disease) may be more sensitive to temperature. Lifestyle habits: such as whether you exercise regularly, the nature of your work (whether you need to sit or stand for a long time), etc., all of which will affect the user's preference for ambient temperature.

[0080] After obtaining user behavior data and user attribute data, it is necessary to pre-process the above data, such as data cleaning, data standardization and feature extraction, to obtain processed data. Then the processed data is input into the user preference recognition model to generate user preference information.

[0081] The user preference recognition model can be obtained by training an initial model based on user behavior data and user attribute data of at least one user acquired in advance and preference information of each user, wherein the initial model can be a decision tree, random forest, gradient boosting machine, neural network, etc. The obtained user preference information may include information such as temperature preference range, wind speed preference, mode preference, etc. Other types of preference information may also be included, which is not specifically limited in this application.

[0082] In another possible implementation, the method further includes: the edge device acquires multi-dimensional parameter data.

[0083] The multiple sensors configured in the air conditioner can collect multi-dimensional parameter data when the air conditioner is running, such as temperature, humidity, air flow, sound, vibration, current, voltage and other sensor parameters. The air conditioner control system uploads the collected multi-dimensional parameter data to the edge device in real time for processing.

[0084] Optionally, the locations where the sensors corresponding to the multi-dimensional parameter data collected above can be installed include: Temperature and humidity sensors: installed at the air inlet and outlet of the air-conditioning equipment to accurately monitor the temperature and humidity of the air. Flow sensor: installed at the air inlet and outlet of the air-conditioning fan to detect the air flow. Position sensor: installed above both ends of the drive shaft of the air-conditioning fan impeller to detect the vibration of the drive shaft. Temperature sensor: installed at the top of the air outlet of the air-conditioning fan to detect the fin temperature to prevent icing.

[0085] The edge device preprocesses and extracts features of the multi-dimensional parameter data to obtain multi-dimensional feature data.

[0086] The edge device preprocesses and extracts features from the received multi-dimensional parameter data.

[0087] Exemplarily, preprocessing may include: Removing duplicate data: ensuring that there are no duplicate records in the data set. Filling missing values: filling missing data using the mean, median, mode, or interpolation method. Correcting errors: correcting data entry errors, such as typos, incorrect unit conversions, etc. Normalization and standardization: scaling the data to a standard range (such as between 0 and 1), or converting the data to a distribution with a mean of 0 and a standard deviation of 1 for subsequent analysis.

[0088] Feature extraction can use deep learning models (such as convolutional neural networks (CNN) or recurrent neural networks (RNN)) to automatically extract key features from multi-dimensional parameter data to obtain multi-dimensional feature data.

[0089] In the air conditioner operation status identification data processing method provided in this embodiment, the edge device sends the pre-acquired multi-dimensional feature data of the air conditioner to the cloud device, and the cloud device performs multi-modal fusion processing on the multi-dimensional feature data to obtain multi-modal fusion feature data, and processes the multi-modal fusion feature data based on the pre-trained state recognition model to obtain the target operation status of the air conditioner, and then dynamically adjusts the operation parameters of the air conditioner based on the target operation status and the pre-acquired user preference information. The above method not only improves the accuracy and efficiency of air conditioner operation status identification, but also optimizes energy consumption management, enhances the adaptability and reliability of the system, and provides a more personalized and intelligent user experience.

[0090] Figure 2 The flowchart of the second embodiment of the method for processing the air conditioner operation status identification data provided in the embodiment of the present application is as follows: Figure 2 As shown, based on the above embodiment, the method further includes:

[0091] S201: The edge device sends the pre-acquired multi-dimensional feature data of the air conditioner to the cloud device.

[0092] S202: The cloud device performs multimodal fusion processing on the multi-dimensional feature data to obtain multimodal fusion feature data.

[0093] The specific implementation method of this step is the same as the specific implementation method in the above embodiments, and will not be repeated here.

[0094] S203: The cloud device performs fault prediction on the multimodal fusion feature data based on the pre-trained fault prediction model to obtain fault information.

[0095] S204: The cloud device generates a fault warning based on the fault information and sends it to the terminal device.

[0096] The cloud device can also input the multimodal fusion feature data into the fault prediction model to predict fault information, thereby ensuring the normal operation of the air conditioner and improving the user experience. When the model predicts a possible serious fault, it can send an early warning notification to the user or maintenance personnel so that timely measures can be taken for prevention or maintenance.

[0097] Exemplarily, the cloud device sends the fault information to the terminal device, and the user can view the details of the fault information through the application of the terminal device.

[0098] Optionally, the predicted fault information and the fault prediction sent to the terminal device may include:

[0099] Compressor failure prediction: Predicts that the compressor may fail due to long-term high-load operation, abnormal current or excessive temperature. For example, if the compressor is at risk of overheating, it is recommended to reduce the cooling power or perform maintenance inspections.

[0100] Refrigerant leakage prediction: By analyzing the changes in refrigerant pressure, flow rate, temperature and other parameters, predict whether there is a leakage. For example, if an abnormal drop in refrigerant pressure is detected, there may be a leakage. Please check and refill the refrigerant as soon as possible.

[0101] Fan failure prediction: Predicts that the fan may drop in speed or stop working due to motor failure, bearing wear or blockage, etc. For example, if the fan speed is abnormal, there may be a risk of failure. It is recommended to check the fan motor and blades.

[0102] Circuit fault prediction: By analyzing the voltage, current, resistance and other parameters in the circuit, predict whether there are problems such as short circuit, open circuit or component aging. For example, if the current in a certain area of ​​the circuit board is abnormal, there may be a short circuit risk, please check the circuit.

[0103] Sensor failure prediction: Predict that the sensor may cause inaccurate measurement data due to aging, damage, or contamination. For example, if the temperature sensor reading fluctuates abnormally, there may be a fault, and it is recommended to calibrate or replace the sensor.

[0104] Filter clogging prediction: By analyzing parameters such as air flow and pressure difference, it is predicted whether the filter will be clogged due to long-term use. For example, if the air flow rate drops, the filter may be clogged and it is recommended to clean or replace the filter.

[0105] Condenser frost prediction: Predicts that frost may form on the condenser surface due to high humidity, low temperature or excessive cooling capacity. For example, if there is a tendency for frost to form on the condenser surface, it is recommended to adjust the cooling capacity or increase the defrost cycle.

[0106] Since fault information may not affect the operation of the air conditioner, when sending a fault warning to the user, it can be determined whether to prompt the user based on the severity of the fault. The user can view all records of fault information through the terminal device. Fault information that does not affect the operation of the air conditioner is only sent to the terminal device without prompting the user. For serious fault information, while sending it to the terminal device, the user can also be prompted by voice or pop-up screen.

[0107] Optionally, based on the above steps, step S104 specifically includes: the cloud device dynamically adjusts the operating parameters based on the target operating state, fault information and pre-acquired user preference information.

[0108] In the air conditioner operation status identification data processing method provided in this embodiment, the edge device sends the pre-acquired multi-dimensional feature data of the air conditioner to the cloud device, and the cloud device performs multi-modal fusion processing on the multi-dimensional feature data to obtain multi-modal fusion feature data, and based on the pre-trained fault prediction model, performs fault prediction on the multi-modal fusion feature data to obtain fault information, and then generates a fault warning based on the fault information and sends it to the terminal device. The above method improves the accuracy of fault prediction, strengthens fault diagnosis and maintenance, and also improves data processing efficiency and enhances user experience.

[0109] Figure 3 The flowchart of the third embodiment of the method for processing the air conditioner operation status identification data provided in the embodiment of the present application is as follows: Figure 3 As shown, the method specifically includes:

[0110] S301: The edge device sends the pre-acquired multi-dimensional feature data of the air conditioner to the cloud device.

[0111] S302: The cloud device performs multimodal fusion processing on the multi-dimensional feature data to obtain multimodal fusion feature data.

[0112] S303: The cloud device processes the multimodal fusion feature data based on a pre-trained state recognition model to obtain a target operating state of the air conditioner.

[0113] The specific implementation method of this step is the same as the specific implementation method in the above embodiments, and will not be repeated here.

[0114] S304: The cloud device inputs the target operating status, fault information and user preference information into a pre-set parameter adjustment model to generate target parameters.

[0115] S305: The cloud device dynamically adjusts the operating parameters based on the target parameters.

[0116] After obtaining the target operating state of the air conditioner, in order to achieve intelligence and user personalization and thus improve user experience, the target operating state, fault information and user preference information can be analyzed based on the pre-configured parameter adjustment model to obtain the target parameters, and based on the target parameters, the operating parameters of the air conditioner can be dynamically adjusted.

[0117] Exemplarily, user preference information may be a target temperature (24°C) set by the user, energy-saving mode preference, and fault information may be sensor failure: the sensor is an important component in the air-conditioning system for sensing and measuring environmental parameters (such as temperature, humidity, pressure, etc.). When a sensor fails, such as inaccurate readings or complete failure, the air-conditioning system may not be able to correctly sense the indoor environmental state, resulting in poor cooling / heating effects or increased energy consumption. Compressor failure: The compressor is the core component of the air-conditioning system, responsible for the compression and circulation of the refrigerant. When the compressor fails, such as overheating, overload or mechanical failure, the cooling / heating capacity of the air-conditioning system will be greatly reduced or even completely lost. Refrigerant leakage: Refrigerant is a key substance for achieving cooling / heating in the air-conditioning system. When the refrigerant leaks, the cooling / heating effect of the air-conditioning system will gradually weaken until it fails completely. Electrical system failure: The electrical system is an important component in the air-conditioning system for controlling and driving the operation of various components. When the electrical system fails, such as a short circuit, open circuit or controller failure, the normal operation of the air-conditioning system will be seriously affected. Other mechanical failures: In addition to the above types of failures, the air conditioning system may also encounter other mechanical failures, such as fan failure, condenser blockage, evaporator frosting, etc. These failures will also affect the performance and efficiency of the air conditioning system.

[0118] In the air-conditioning operation status identification data processing method provided in this embodiment, the edge device sends the pre-acquired multi-dimensional feature data of the air conditioner to the cloud device, and the cloud device performs multi-modal fusion processing on the multi-dimensional feature data to obtain multi-modal fusion feature data, and inputs the target operation status, fault information and user preference information into the pre-set parameter adjustment model to generate the target parameters, and finally dynamically adjusts the operation parameters based on the target parameters. Through the above method, not only the real-time and high efficiency are guaranteed, but also accurate identification and prediction are achieved, flexibility and personalization are improved, and the operation efficiency and user experience of the air-conditioning system are significantly improved.

[0119] The target parameters obtained through analysis may include cooling power, wind speed, air supply direction, etc. After obtaining the target parameters, the cloud device can remotely control the air conditioner control system to adjust the parameters, or send the target parameters to the air conditioner control system. After receiving the parameters, the control system will immediately dynamically adjust the operating parameters of the air conditioner. For example, reduce the cooling power to a certain value, adjust the wind speed to a medium wind speed, and optimize the air supply direction to ensure that the indoor temperature drops evenly.

[0120] Figure 4 The flowchart of the fourth embodiment of the method for processing the air conditioner operation status identification data provided in the embodiment of the present application is as follows: Figure 4 As shown, based on the above embodiments, the method includes:

[0121] S401: The server constructs an initial state recognition model.

[0122] In order to accurately identify the operating status of the air conditioner, it is necessary to train an accurate status recognition model. The server pre-builds the initial status recognition model.

[0123] Exemplarily, the initial state recognition model may use a support vector machine, a random forest, a neural network or the like, or may be a deep learning model (such as a convolutional neural network, a recurrent neural network or a long short-term memory network).

[0124] S402: The server performs iterative forward propagation on the initial state recognition model based on the pre-acquired first training data, and calculates a first target loss function value.

[0125] S403: The server updates the parameters of the initial state recognition model based on the first objective loss function value until a preset number of iterations is reached to obtain a state recognition model.

[0126] After the initial state recognition model is constructed, the parameters of the initial state recognition model are initialized, and the initial state recognition model is iteratively trained using the first training data. During the training process, the first target loss function value is calculated in real time, and based on the first target loss function value, the parameters of the initial state recognition model are adjusted until a preset number of iterations is reached, and then the training is terminated.

[0127] Optionally, during the training process, regularization, dropout and other techniques may be introduced to improve the generalization ability of the model. The parameters for initializing the initial state recognition model may include: learning rate, number of iterations, batch size and the like.

[0128] First, the training data can be divided into training data and verification data. After the training is completed, the verification data can be input into the trained model for evaluation and optimization. The accuracy, recall rate, and F1 score can be used as evaluation indicators. The model parameters can be adjusted according to the evaluation results to obtain the state recognition model.

[0129] The first objective loss function can be expressed as:

[0130]

[0131] Among them, α, β, γ represent weight coefficients, N is the number of samples, C is the number of categories, κ ic is the true label (0 or 1) of whether the i-th sample belongs to the c-th class, is the probability that the model predicts that the i-th sample belongs to the c-th class. represents the maximum value of the predicted probability of the i-th sample, Q() represents the indicator function, when κ i+1,c =κic It takes the value 1 when , otherwise it takes the value 0, which is used to penalize the inconsistency of the prediction results at adjacent time points.

[0132] The above loss function not only measures the difference between the running state predicted by the model and the actual running state, but also ensures that the model prediction is time consistent, thereby improving the accuracy of the model prediction.

[0133] In the air conditioner operation state recognition data processing method provided in this embodiment, the server constructs an initial state recognition model, performs iterative forward propagation on the initial state recognition model based on the first training data acquired in advance, calculates the first target loss function value, and updates the parameters of the initial state recognition model based on the first target loss function value until a preset number of iterations is reached to obtain a state recognition model. The state recognition model is constructed by the above-mentioned iterative training method, which has the advantages of improving model accuracy, enhancing model robustness, improving training efficiency, flexibility and scalability.

[0134] Figure 5 The flowchart of the fifth embodiment of the method for processing the air conditioner operation status identification data provided in the embodiment of the present application is as follows: Figure 5 As shown, based on the above embodiments, the method includes:

[0135] S501: The server builds an initial fault prediction model.

[0136] S502: The server performs iterative forward propagation on the initial fault prediction model based on the pre-acquired second training data, and calculates a second target loss function value.

[0137] S503: The server updates the parameters of the initial fault prediction model based on the second objective loss function value until the second objective loss function converges to obtain the fault prediction model.

[0138] In order to accurately predict the fault information of the air conditioner, it is necessary to train an accurate fault prediction model. The server pre-builds an initial fault prediction model.

[0139] Exemplarily, the initial fault prediction model may use a support vector machine, a random forest, a neural network or the like, or may be a deep learning model (such as a convolutional neural network, a recurrent neural network or a long short-term memory network).

[0140] After the initial fault prediction model is constructed, the parameters of the initial fault prediction model are initialized, and the initial fault prediction model is iteratively trained using the second training data. During the training process, the second target loss function value is calculated in real time, and based on the second target loss function value, the parameters of the initial fault prediction model are adjusted until a preset number of iterations is reached, and then the training is terminated.

[0141] Optionally, during the training process, regularization, dropout and other techniques may be introduced to improve the generalization ability of the model. Parameters for initializing the initial fault prediction model may include: learning rate, number of iterations, batch size and the like.

[0142] The second training data can be divided into training data and verification data. After the training is completed, the verification data can be input into the trained model for evaluation and optimization. The accuracy, recall rate, and F1 score can be used as evaluation indicators. The model parameters can be adjusted according to the evaluation results to obtain a fault prediction model.

[0143] It should be noted that the second training data and the first training data may be the same or different, and this embodiment of the present application does not specifically limit this.

[0144] Before an air conditioner fails, there are usually some precursor signals or patterns. A loss term can be designed to encourage the model to capture these precursor signals and give early warnings when they occur:

[0145]

[0146] Where M is the total number of pre-fault time windows, is the prediction uncertainty of the model at the ith time step, is the prediction error rate of the model at the i-th time step.

[0147] The development of a fault is usually a gradually deteriorating process. A loss term can be designed to encourage the model to capture the trend of this fault evolution and reflect this trend in the prediction:

[0148]

[0149] Where N is the number of samples, T is the number of future time steps, and are the model predicted and actual fault severity at the t-th time step in the future, respectively.

[0150] In air conditioner failure prediction, failures may manifest in many different modes. A loss term can be designed to encourage the model to consider multiple possible failure modes when predicting and give the corresponding probability distribution:

[0151]

[0152] Among them, P ture (M|x i )||P pred (M|x i ) means that given input x iWhen , the actual and predicted failure mode distributions are obtained, and M represents the set of failure modes.

[0153] Then the second objective loss function is expressed as:

[0154] L2=νL precursor +ιL evolution +θL multimodal

[0155] Among them, v, ι, and θ represent weight coefficients.

[0156] In the air conditioner operation status identification data processing method provided in this embodiment, the server constructs an initial fault prediction model, performs iterative forward propagation on the initial fault prediction model based on the pre-acquired second training data, calculates the second target loss function value, and updates the parameters of the initial fault prediction model based on the second target loss function value until the second target loss function converges to obtain a state identification model. Through the above method, the accuracy of fault prediction is improved and the generalization ability of the model is enhanced.

[0157] Figure 6 The flowchart of the sixth embodiment of the method for processing the air conditioner operation status identification data provided in the embodiment of the present application is as follows: Figure 6 As shown, based on the above embodiments, the method includes:

[0158] S601: The server constructs an initial parameter adjustment model.

[0159] In order to accurately adjust the operating parameters of the air conditioner, it is necessary to train an accurate parameter adjustment model. The server pre-builds the initial parameter adjustment model.

[0160] S602: The server randomly extracts experience data from a pre-acquired experience pool and inputs the data into an initial parameter adjustment model to calculate a reward value.

[0161] S603: The server updates the parameters of the initial parameter adjustment model through back propagation, and continues to extract experience data from the experience pool to iteratively calculate the initial parameter adjustment model until the accumulated reward value reaches a preset threshold, thereby obtaining a parameter adjustment model.

[0162] First, you need to define a simulated or actual environment that can receive the agent's actions and return new states and rewards. In the air conditioning control task, the environment can be a simulated air conditioning system or an actual physical environment. Set the initial parameters of the agent, including the structure of the neural network, learning rate, discount factor, etc. Then design the reward function:

[0163] R=-αP current -β(|ΔT|+|Δspeed|+…)-|Tactual -T desired |+R preference

[0164] Among them, T actual Indicates the current indoor temperature, T desired Indicates the center value of the user's expected temperature range, P current represents the current operating power of the air conditioner, α is a positive coefficient used to adjust the weight of the energy-saving reward. ΔT represents the temperature change, Δspeed represents the wind speed change, and β is a positive coefficient. This reward item will penalize any large changes in operating parameters. preference Indicates the user's preference for reward.

[0165] Clarify the definitions of state space and action space. The state space should include all factors that affect the air conditioning operation effect, such as indoor environment status, air conditioning operation status, fault prediction information and user preferences. The action space includes all adjustable operation parameters, such as set temperature, wind speed, operation mode, etc.

[0166] During training, the model continuously interacts with the environment, selects actions based on the current state, and observes reward signals from the environment. Based on the reward signals, the model adjusts its parameters to optimize the control strategy.

[0167] The training process may require multiple iterations until the model converges or reaches a preset number of training rounds. During the training process, some techniques can be used to speed up the training process, such as batch processing, learning rate adjustment, etc.

[0168] Optionally, after the model is deployed in the actual environment, real-time operation data and user feedback can be collected to optimize the model online. Online optimization can make the model adapt to the changing environment and user needs.

[0169] In the data processing method for identifying the operation status of an air conditioner provided in this embodiment, the server constructs an initial parameter adjustment model, randomly extracts experience data from a pre-acquired experience pool and inputs it into the initial parameter adjustment model, calculates a reward value, updates the parameters of the initial parameter adjustment model through back propagation, and continues to extract experience data from the experience pool to iteratively calculate the initial parameter adjustment model until the accumulated reward value reaches a preset threshold, thereby obtaining a parameter adjustment model. This method improves model performance, flexibility and scalability, and intelligence by efficiently utilizing data, stabilizing the training process, and improving model performance.

[0170] Figure 7 The flowchart of the seventh embodiment of the method for processing the air conditioner operation status identification data provided in the embodiment of the present application is as follows: Figure 7 As shown, the method specifically:

[0171] S701: The edge device sends the pre-acquired multi-dimensional feature data of the air conditioner to the cloud device.

[0172] The specific implementation method of this step is the same as the specific implementation method in the above embodiments, and will not be repeated here.

[0173] S702: For the feature data of each dimension, the cloud device calculates the weight of the feature data to obtain the initial weight of the feature data.

[0174] In multi-dimensional feature data, different features have different importance for identifying the operating status of the air conditioner. For example, some data such as compressor current and condenser temperature directly reflect the key operating status of the air conditioner, while some features may only be indirectly related or have little impact. Through the attention mechanism, the system can automatically learn and identify the importance differences of multi-dimensional features and assign corresponding weights to them.

[0175] Specifically, for the feature data of each dimension, the cloud device first performs a linear transformation on the feature data to obtain three linear transformation matrices, where the three linear transformation matrices include a query matrix, a key matrix, and a value matrix. The dot product of the query matrix and the key matrix is ​​then calculated to obtain a score matrix. The score matrix is ​​normalized with attention weights to obtain the initial weights of the feature data.

[0176] Exemplarily, the above linear transformation can be calculated by the following formula:

[0177] Q=XW Q

[0178] K=XW K

[0179] V=XW V

[0180] Among them, Q represents the query matrix, K represents the key matrix, V represents the value matrix, and X represents the feature data.

[0181] The formula for calculating the dot product is expressed as:

[0182] scores=QK T

[0183] Optionally, since the result of the dot product may be very large, dot product scaling can also be performed, which can be achieved through the following formula:

[0184]

[0185] Among them, d k Represents dimension.

[0186] Normalizing the score matrix for attention weights can be achieved using the following formula:

[0187]

[0188] Wherein, θ represents a nonlinear function, such as sigmoid or tanh, which can be determined according to actual conditions and is not specifically limited in the present embodiment. i represents feature data, Γ represents the target range, such as [0, 1], which can also be set according to actual conditions, n represents the total number of data in the feature data, and μ is a small positive number used to avoid division by zero and maintain numerical stability.

[0189] S703: The cloud device dynamically adjusts the initial weight of the feature data based on the pre-acquired environmental parameters to obtain the target weight of the feature data.

[0190] In this step, since the use of air conditioning is highly correlated with the external environment, the weights in the above steps can also be dynamically adjusted based on environmental parameters.

[0191] Exemplary, a mechanism for automatically adjusting feature weights based on monitoring data, for example, when the outdoor temperature exceeds a preset threshold, the weight of the outdoor temperature feature in the fusion process is increased.

[0192] According to historical data and actual application scenarios, the warning threshold of outdoor temperature is set. For example, when the outdoor temperature reaches above 35℃, it is considered necessary to increase the weight of outdoor temperature features in the fusion process.

[0193] When the outdoor temperature exceeds the preset threshold, the weight adjustment rule is triggered. The rule can be set to increase the weight of the outdoor temperature feature by a certain proportion, and the weights of other features may need to be adjusted accordingly to keep the total weight at 1.

[0194] Optionally, machine learning or deep learning models can be used to predict the importance of different features in the current environment. For example, a lightweight neural network can be trained to predict feature weights based on current and historical environment data. Based on the predicted weights, the proportion of feature fusion can be adjusted in real time.

[0195] For example, when the fluctuation range of a certain environmental parameter is large, it indicates that the parameter has a greater impact on the operating status of the air conditioner, so the weight of the feature data related to the parameter can be increased. The rate of change of the environmental parameter can also be used as a basis for adjusting the weight. If a certain parameter changes significantly in a short period of time, it means that the impact of the parameter on the operating status of the air conditioner is more urgent, and the weight of the corresponding feature data needs to be increased for a quick response. By analyzing historical data, you can also understand the correlation between different feature data and recognition results. For feature data that is highly correlated with the recognition results, its weight can be increased to improve the accuracy of recognition.

[0196] S704: The cloud device performs weighted fusion on the multi-dimensional feature data based on the target weight of the feature data of each dimension to obtain multi-modal fusion feature data.

[0197] In this step, after obtaining the target weight of the feature data of each dimension, multimodal fusion feature data can be obtained through weighted fusion.

[0198] In the air conditioner operation status identification data processing method provided in this embodiment, the edge device sends the pre-acquired multi-dimensional feature data of the air conditioner to the cloud device. For the feature data of each dimension, the cloud device performs weight calculation on the feature data to obtain the initial weight of the feature data. The cloud device performs weighted fusion on the multi-dimensional feature data based on the target weight of the feature data of each dimension to obtain multimodal fusion feature data. Through the above method, the recognition accuracy and efficiency of the air conditioner operation status are improved, the energy consumption management is optimized, the adaptability and reliability of the system are enhanced, and a more personalized and intelligent user experience is provided.

[0199] Figure 8 A schematic diagram of the structure of the intelligent management system provided in the embodiment of the present application is shown in FIG. Figure 8 As shown, the intelligent management system 800 includes: an edge device 801 for data processing, a cloud device 802, a terminal device 803 and a server 804 for model training, the edge device 801, the cloud device 802 and the server 804 are communicatively connected, and the cloud device 802 and the terminal device 803 are communicatively connected, and are used to execute the air conditioning operation status identification data processing method in any of the aforementioned embodiments.

[0200] Optional,

[0201] Fig. 9 A schematic diagram of the structure of the cloud device provided in the embodiment of the present application, such as Fig. 9 As shown, the cloud device 900 includes: a processor 902, a memory 901 and a communication interface 903 that are communicatively connected to the processor 902;

[0202] The memory 901 stores computer executable instructions;

[0203] The processor 902 executes the computer-executable instructions stored in the memory 901 to implement the air-conditioning operation status identification data processing method in any method embodiment.

[0204] An embodiment of the present application also provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to execute the air conditioning operation status identification data processing method provided by the various embodiments described above.

[0205] The computer-readable storage medium mentioned above can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory, electrically erasable programmable read-only memory, erasable programmable read-only memory, programmable read-only memory, read-only memory, magnetic memory, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium that can be accessed by a general or special-purpose computer.

[0206] Optionally, a readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (ASIC). Of course, the processor and the readable storage medium can also exist in the device as discrete components.

[0207] An embodiment of the present application also provides a computer program product, which includes a computer program. The computer program is stored in a computer-readable storage medium. At least one processor can read the computer program from the computer-readable storage medium. When at least one processor executes the computer program, the technical solution provided by any of the above method embodiments can be implemented.

[0208] In the present application, "at least one" means one or more, and "plurality" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship; in the formula, the character " / " indicates that the previous and next associated objects are in a "division" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, c can be single or multiple.

[0209] It is to be understood that the various numerical numbers involved in the embodiments of the present application are only for the convenience of description and are not intended to limit the scope of the embodiments of the present application. In the embodiments of the present application, the size of the sequence number of each process does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0210] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present application. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0211] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A method for processing air conditioner operation status identification data, characterized in that: Applied to an intelligent management system, the intelligent management system includes an edge device for data processing, a cloud device, a terminal device and a server for model training, the edge device, the cloud device and the server are communicatively connected, and the cloud device and the terminal device are communicatively connected; the method includes: The edge device sends the pre-acquired multi-dimensional feature data of the air conditioner to the cloud device, wherein the multi-dimensional feature data includes feature data of at least two dimensions; The cloud device performs multimodal fusion processing on the multi-dimensional feature data to obtain multimodal fusion feature data; The cloud device processes the multimodal fusion feature data based on a pre-trained state recognition model to obtain a target operating state of the air conditioner; The cloud device dynamically adjusts the operating parameters of the air conditioner based on the target operating state and pre-acquired user preference information.

2. The method according to claim 1, characterized in that The method further comprises: The cloud device performs fault prediction on the multimodal fusion feature data based on a pre-trained fault prediction model to obtain fault information; Accordingly, the cloud device dynamically adjusts the operating parameters of the air conditioner based on the target operating state and the pre-acquired user preference information, including: The cloud device dynamically adjusts the operating parameters based on the target operating state, the fault information and pre-acquired user preference information.

3. The method according to claim 2, characterized in that The cloud device dynamically adjusts the operating parameters based on the target operating state, the fault information, and the pre-acquired user preference information, including: The cloud device inputs the target operating state, the fault information and the user preference information into a preset parameter adjustment model to generate target parameters; The cloud device dynamically adjusts the operating parameters based on the target parameters.

4. The method according to claim 1, characterized in that: The method further comprises: The server constructs an initial state recognition model; The server performs iterative forward propagation on the initial state recognition model based on the pre-acquired first training data to calculate a first target loss function value; The server updates the parameters of the initial state recognition model based on the first objective loss function value until a preset number of iterations is reached to obtain the state recognition model.

5. The method according to claim 2, characterized in that: The method further comprises: The server constructs an initial fault prediction model; The server performs iterative forward propagation on the initial fault prediction model based on the pre-acquired second training data to calculate a second objective loss function value; The server updates the parameters of the initial fault prediction model based on the second objective loss function value until the second objective loss function converges to obtain the state recognition model.

6. The method according to claim 3, characterized in that The method further comprises: The server constructs an initial parameter adjustment model; The server randomly extracts experience data from a pre-acquired experience pool and inputs the data into the initial parameter adjustment model to calculate a reward value; The server updates the parameters of the initial parameter adjustment model through back propagation, and continues to extract experience data from the experience pool to iteratively calculate the initial parameter adjustment model until the accumulated reward value reaches a preset threshold, thereby obtaining the parameter adjustment model.

7. The method according to claim 1 or 2, characterized in that: Before the cloud device dynamically adjusts the operating parameters of the air conditioner based on the target operating state and the pre-acquired user preference information, the method further includes: The user preference information is obtained based on the pre-acquired user behavior data and user attribute data through processing by a pre-set user preference recognition model, wherein the user preference recognition model is trained based on the pre-acquired user data set.

8. The method according to any one of claims 1 to 6, characterized in that: The cloud device performs multimodal fusion processing on the multi-dimensional feature data to obtain multimodal fusion feature data, including: For the feature data of each dimension, the cloud device performs weight calculation on the feature data to obtain an initial weight of the feature data; The cloud device dynamically adjusts the initial weight of the feature data based on the pre-acquired environmental parameters to obtain the target weight of the feature data; The cloud device performs weighted fusion on the multi-dimensional feature data based on the target weight of the feature data of each dimension to obtain the multimodal fusion feature data.

9. The method according to any one of claims 1 to 6, characterized in that: Before the edge device sends the pre-acquired multi-dimensional feature data of the air conditioner to the cloud device, the method further includes: The edge device acquires multi-dimensional parameter data; The edge device preprocesses and extracts features from the multi-dimensional parameter data to obtain the multi-dimensional feature data.

10. An intelligent management system, characterized in that: include: Edge devices, cloud devices, terminal devices for data processing, and servers for model training; The edge device, the cloud device and the server are communicatively connected, and the cloud device and the terminal device are communicatively connected, and are used to execute the air conditioning operation status identification data processing method described in any one of claims 1 to 9.

Images