Method and device for controlling air conditioner and air conditioner

[0028] Example 1

[0029] According to an embodiment of the present application, an embodiment of a method for controlling an air conditioner is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and, although The logical sequence is shown in the flowchart, but in some cases, the steps shown or described may be performed in a different order than here.

[0030] figure 1 Is the method of controlling the air conditioner according to the embodiment of the application, such as figure 1 As shown, the method includes the following steps:

[0031] Step S102: Read the pre-stored control model from the built-in chip of the air conditioner, where the control model is a neural network model, which is obtained through machine learning training using multiple sets of data. Each set of data in the multiple sets of data includes: The simulation data of the simulation model.

[0032] In an alternative solution, the above-mentioned built-in chip can be embedded in the controller of the air conditioner, or on either side of the controller of the air conditioner; the controller can be the controller of the internal unit of the air conditioner, or It can be the controller of the outdoor unit of the air conditioner.

[0033] The above simulation model can be a mathematical model of an air conditioner built in a laboratory environment. The simulation model can quickly simulate various working conditions and fault conditions of the air conditioner, avoiding the problem of increased costs caused by the use of real machine operations. This application obtains a large amount of simulation data through the simulation model of the air conditioner, which greatly improves the speed of obtaining the simulation data, and at the same time provides a large amount of training data for the training of the neural network model.

[0034] The aforementioned control model may be a neural network model. The neural network model is described based on the mathematical model of the neuron. It has powerful parallel distributed processing capabilities, high robustness and fault tolerance, distributed storage and learning capabilities, and can fully approximate complex nonlinear relationships.

[0035] After obtaining a large amount of simulation data through the simulation model of the air conditioner, use the simulation data to train the neural network model to obtain the trained neural network model, and the desired output result can be obtained.

[0036] It is easy to notice that because the control model is pre-stored in the built-in chip, that is, the control model can be generated outside the built-in chip and then downloaded to the built-in chip. Considering that the calculation speed of the built-in chip is limited, and the artificial intelligence algorithm to control the neural network model requires a large amount of input and output data to predict more accurately, users can connect to the cloud server through the wireless network module, and perform the training of the control model in the cloud server. Test and update. In addition, the cloud server can also assist in calculating a large amount of simulation data for neural network learning and training, making model predictions more accurate. And if the control model is generated in the controller or built-in chip of the air conditioner, it will inevitably affect the resource allocation of the built-in chip or processor.

[0037] In step S104, the control strategy of the air conditioner is adjusted according to the control model to make the air conditioner operate under the target working condition, wherein the control strategy is used to determine the working mode of the air conditioner to be used.

[0038] In an optional solution, the above control strategy can be the traditional control strategy of the air conditioner, or the initial control strategy after the air conditioner is started. The control strategy can be through the compressor, the air deflector, the fan of the indoor unit, and the fan of the outdoor unit. , Electronic expansion valve, evaporator, condenser and other actuators for implementation.

[0039] The above-mentioned working modes to be used can be rapid heating and cooling mode, best power saving mode, elderly mode, sleep mode, maternal and child mode, etc.; the above-mentioned target working conditions can be various actuators running in the working mode to be used. Under the best control parameters, the control parameters can be obtained through artificial intelligence algorithms, such as optimization control. For example, the best parameters can be the maximum energy efficiency based on the same heat exchange amount, and the maximum heat exchange amount based on higher energy efficiency. Compressor frequency under control, air deflector position, fan frequency of indoor unit, fan frequency of outdoor unit, opening degree of electronic expansion valve, etc.

[0040] After reading the control model pre-stored in the built-in chip of the air conditioner, the control model can be used to adjust the current control strategy of the air conditioner in real time to make it run under the target operating conditions.

[0041] It should be noted that by adjusting the traditional control strategy, unpredictable errors in the control model derived from artificial intelligence algorithms can be avoided.

[0042] In an alternative embodiment, the user turns on the air conditioner, and the selected working mode to be used is the best power saving mode, and the air conditioner runs under the initial control parameters and initial state parameters pre-configured for the rapid heating and cooling mode. When the air conditioner controller reads the control model from the built-in chip, it will use the control model to continuously adjust the compressor frequency, expansion valve opening, fan speed and other parameters, so that the air conditioner can ensure the same amount of heat exchange. The control strategy operation that can achieve the maximum energy efficiency (=capacity/power) to achieve low-power operation, achieve the purpose of saving power to the greatest extent, and meet the needs of users for energy saving and power saving.

[0043] It should be noted that the "capacity" of the air conditioner in this application refers to the refrigeration capacity or heating capacity of the air conditioner, which is also called the heat exchange amount.

[0044] In the above embodiment, the pre-stored control model is first read from the built-in chip of the air conditioner. The control model is a neural network model. The neural network model is obtained through machine learning training using multiple sets of data. Including: the simulation data of the simulation model of the air conditioner; and then adjust the control strategy of the air conditioner according to the control model to make the air conditioner run under the target working condition, where the control strategy is used to determine the working mode of the air conditioner to be used. Compared with the related technology, this application adjusts the current control strategy of the air conditioner through the control model pre-stored in the built-in chip of the air conditioner, and solves the problem that the control strategy of the air conditioner in the related technology is relatively fixed and single, resulting in the difficulty of achieving the best performance of the air conditioner. The technical problem achieved the goal of maximizing the capacity of the air conditioner. It is easy to notice that this application realizes the rapid prediction of the operating state of the air conditioner through the simulation model and the neural network model of the air conditioner, and uses intelligent control algorithms, such as optimization control, to generate the best operating strategy for different working modes from the prediction results, and executes each of the existing air conditioners. Coupling control between devices maximizes the ability of the air conditioner.

[0045] Optionally, the simulation data includes: simulation control parameters and simulation state parameters, where the simulation control parameters are input parameters of the simulation model, the simulation control parameters are processed by the simulation model, and the simulation state parameters are output.

[0046] In an optional solution, the above-mentioned simulation control parameters may be compressor frequency, air deflector position, fan frequency of indoor unit, fan frequency of outdoor unit, opening of electronic expansion valve, room temperature and humidity, etc. The above-mentioned simulation state parameters It can be used for heat exchange, energy efficiency, refrigerant tube temperature, motor current, etc. of the air conditioner.

[0047] figure 2 A schematic diagram of a simulation model of an optional air conditioner is shown. Such as figure 2 As shown, the simulation model of the air conditioner can obtain the heat exchange by inputting the compressor frequency, the position of the air deflector, the fan frequency of the indoor unit, the fan frequency of the outdoor unit, the opening of the electronic expansion valve, the room temperature and humidity and other simulation control parameters. Simulation status parameters such as quantity, energy efficiency, refrigerant tube temperature, motor current, etc. A large number of different simulation control parameters and corresponding sets of simulation state parameters can be simulated through the simulation model of the air conditioner.

[0048] Optionally, the neural network model is obtained through machine learning training using simulation control parameters and simulation state parameters, and the actual state parameters obtained by using the neural network model to process actual control parameters that are adapted to the simulation control parameters, the simulation state parameters Make corrections.

[0049] In an optional solution, the above-mentioned actual control parameters may be derived from big data, such as actual control parameters of multiple air conditioners in an Internet of Things environment.

[0050] Big data technology can integrate control parameters and state parameters in different regions and climates across the country.

[0051] image 3 A schematic diagram of an optional multilayer neural network model is shown. Such as image 3 As shown, the input parameters of the input layer include simulation control parameters such as compressor frequency, room temperature and humidity, air deflector position, electronic expansion valve opening, internal and external fan speed, etc. The output parameters of the output layer include air conditioning capacity, power, energy efficiency, Simulation state parameters such as refrigerant tube temperature and motor current. The hidden layer can be one or more layers. Therefore, by using simulation control parameters and simulation state parameters through machine learning training, the aforementioned neural network model can be obtained, and the corresponding relationship between the input parameters of the input layer and the output parameters of the output layer can be determined.

[0052] Further, the corresponding relationship between the input parameters and the output parameters can be corrected by using the neural network model to adapt the simulation control parameters to the training results obtained after the actual control parameters are trained, that is, the actual state parameters, so that the simulation The neural network model trained from the simulation data of the model is closer to the real operating conditions of the air conditioner.

[0053] Optionally, when the target operating condition is the state parameter maximization, step S104 adjusts the control strategy of the air conditioner according to the control model to make the air conditioner operate under the target operating condition, which may specifically include the following steps:

[0054] Step S1041: Acquire initial control parameters and initial state parameters pre-configured for the working mode to be used in the control strategy.

[0055] In an optional solution, the above-mentioned state parameters may be initial heat exchange, initial energy efficiency, and the like.

[0056] Step S1042, controlling the air conditioner to operate under the initial control parameters.

[0057] In step S1043, the initial control parameter is continuously modified based on the control model until the state parameter corresponding to the modified initial control parameter is greater than the initial state parameter.

[0058] The above scheme can make the air conditioner operate in the rapid heating and cooling mode, that is, the compressor frequency, expansion valve opening, fan speed and other parameters can be controlled within the operating range of the air conditioner reliability parameters. The air conditioner can ensure higher energy efficiency (= heat exchange /Power) based on the maximum heat exchange control strategy that can be achieved.

[0059] In an optional embodiment, the working mode to be used selected by the user is the rapid heating and cooling mode, and the air conditioner runs under the initial control parameters and initial state parameters pre-configured for the rapid heating and cooling mode. After the air conditioner controller reads the control model from the built-in chip, it will use the control model to continuously adjust the control parameters, for example, adjust the compressor frequency, expansion valve opening, fan speed and other parameters, until the modified control parameters The corresponding heat exchange is greater than the initial heat exchange, that is, the air conditioner operates according to the control strategy that can ensure the maximum heat exchange on the basis of ensuring higher energy efficiency (= heat exchange/power), and adjusts the state parameters under this time Compare with the initial state parameters. If the state parameter corresponding to the modified control parameter is less than the initial state parameter, it means that the adjustment did not maximize the heat transfer. Then continue iterative adjustment until the state parameter corresponding to the modified control parameter is greater than the initial state parameter. This realizes the rapid heating and cooling of the air conditioner, and meets the needs of users for rapid cooling and heating of the room.

[0060] Optionally, when the target operating condition is the state parameter minimization, step S104 adjusts the control strategy of the air conditioner according to the control model to make the air conditioner operate under the target operating condition, which may specifically include the following steps:

[0061] Step S1044: Acquire initial control parameters and initial state parameters pre-configured for the working mode to be used in the control strategy.

[0062] In an optional solution, the above-mentioned state parameter may be initial power and the like.

[0063] Step S1045, controlling the air conditioner to operate under the initial control parameters.

[0064] In step S1046, the initial control parameter is continuously modified based on the control model until the state parameter corresponding to the modified initial control parameter is less than the initial state parameter.

[0065] The above scheme can make the air conditioner operate in the best energy-saving mode, that is, within the operating range of air conditioner reliability parameters, control the compressor frequency, expansion valve opening, fan speed and other parameters, and the air conditioner can guarantee the same amount of heat exchange. The control strategy of the minimum power reached is run.

[0066] In an optional embodiment, the working mode to be used selected by the user is the best energy-saving mode, and the air conditioner runs under the initial control parameters and initial state parameters pre-configured for the optimal energy-saving mode. After the air conditioner controller reads the control model from the built-in chip, it will use the control model to continuously adjust the control parameters, for example, adjust the compressor frequency, expansion valve opening, fan speed and other parameters, until the modified control parameters The corresponding power is less than the initial power, that is, the air conditioner operates according to a control strategy that can ensure the minimum power that can be achieved on the basis of the same heat exchange amount, and the state parameters under this adjustment are compared with the initial state parameters. If the state parameter corresponding to the modified control parameter is greater than the initial state parameter, it means that the adjustment has not reached the power minimization, so continue to iteratively adjust until the state parameter corresponding to the modified control parameter is less than the initial state parameter. The air conditioner saves electricity to the greatest extent and meets the needs of users for energy saving.

[0067] Through the above steps, the rapid heating and cooling mode and the best energy-saving working mode of the air conditioner are realized, and each working mode can be selected according to the needs of the user.

[0068] It should be noted that the above adjustment process needs to ensure that the various control parameters are operated within the reliability range, which is the operating prerequisite for realizing the rapid heating and cooling mode of the air conditioner and the best energy-saving working mode.

[0069] Optionally, before performing step S104 to adjust the control strategy of the air conditioner according to the control model, the above method may further include:

[0070] In step S1031, if it is determined that the reading of the control model fails, resume using the initial operating parameters to control the operation of the air conditioner.

[0071] In an optional solution, the above-mentioned reading failure may be interruption of the transmission path of the control model, malfunction of the built-in chip, failure of the reading function of the controller, etc.

[0072] Under normal circumstances, the controller can directly read the control model pre-stored in the built-in chip, and then use the control model to adjust and optimize the control strategy of the air conditioner. However, if the controller fails to read the control model, the air conditioner can still operate according to the original control strategy, thereby ensuring the reliable operation of the air conditioner.

[0073] Optionally, the control model is updated through an external device connected in communication with the built-in chip.

[0074] In an optional solution, the above-mentioned external device may be a cloud server, a computer, etc.

[0075] In the above solution, the user can download the control model online, select the control model that meets the user's living environment and usage requirements, and realize the update of the control model.

[0076] Taking into account the differences in users’ living locations and climatic environment characteristics, the optimal operating conditions of air conditioners are different. For example, in cold winter, whether air conditioners in different regions are frosted and the thickness of the frost layer is in the air conditioning heating mode. Differences have different effects on the performance of air conditioning equipment. Therefore, users can use external devices to update the control model.

[0077] It should be noted that the built-in chip can selectively download the control model corresponding to the special mode (for example: child mode, elderly mode, mother-infant mode, sleep mode, etc.) to meet the needs of individual mode, improve user experience, and The chip capacity can also be reduced to reduce hardware costs, thereby solving the problem that the control model is difficult to update.

[0078] It is easy to notice that cloud servers can use big data technology to integrate control strategies operating in different regions of the country and in different climates, which helps to update and optimize the control model, and also helps to reduce chip costs (capacity and computing speed).

[0079] Optionally, after performing step S104 to adjust the control strategy of the air conditioner according to the control model, the above method may further include: step S105, using a self-learning model to optimize the adjusted control strategy, wherein the self-learning model can be generated based on the following steps:

[0080] Step S1051: Obtain historical information of the air conditioner and historical control strategies under corresponding historical target operating conditions, where the historical information includes at least one of the following: historical climate information, historical room information, and historical key information.

[0081] In an optional solution, the historical room information may be the size of the room and the temperature change rate of the room, and the historical key information may be the remote control key selection of the air conditioner by the user.

[0082] In step S1052, an initial self-learning model is trained based on historical information and historical control strategies to obtain a trained self-learning model.

[0083] In the above scheme, the self-learning model collects the actual control parameters and actual state parameters of the air conditioner, learns the user's usage habits to optimize the adjusted control strategy, so that the actual operation of the air conditioner is more in line with the user's usage habits, and realizes the intelligent, Individual needs.

[0084] In an optional embodiment, the self-learning model learns the historical use environment of the air conditioner, such as climate information, and the corresponding historical control strategy. Because the operating states of different types of air conditioners are different in different regions, the self-learning model can be combined with the Wifi module to obtain local climate characteristics from the cloud, learn the air conditioner climate and analyze big data, and then optimize the adjusted control strategy to make The actual operation reaches the optimal operating state of the air conditioner.

[0085] In another optional embodiment, the self-learning model can learn the size of the room every time the air conditioner is turned on until the room temperature reaches the set temperature. For example, the same indoor and outdoor temperature and set temperature can be used for the second time. The start-up temperature reaches the set temperature faster than the first time, indicating that the air conditioner has a larger cooling/heating capacity. Next time, the neural network model and artificial intelligence algorithm are used to control the coupling between the compressor, internal fan, electronic expansion valve and other actuators. For example, reduce the operating frequency of the compressor, reduce the speed of the internal fan, and change the size of the electronic expansion valve to accurately reduce the heat exchange output of the air conditioner. In the same way, the second start-up is slower than the first time the temperature reaches the set temperature, indicating that the cooling/heating capacity of the air conditioner is larger. Next time, the coupling control algorithm between the actuators is used to increase the heat exchange output of the air conditioner to make the room and Match the output capacity of the air conditioner. Through continuous learning, the self-learning model can determine the actual load required by the actual room size.

[0086] In another optional embodiment, the self-learning model records the user's remote control and APP button usage frequency, uses big data technology to perform statistical analysis on the records, and learns the user's usage habits locally in the air conditioner, and the next time the air conditioner runs At times, it will give priority to the control strategy that users are more accustomed to, reduce air conditioning-related processing procedures, and improve the comfort and user experience of air conditioning.

[0087] Figure 4 A schematic diagram of the principle of an optional control strategy for adjusting the air conditioner according to the control model is shown. Such as Figure 4 As shown, in view of the fact that the multi-layer neural network model requires a large amount of input data and output data to obtain more accurate prediction results, users can obtain the control model through external equipment (for example: cloud server), and integrate the whole country through big data technology. Cloud data obtained from air-conditioning control strategies running in different regions and different environments assists in training the neural network model, thereby making the prediction results of the neural network model more accurate. Then, the trained control model is written into the built-in chip through the Wifi module, and the model can be updated regularly or irregularly. In order to make the air conditioner enter the rapid heating and cooling mode or the best power saving mode, the control model continuously adjusts the operating control parameters such as compressor frequency, electronic expansion valve opening, internal and external fan speed, and air deflector position under the traditional control strategy. Until the operating state parameters that meet the rapid heating and cooling mode or the best power saving mode are obtained. Finally, the self-learning model further optimizes the adjusted control strategy so that the operation of the air conditioner meets the regional characteristics and user habits.

[0088] In the above embodiment, the pre-stored control model is first read from the built-in chip of the air conditioner. The control model is a neural network model. The neural network model is obtained through machine learning training using multiple sets of data. Including: the simulation data of the simulation model of the air conditioner; and then adjust the control strategy of the air conditioner according to the control model to make the air conditioner run under the target working condition, where the control strategy is used to determine the working mode of the air conditioner to be used. Compared with the related technology, this application adjusts the current control strategy of the air conditioner through the control model pre-stored in the built-in chip of the air conditioner, and solves the problem that the control strategy of the air conditioner in the related technology is relatively fixed and single, resulting in the difficulty of achieving the best performance of the air conditioner. The technical problem achieved the goal of maximizing the capacity of the air conditioner. It is easy to notice that this application realizes the rapid prediction of the operating state of the air conditioner through the simulation model and the neural network model of the air conditioner, and uses intelligent control algorithms, such as optimization control, to generate the best operating strategy for different working modes from the prediction results, and executes each of the existing air conditioners. Coupling control between devices to maximize the capacity of the air conditioner; obtain a large amount of training data through the simulation model of the air conditioner; modify the actual control parameters to make the neural network model more accurate; optimize the control to make the air conditioner run at the best level Under circumstances; updating the control model through external equipment can download personalized working modes according to the user’s living environment and use needs, saving local resources of the air conditioner; optimizing the air conditioning control strategy through the self-learning model, further reducing user operations and improving User experience.

[0089] Example 2

[0090] According to an embodiment of the present invention, a device for controlling an air conditioner is provided, Figure 5 It is a schematic diagram of a device for controlling an air conditioner according to an embodiment of the present application. Such as Figure 5 As shown, the device 500 includes a reading module 502 and an adjustment module 504.

[0091] Among them, the reading module 502 is used to read the pre-stored control model from the built-in chip of the air conditioner. The control model is a neural network model. The neural network model is obtained through machine learning training using multiple sets of data. The set of data includes: simulation data of the simulation model of the air conditioner; the adjustment module 504 is used to adjust the control strategy of the air conditioner according to the control model to make the air conditioner run under the target working condition, where the control strategy is used to determine the working mode of the air conditioner to be used .

[0092] Optionally, the simulation data includes: simulation control parameters and simulation state parameters, where the simulation control parameters are input parameters of the simulation model, the simulation control parameters are processed by the simulation model, and the simulation state parameters are output.

[0093] Optionally, the neural network model is obtained through machine learning training using simulation control parameters and simulation state parameters, and the actual state parameters obtained by using the neural network model to process actual control parameters that are adapted to the simulation control parameters, the simulation state parameters Make corrections.

[0094] Optionally, in the case that the target operating condition is the state parameter maximization, the adjustment module includes: a first acquisition module, configured to acquire the initial control parameters and initial state parameters pre-configured for the working mode to be used in the control strategy; The control module is used to control the air conditioner to operate under the initial control parameters; the first correction module is used to continuously correct the initial control parameters based on the control model until the state parameters corresponding to the corrected initial control parameters are greater than the initial state parameters.

[0095] Optionally, when the target operating condition is the state parameter minimization, the adjustment module includes: a second acquisition module, configured to acquire the initial control parameters and initial state parameters pre-configured for the working mode to be used in the control strategy; second The control module is used to control the air conditioner to operate under the initial control parameters; the second correction module is used to continuously correct the initial control parameters based on the control model until the state parameters corresponding to the corrected initial control parameters are less than the initial state parameters.

[0096] Optionally, the above-mentioned device may further include: a restoration module, which is used to restore the use of the initial operating parameters to control the operation of the air conditioner if it is determined that the reading of the control model fails before adjusting the control strategy of the air conditioner according to the control model.

[0097] Optionally, the control model is updated through an external device connected in communication with the built-in chip.

[0098] Optionally, the above-mentioned device may further include: an optimization module for optimizing the adjusted control strategy using a self-learning model after adjusting the control strategy of the air conditioner according to the control model, wherein the self-learning model is generated based on the following modules: third acquisition The module is used to obtain historical information of the air conditioner and the historical control strategy under the corresponding historical target operating conditions. The historical information includes at least one of the following: historical climate information, historical room information, and historical key information; a training module for The initial self-learning model is trained based on historical information and historical control strategies to obtain a trained self-learning model.

[0099] It should be noted that the above-mentioned reading module 502 and adjustment module 504 correspond to step S102 to step S104 in Embodiment 1. The examples and application scenarios implemented by these two modules are the same as the corresponding steps, but are not limited to the above-mentioned embodiment. 1 Disclosure.

[0100] Example 3

[0101] According to an embodiment of the present invention, there is provided an air conditioner, including:

[0102] The built-in chip is set on the controller of the air conditioner for pre-stored control models. The control model is a neural network model. The neural network model is obtained through machine learning training using multiple sets of data. Each of the multiple sets of data includes: Simulation data of the simulation model of the air conditioner.

[0103] The controller is communicatively connected with the built-in chip to read the control model and adjust the control strategy of the air conditioner according to the control model to make the air conditioner run under the target working condition. The control strategy is used to determine the working mode of the air conditioner to be used.

[0104] Further, the air conditioner may also include other modules in Embodiment 2, which will not be repeated here.