Intelligent automatic control method and device for indoor electrical equipment based on Transform

Abstract

The invention relates to an intelligent automatic control method and device for indoor electrical equipment based on Transform, and belongs to the technical field of intelligent equipment control. Comprising the steps and modules of data processing, transformer model construction, model training, model evaluation and model prediction. According to the method and device, the use rule of indoor electrical equipment is learned by utilizing a transformer model, and a corresponding result is predicted in combination with the current indoor environment and the use rule of the electrical equipment; meanwhile, prediction results are transmitted to the lighting system, the temperature control system, the curtain system and the ventilation system in an instruction mode, so that the effect of intelligently adjusting all intelligent electrical equipment is achieved, and the effects of energy conservation and emission reduction are achieved while a suitable living environment is created for a user.

Description

Technical Field

[0001] The present invention relates to an indoor electrical equipment automatic control method and device, in particular to an indoor electrical equipment intelligent automatic control method and device based on Transformer, belonging to the technical field of intelligent equipment control. Background Art

[0002] With the rapid development of my country's economy and the continuous improvement of living standards, in order to meet the various needs of indoor buildings, a large number of electrical equipment need to be configured indoors, such as lighting systems, temperature control systems, curtain systems and ventilation systems, etc. In order to realize the control operation of all indoor electrical equipment, corresponding control devices need to be configured.

[0003] In the prior art, the control of indoor electrical equipment often requires users to switch on or off or adjust each electrical equipment according to actual needs.

[0004] However, in actual application, although the usage patterns of indoor electrical equipment vary depending on the indoor type, the usage patterns of the indoor equipment are often relatively fixed. The existing indoor automatic control system does not take into account the usage patterns of the overall electrical equipment in each room and optimizes its control intelligently. Therefore, on the one hand, it creates an additional operating burden for users and fails to achieve true energy-saving and environmental protection effects. On the other hand, the degree of intelligence in the building is not enough, the user experience is poor, and it cannot meet the users' growing demand for intelligence. Summary of the invention

[0005] In view of the above defects in the above-mentioned prior art, the present invention proposes a method and device for intelligent automatic control of indoor electrical equipment based on Transformer. The method and device use the transformer model to learn the usage rules of indoor electrical equipment, combine the current indoor environment and the usage rules of electrical equipment to predict the corresponding results, and transmit the predicted results to the lighting system, temperature control system, curtain system and ventilation system respectively through instructions, so as to achieve the effect of intelligently adjusting each intelligent electrical equipment, create a suitable living environment for users, and achieve the effect of energy saving and emission reduction.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0007] A transformer-based intelligent automatic control method for indoor electrical equipment includes the following steps:

[0008] S1. Data processing: standardize the historical data collected by the indoor environment detection sensor and divide the standardized data set into a training set and a test set;

[0009] S2, transformer model construction: Use the loop mechanism to sequentially construct the feature embedding layer, encoder layer and classifier layer corresponding to the lighting system, temperature control system-mode, temperature control system-wind speed, temperature control system-intensity, curtain system and ventilation system;

[0010] S3, model training: set the cross loss function and gradient zeroing required for model training, and use the loop mechanism to train the lighting system model, temperature control system-mode model, temperature control system-wind speed model, temperature control system-intensity model, curtain system model and ventilation system model in sequence;

[0011] S4, model evaluation: Use the lighting system, temperature control system-mode, temperature control system-wind speed, temperature control system-intensity, curtain system and ventilation system test sets to obtain the accuracy of each model. If the accuracy of the model is improved by more than 0.01 each time, the best model weight dictionary is saved, and the model training step S3 is jumped at the same time. If the accuracy is improved by less than 0.01 more than 100 times, the model training and model evaluation are stopped;

[0012] S5, model prediction: Based on the data collected by the indoor environment detection sensor at this moment, use the best model to output the corresponding prediction results according to relevant needs, and execute corresponding commands based on the relevant results.

[0013] Furthermore, in the data processing step of S1, the data collected by the indoor environment detection sensor include indoor light intensity, humidity, temperature, carbon dioxide concentration, formaldehyde concentration, humidity, indoor light intensity, whether there are people, PM10 particles, PM2.5 concentration, atmospheric pressure, volatiles, lighting system, temperature control system, curtain system and ventilation system data.

[0014] Furthermore, in the data processing step of S1, the data standardization includes uniformly converting the int type in the data into the float type, standardizing the data required in the ventilation system using the Standardscaler function, and converting the data of the lighting system, temperature control system, curtain system, and ventilation system into corresponding numerical labels, as shown in the following table:

[0015]

[0016]

[0017] Among them, the Standardscaler function is a Python function for feature scaling, which scales each feature to a range with a mean of 0 and a standard deviation of 1. This helps to eliminate differences between features and enables the model to predict results more accurately.

[0018] Furthermore, in the data processing step of S1, 80% of the standardized data is used as a training set, and 20% of the standardized data is used as a test set.

[0019] Furthermore, in the transformer model construction step of S2, the feature embedding layers corresponding to the lighting system, temperature control system-mode, temperature control system-wind speed, temperature control system-temperature, curtain system and ventilation system have corresponding data features as shown in the following table:

[0020]

[0021] Furthermore, in the transformer model construction step of S2, the Encoder encoder layer includes only one Encoder encoder layer, wherein the Encoder encoder layer includes a sub-layer consisting of a multi-head attention layer and a feedforward neural network, and each sub-layer is followed by a residual connection and a layer normalization operation, which are collectively referred to as an Add&Norm operation.

[0022] Furthermore, in the transformer model construction step of S2, the classifier layer includes a dropout layer and a linear fully connected layer.

[0023] Furthermore, in the S5 model prediction step,

[0024] Assume that the model predicts the temperature in the temperature control system. If the PIR infrared human body sensing sensor is turned on and the collected data shows that there is no one in the room, then the temperature of the temperature control system predicted by the model is set to be equal to the temperature collected by the current indoor environment detection. Otherwise, the best model is used to predict the mode m and intensity t in the temperature control system. strength , and then use the following formula to calculate the new temperature of the temperature control system. The calculation formula is as follows.

[0025]

[0026] Among them, t new Indicates the temperature to which the temperature control system needs to be adjusted, t c Indicates the temperature of the temperature control system collected by the current indoor environment detection, t min Indicates the minimum temperature of the temperature control system is 16℃, t max Indicates the maximum temperature of the temperature control system is 32℃, t strength Indicates the wind speed of the temperature control system, m=0, 1, 2 respectively indicate that the temperature control system mode is no operation, cooling and heating mode;

[0027] Assume that the model predicts the lighting system, curtain system, ventilation system, temperature control system-mode, and temperature control system-wind speed. If the PIR infrared human body sensing sensor is on and there is no one in the room, the model prediction result is set as the lighting system is off, the curtain system is off, the ventilation system is off, the temperature control system-mode is no operation, and the temperature control system-wind speed is 0. Otherwise, the result prediction is made according to the best model of each system.

[0028] A transformer-based intelligent automatic control device for indoor electrical equipment, comprising:

[0029] Data processing module: used to perform data standardization on the historical data collected by the indoor environment detection sensor, and divide the standardized data set into a training set and a test set;

[0030] Transformer model building module: used to use the loop mechanism to sequentially build the feature embedding layer, encoder layer and classifier layer corresponding to the lighting system, temperature control system-mode, temperature control system-wind speed, temperature control system-intensity, curtain system and ventilation system;

[0031] Model training module: used to set the cross-loss function and gradient zeroing required for model training, and use the loop mechanism to train the lighting system model, temperature control system-mode model, temperature control system-wind speed model, temperature control system-intensity model, curtain system model and ventilation system model in sequence;

[0032] Model evaluation module: used to obtain the accuracy of each model using the lighting system, temperature control system-mode, temperature control system-wind speed, temperature control system-intensity, curtain system and ventilation system test sets. If the accuracy of the model is improved by more than 0.01 each time, the best model weight dictionary is saved and the model training step S3 is jumped to. If the accuracy is improved by less than 0.01 for more than 100 times, the model training and model evaluation are stopped.

[0033] Model prediction module: It is used to output corresponding prediction results based on the data collected by the indoor environment detection sensor at the moment, using the best model according to relevant needs, and execute corresponding commands based on the relevant results.

[0034] A computer-readable storage medium includes a stored program, wherein the program executes the above-mentioned transformer-based indoor electrical equipment intelligent automatic control method when running.

[0035] The technical solution of the present invention can achieve the following beneficial effects:

[0036] On the one hand, the present invention can automatically adjust the state of electrical equipment according to the actual environment without the need for manual or voice control by the user, thereby creating a more suitable living environment for the user; on the other hand, the model prediction results are relatively accurate and can provide accurate instructions for electrical equipment to achieve the effect of energy saving and emission reduction. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a logic flow chart of the transformer-based intelligent automatic control method for indoor electrical equipment of the present invention;

[0038] Figure 2 This is the transformer model diagram of the present invention. DETAILED DESCRIPTION

[0039] The following is combined with Figure 1-2 The present invention is further described in detail with specific implementations to facilitate a clear understanding of the present invention, but they do not constitute a limitation on the present invention.

[0040] As attached Figure 1-2 As shown, the present invention can realize the intelligent control functions of the lighting system, the temperature control system, the curtain system and the ventilation system. In order to enable those skilled in the art to more clearly understand the specific method of the present invention, the implementation process of the present invention is described in detail by taking the three specific embodiments of the temperature control system-mode, the temperature control system-intensity and the temperature control system temperature as examples.

[0041] Example 1: Temperature Control System-Mode

[0042] like Figure 1 As shown, the transformer-based temperature control system-mode intelligent control method provided in an embodiment of the present invention includes the following steps.

[0043] S1. Data processing: Standardize the historical data collected by the indoor environment detection sensor, including temperature, temperature control system-mode, etc., and divide the standardized data set into a training set and a test set in a ratio of 8:2. That is, 80% of the standardized data is used as the training set, and 20% of the standardized data is used as the test set. Among them, data standardization includes uniformly converting the int type in the data into the float type, standardizing the required data in the ventilation system using the Standardscaler function, and converting the temperature control system data into corresponding numerical labels, as shown in the following table:

[0044] system data Value Label Temperature control system-mode 0,1,2 (0 means no operation; 1 means cooling; 2 means heating) 0；1

[0045] Among them, the Standardscaler function is a Python function for feature scaling, which scales each feature to a range with a mean of 0 and a standard deviation of 1.

[0046] S2. Transformer model construction: Construct the feature embedding layer, encoder layer and classifier layer of the temperature control system-mode. The model is as follows Figure 2 As shown in the figure, the encoder layer includes sub-layers composed of a multi-head attention layer (Multi-HeadAttention) and a feed-forward neural network (Feed-Forward Network). Each sub-layer is followed by a residual connection and a layer normalization (LayerNorm) operation, which are collectively called the Add&Norm operation; the classifier layer includes a dropout layer and a linear fully connected layer. The feature embedding of the temperature control system-mode refers to the embedded value of the temperature.

[0047] S3, model training: set the cross loss function and gradient clearing required for temperature control system-mode model training, and train the model using the temperature control system-mode training data set.

[0048] S4, model evaluation: Use the temperature control system-mode test set to obtain the accuracy of the lighting system training model. If the accuracy is improved by more than 0.01 each time, the temperature control system-mode training model is saved, and the next round of temperature control system-mode model training is continued. When the accuracy is improved by less than 0.01 for more than 100 times, the temperature control system-mode model training and evaluation is stopped, and the best temperature control system-mode model is saved.

[0049] S5, model prediction: according to the temperature, PIR and PIR status values ​​collected by the indoor environment detection sensor at this moment, if the PIR is in the open state and the PIR value is equal to 0, the model output value is 0, and no operation command is executed on the temperature control system-mode. Otherwise, the corresponding prediction result (0 means no operation; 1 means cooling; 2 means heating) is output using the optimal temperature control system-mode model. If the prediction result is 0, no adjustment command is executed on the temperature control system mode. If the prediction result is 1, a cooling command is executed on the temperature control system mode. Otherwise, a heating command is executed on the temperature control system mode.

[0050] Example 2: Temperature Control System - Strength

[0051] like Figure 1 The transformer-based temperature control system-intensity intelligent control method provided in an embodiment of the present invention includes the following steps.

[0052] S1. Data processing: Standardize the historical data collected by the indoor environment detection sensor, including temperature, temperature control system-intensity, etc., and divide the standardized data set into a training set and a test set in an 8:2 ratio, that is, 80% of the standardized data is used as the training set, and 20% of the standardized data is used as the test set. Among them, data standardization includes uniformly converting the int type in the data into a float type, standardizing the data required in the ventilation system using the Standardscaler function, and converting the data of the lighting system, temperature control system, curtain system, and ventilation system into corresponding numerical labels, as shown in the following table:

[0053] system data Value Label Temperature Control System - Strength 0,1,2,3 (cooling / heating degree: 0; 1; 2; 3 levels) 0；1；2；3

[0054] The Standardscaler function is a Python function for feature scaling that scales each feature to a range with a mean of 0 and a standard deviation of 1.

[0055] S2. Transformer model construction: Construct the temperature control system-intensity feature embedding layer, encoder layer and classifier layer. The model is as follows Figure 2 As shown in the figure, the encoder layer includes sub-layers composed of a multi-head attention layer (Multi-HeadAttention) and a feed-forward neural network (Feed-Forward Network). Each sub-layer is followed by a residual connection and a layer normalization (LayerNorm) operation, which are collectively called the Add&Norm operation; the classifier layer includes a dropout layer and a linear fully connected layer. The feature embedding of the temperature control system-module strength refers to the embedded value of the temperature.

[0056] S3, model training: set the cross loss function and gradient clearing required for temperature control system-intensity model training, and train the model using the temperature control system-intensity training data set.

[0057] S4, model evaluation: Use the temperature control system-intensity test set to obtain the accuracy of the lighting system training model. If the accuracy is improved by more than 0.01 each time, the temperature control system-mode training model is saved, and the next round of temperature control system-intensity model training is continued. When the accuracy is improved by less than 0.01 for more than 100 times, the temperature control system-intensity model training and evaluation is stopped, and the best temperature control system-intensity model is saved.

[0058] S5. Model prediction: According to the temperature, PIR and PIR status values ​​collected by the indoor environment detection sensor at this moment, if the PIR is in the on state and the PIR infrared human body sensing sensor value is equal to 0, the model output value is 0, and the -intensity value of the temperature control system is 0. Otherwise, the optimal temperature control system -mode model is used to output the corresponding prediction result, where the prediction result of 0 indicates that the temperature control system intensity value is 0; the prediction result of 1 indicates that the temperature control system intensity value is 1; the prediction result of 2 indicates that the temperature control system intensity value is 2; the prediction result of 3 indicates that the temperature control system intensity value is 3.

[0059] Example 3: Intelligent temperature control of temperature control system

[0060] Since the intelligent control of the temperature of the temperature control system is related to the current temperature of the temperature control system, the temperature control system mode and the temperature control system strength, the premise for the specific implementation of Example 3 is that both Specific Example 1 and Specific Example 2 have been implemented.

[0061] Model prediction: Based on the temperature, PIR and PIR status values ​​collected by the indoor environment detection sensor at this moment, if the PIR is in the open state and the PIR value is equal to 0, the model output value is the collected temperature of the indoor environment detection sensor. Otherwise, first implement specific example 1 to obtain the prediction result of the temperature control system-mode, then implement specific example 2 to obtain the prediction result of the temperature control system-intensity, and finally use the following formula to calculate the temperature to which the temperature control system needs to be adjusted.

[0062]

[0063] Among them, t new Indicates the temperature to which the temperature control system needs to be adjusted, t c Indicates the temperature of the temperature control system collected by the current indoor environment detection, t min Indicates the minimum temperature of the temperature control system is 16℃, t max Indicates the maximum temperature of the temperature control system is 32℃, t strength Indicates the strength of the temperature control system, m=0, 1, and 2 respectively indicate that the temperature control system mode is no operation, cooling, and heating mode.

[0064] Assuming that the model predicts the lighting system, curtain system, ventilation system, temperature control system-wind speed, the model prediction is performed according to the steps of specific example 1 and specific example 2.

[0065] In addition, the present invention can also use the four steps of data processing, transformer model construction, model training and model evaluation of Example 1 and Specific Example 2 in a cyclic manner to obtain the optimal models of the lighting system, curtain system, ventilation system, temperature control system-wind speed, temperature control system-mode, and temperature control system-intensity in sequence, and then obtain the optimal models of different systems according to actual needs, and use the optimal models to predict model results.

[0066] Based on 1400 historical data collected by indoor environment detection sensors, the present invention uses the naive Bayes model, MLP model, DeepML model and the method of the present invention (Transformer model) to train and evaluate the models of the lighting system, temperature control system, curtain system and ventilation system. The model evaluation accuracy is shown in the following table:

[0067]

[0068] It can be concluded from the above table that the accuracy of the indoor electrical equipment automatic control method based on the transformer model is higher than that of other models. Therefore, the indoor electrical equipment automatic control method based on the transformer model can more accurately predict the results and provide accurate instructions for the intelligent automatic control of electrical equipment.

[0069] Embodiment 4:

[0070] A transformer-based intelligent automatic control device for indoor electrical equipment in this embodiment includes:

[0071] Data processing module: used to perform data standardization on the historical data collected by the indoor environment detection sensor, and divide the standardized data set into a training set and a test set;

[0072] Transformer model building module: used to use the loop mechanism to sequentially build the feature embedding layer, encoder layer and classifier layer corresponding to the lighting system, temperature control system-mode, temperature control system-wind speed, temperature control system-intensity, curtain system and ventilation system;

[0073] Model training module: used to set the cross-loss function and gradient zeroing required for model training, and use the loop mechanism to train the lighting system model, temperature control system-mode model, temperature control system-wind speed model, temperature control system-intensity model, curtain system model and ventilation system model in sequence;

[0074] Model evaluation module: used to obtain the accuracy of each model using the lighting system, temperature control system-mode, temperature control system-wind speed, temperature control system-intensity, curtain system and ventilation system test sets. If the accuracy of the model is improved by more than 0.01 each time, the best model weight dictionary is saved and the model training step S3 is jumped to. If the accuracy is improved by less than 0.01 for more than 100 times, the model training and model evaluation are stopped.

[0075] Model prediction module: It is used to output corresponding prediction results based on the data collected by the indoor environment detection sensor at the moment, using the best model according to relevant needs, and execute corresponding commands based on the relevant results.

[0076] Embodiment 5:

[0077] A computer-readable storage medium of this embodiment includes a stored program, wherein when the program is run, the transformer-based intelligent automatic control method for indoor electrical equipment of any one of the above embodiments 1-3 is executed.

[0078] The above is only a preferred embodiment of the present invention, and does not impose any formal limitation on the structure of the present invention. The layout type and the number of uses of the present invention are not limited to this example, and can be optimized and selected according to the actual project. Any modification, equivalent change and decoration of the above embodiment based on the technical principle of the present invention that does not deviate from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A transformer-based intelligent automatic control method for indoor electrical equipment, characterized in that: The steps include: S1. Data processing: standardize the historical data collected by the indoor environment detection sensor and divide the standardized data set into a training set and a test set; S2, transformer model construction: Use the loop mechanism to sequentially construct the feature embedding layer, encoder layer and classifier layer corresponding to the lighting system, temperature control system-mode, temperature control system-wind speed, temperature control system-intensity, curtain system and ventilation system; S3, model training: set the cross loss function and gradient zeroing required for model training, and use the loop mechanism to train the lighting system model, temperature control system-mode model, temperature control system-wind speed model, temperature control system-intensity model, curtain system model and ventilation system model in sequence; S4, model evaluation: Use the lighting system, temperature control system-mode, temperature control system-wind speed, temperature control system-intensity, curtain system and ventilation system test sets to obtain the accuracy of each model. If the accuracy of the model is improved by more than 0.01 each time, the best model weight dictionary is saved, and the model training step S3 is jumped at the same time. If the accuracy is improved by less than 0.01 more than 100 times, the model training and model evaluation are stopped; S5, model prediction: Based on the data collected by the indoor environment detection sensor at this moment, use the best model to output the corresponding prediction results according to relevant needs, and execute corresponding commands based on the relevant results.

2. According to claim 1, a transformer-based intelligent automatic control method for indoor electrical equipment is characterized in that: In the data processing step of S1, the data collected by the indoor environment detection sensor include indoor light intensity, humidity, temperature, carbon dioxide concentration, formaldehyde concentration, humidity, indoor light intensity, whether there are people, PM10 particles, PM2.5 concentration, atmospheric pressure, volatiles, lighting system, temperature control system, curtain system and ventilation system data.

3. According to claim 2, a transformer-based intelligent automatic control method for indoor electrical equipment is characterized in that: In the data processing step of S1, the data standardization includes uniformly converting the int type in the data into the float type, standardizing the data required in the ventilation system using the Standardscaler function, and converting the data of the lighting system, temperature control system, curtain system, and ventilation system into corresponding numerical labels, as shown in the following table: Among them, the Standardscaler function is a Python function for feature scaling, which scales each feature to a range with a mean of 0 and a standard deviation of 1.

4. According to claim 3, a transformer-based intelligent automatic control method for indoor electrical equipment is characterized in that: In the data processing step of S1, 80% of the standardized data is used as a training set, and 20% of the standardized data is used as a test set.

5. The transformer-based intelligent automatic control method for indoor electrical equipment according to claim 4 is characterized in that: In the transformer model construction step of S2, the feature embedding layers corresponding to the lighting system, temperature control system-mode, temperature control system-wind speed, temperature control system-temperature, curtain system and ventilation system have corresponding data features as shown in the following table:

6. The transformer-based intelligent automatic control method for indoor electrical equipment according to claim 5, characterized in that: In the transformer model construction step of S2, the Encoder encoder layer only includes one Encoder encoder layer, wherein the Encoder encoder layer includes a sublayer consisting of a multi-head attention layer and a feedforward neural network, and each sublayer is followed by a residual connection and a layer normalization operation, which are collectively referred to as an Add&Norm operation.

7. The transformer-based intelligent automatic control method for indoor electrical equipment according to claim 6 is characterized in that: In the transformer model construction step of S2, the classifier layer includes a dropout layer and a linear fully connected layer.

8. The transformer-based intelligent automatic control method for indoor electrical equipment according to claim 7, characterized in that: In the S5 model prediction step, Assume that the model predicts the temperature in the temperature control system. If the PIR infrared human body sensing sensor is turned on and the collected data shows that there is no one in the room, then the temperature of the temperature control system predicted by the model is set to be equal to the temperature collected by the current indoor environment detection. Otherwise, the best model is used to predict the mode m and intensity t in the temperature control system. strength , and then use the following formula to calculate the new temperature of the temperature control system. The calculation formula is as follows: Among them, t new Indicates the temperature to which the temperature control system needs to be adjusted, t c Indicates the temperature of the temperature control system collected by the current indoor environment detection, t min Indicates the minimum temperature of the temperature control system is 16℃, t max Indicates the maximum temperature of the temperature control system is 32℃, t strength Indicates the wind speed of the temperature control system, m=0, 1, 2 respectively indicate that the temperature control system mode is no operation, cooling and heating mode; Assume that the model predicts the lighting system, curtain system, ventilation system, temperature control system-mode, and temperature control system-wind speed. If the PIR infrared human body sensing sensor is on and there is no one in the room, the model prediction result is set as the lighting system is off, the curtain system is off, the ventilation system is off, the temperature control system-mode is no operation, and the temperature control system-wind speed is 0. Otherwise, the result prediction is made according to the best model of each system.

9. A transformer-based intelligent automatic control device for indoor electrical equipment, characterized in that: include: Data processing module: used to perform data standardization on the historical data collected by the indoor environment detection sensor, and divide the standardized data set into a training set and a test set; Transformer model building module: used to use the loop mechanism to sequentially build the feature embedding layer, encoder layer and classifier layer corresponding to the lighting system, temperature control system-mode, temperature control system-wind speed, temperature control system-intensity, curtain system and ventilation system; Model training module: used to set the cross-loss function and gradient zeroing required for model training, and use the loop mechanism to train the lighting system model, temperature control system-mode model, temperature control system-wind speed model, temperature control system-intensity model, curtain system model and ventilation system model in sequence; Model evaluation module: used to obtain the accuracy of each model using the lighting system, temperature control system-mode, temperature control system-wind speed, temperature control system-intensity, curtain system and ventilation system test sets. If the accuracy of the model is improved by more than 0.01 each time, the best model weight dictionary is saved and the model training step S3 is jumped to. If the accuracy is improved by less than 0.01 for more than 100 times, the model training and model evaluation are stopped. Model prediction module: It is used to output corresponding prediction results based on the data collected by the indoor environment detection sensor at the moment, using the best model according to relevant needs, and execute corresponding commands based on the relevant results.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is run, the transformer-based intelligent automatic control method for indoor electrical equipment described in any one of claims 1 to 8 is executed.

Images