Air conditioner target temperature prediction method and system based on historical working condition data

Abstract

The invention discloses an air conditioner target temperature prediction method and system based on historical working condition data, relates to the technical field of air conditioner target temperature prediction, and solves the problem that target temperatures in different scenes are difficult to predict according to environmental working conditions when an air conditioner runs. The operation energy consumption of the air conditioning equipment when the actual reachable temperature is deviated from the set temperature value is reduced; according to the method, original data of all dimensions are processed and matched with working condition dimension data and temperature reaching values of each time of starting of the air conditioner, and abnormal values are filtered; performing correlation analysis on the working condition dimension data after data processing to obtain correlation coefficients, and selecting target features from high to low based on the correlation coefficients; training the target features through a decision tree regression algorithm to obtain a temperature prediction model, and predicting a target temperature matched with the working condition dimension data when the air conditioner is started on the basis of the temperature prediction model; long-term high-load operation of air conditioner parts can be reduced, and stability and energy conservation of air conditioner operation are promoted.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent air conditioning, and relates to an air conditioning target temperature prediction technology, and specifically to an air conditioning target temperature prediction method and system based on historical operating condition data. Background Art

[0002] With the rapid development of modern society and the complex and ever-changing environment, air conditioners have gradually become indispensable household appliances in providing healthy temperature and humidity and improving human comfort. Most existing air conditioners use built-in fuzzy logic control, which is guided by the user's set temperature and adjusts the operating status of each component according to environmental parameters such as real-time temperature difference. This method does not take into account the cooling or heating capacity in the actual user environment. When the working conditions are complex or the equipment performance is degraded, it is easy for the air conditioner to fail to meet the set value but the compressor and other components still maintain high-speed operation, which causes energy consumption to continue to increase.

[0003] The prior art (invention patent with publication number: CN112268355B) discloses an air-conditioning target temperature adjustment method and terminal device, and its technical solution is: within a first preset time, the actual temperature of the environment is obtained in real time; when the limit value and the alarm value of the actual temperature of the environment meet the preset conditions, the target temperature of the air-conditioning is adjusted so that the target temperature of the air-conditioning approaches the alarm value and the limit value of the actual temperature of the environment does not exceed the alarm value within the next first preset time; wherein, when the air-conditioning is in cooling mode, the limit value of the actual temperature of the environment is the maximum value of the actual temperature of the environment within the first preset time, and the alarm value is the upper alarm limit value; when the air-conditioning is in heating mode, the limit value of the actual temperature of the environment is the minimum value of the actual temperature of the environment within the first preset time, and the alarm value is the lower alarm limit value. This invention is difficult to predict the target temperature in different scenarios according to the environmental conditions, and reduce the operating energy consumption of the air-conditioning equipment when the actual achievable temperature deviates from the set temperature value.

[0004] Therefore, the present invention discloses an air conditioning target temperature prediction method and system based on historical operating condition data, which are used to solve the above technical problems. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes an air-conditioning target temperature prediction method and system based on historical operating condition data, which is used to solve the technical problem that it is difficult to predict the target temperature in different scenarios according to the environmental operating conditions when the air-conditioning is running, and reduce the operating energy consumption of the air-conditioning equipment when the actual achievable temperature deviates from the set temperature value. The present invention solves the above problems by screening the original data of each dimension related to the temperature change of the air-conditioning; processing the original data of each dimension to match the operating condition dimension data and the temperature value each time the air-conditioning is turned on, and filtering the abnormal values; performing correlation analysis on the data after data processing to obtain the correlation coefficient, and selecting the target feature from high to low based on the correlation coefficient; training the target feature through a decision tree regression algorithm to obtain a temperature prediction model, and predicting the target temperature matched by the operating condition dimension data when the air-conditioning is turned on based on the temperature prediction model.

[0006] To achieve the above object, a first aspect of the present invention provides an air conditioning target temperature prediction method based on historical operating condition data, comprising:

[0007] S1: Filter the original data of various dimensions related to air conditioning temperature changes;

[0008] S2: Process the original data of each dimension to match the operating condition dimension data and the temperature value each time the air conditioner is turned on, and filter out abnormal values; the temperature value is the temperature that the air conditioner can reach;

[0009] S3: Perform correlation analysis on the data of each working condition dimension after data processing to obtain the correlation coefficient, and select the target feature from high to low based on the correlation coefficient; train the target feature through the decision tree regression algorithm to obtain a temperature prediction model, and predict the target temperature matched by the working condition dimension data when the air conditioner is turned on based on the temperature prediction model.

[0010] Preferably, the screening of original data of various dimensions related to the change of air-conditioning temperature includes:

[0011] The outdoor ambient temperature, indoor temperature, indoor humidity, and various setting parameters and component status values ​​of the air conditioner space are collected in real time through sensors; the dimensional raw data includes outdoor ambient temperature, indoor temperature, indoor humidity, setting parameters and component status values;

[0012] The original dimensional data of each device is uploaded to the cloud, and the original dimensional data that are preliminarily screened out and strongly related to the cooling and heating capabilities are marked as operating condition data dimensions based on expert experience.

[0013] Preferably, the processing of the original data of each dimension to match the working condition dimension data and the temperature value reached each time the air conditioner is turned on includes:

[0014] Mark the data between each startup and shutdown of the air conditioner in the historical data as reference data, obtain the lowest indoor temperature value that can be achieved in the cooling mode or the highest indoor temperature value that can be achieved in the heating mode from several reference data, mark the lowest indoor temperature value or the highest indoor temperature value as the temperature value in the corresponding reference data, and match the operating condition dimension data at each startup moment with the corresponding temperature value.

[0015] Preferably, filtering outliers includes:

[0016] A1: prioritize the corresponding matched working condition dimension data in the following order: set temperature, set windshield, and outdoor temperature;

[0017] A2: Circularly judge each working condition dimension data and the other working condition dimension data, mark the working condition dimension data being judged as the target data, and mark the working condition dimension data in the other working condition dimension data being judged as the data to be judged; judge whether the set temperature of the target data and the data to be judged is the same; if yes, jump to A3; if no, do nothing;

[0018] A3: In cooling mode, when the wind speed of the data to be judged is not less than the wind speed of the target data, the outdoor temperature is not greater than the outdoor temperature of the target data, and the temperature value is greater than the temperature value of the target data; or in heating mode, when the wind speed of the data to be judged is not less than the wind speed of the target data, the outdoor temperature is not less than the outdoor temperature of the target data, and the temperature value is less than the temperature value of the target data, the data to be judged is regarded as abnormal data, and the temperature value of the abnormal data is corrected to the temperature value of the current cycle data.

[0019] Preferably, the correlation analysis of the data of each operating condition dimension after data processing to obtain the correlation coefficient includes:

[0020] Add two features, day and night conditions and the difference between outdoor temperature and set temperature, to the data of each working condition dimension;

[0021] The Pearson correlation coefficient calculation method is used to calculate the characteristic correlation between the dimension data of each working condition. The Pearson correlation coefficient calculation method is as follows:

[0022]

[0023] Among them, X and Y are the two types of operating condition characteristics to be calculated, Cov is the calculated covariance, σ is the standard deviation of the operating condition variable, and ρ is the correlation coefficient.

[0024] Preferably, the selecting target features from high to low based on the correlation coefficient includes:

[0025] The correlation coefficients calculated based on the data of each operating condition dimension are arranged in a matrix form, the values ​​in the last column of the matrix are marked as the positive and negative correlation coefficients between the temperature value and each operating condition feature, and the three feature dimensions with the largest absolute values ​​of the positive and negative correlation coefficients are marked as target features.

[0026] Preferably, the temperature prediction model is obtained by training the target features through a decision tree regression algorithm, including:

[0027] B1: A data set is independently established for a single device, with the target features as the sample features of the data set, the sample labels as the corresponding temperature values, and 80% of the data set is randomly sampled as the training set and 20% as the validation set;

[0028] B2: Establish training criteria, and use the training criteria to train the working condition characteristics of the data set to obtain a temperature prediction model; perform input verification and evaluation on the verification set under the temperature prediction model generated by learning the training set, and determine whether the average absolute error between the predicted target temperature value and the temperature value corresponding to the label is not greater than the minimized error; if yes, jump to B3; if no, continue to train and verify the temperature prediction model; the minimized error is obtained through experience;

[0029] B3: Perform grid search optimization training on the temperature prediction model and set the optimized model parameters; the target optimized model parameters include the maximum depth of the tree and the minimum number of samples of leaf nodes.

[0030] Preferably, the establishing of training criteria comprises:

[0031] The regression tree algorithm model is used for training. The initial maximum depth parameter of the regression tree is set to 8. Starting from the root node, the threshold of each node feature is calculated based on the objective function criterion of minimizing the square error at each depth of the tree. The objective function is:

[0032]

[0033] Among them, f is the objective function, y i is the label value of the current data, y p is the target temperature mean corresponding to the total data of the current interval divided by the threshold, which is used as the predicted value of this interval; Min means taking the minimum value, x i ∈R n Indicates the categories of all feature points in the current stage.

[0034] Preferably, the method of predicting the target temperature matched by the operating condition dimension data when the air conditioner is turned on based on the temperature prediction model includes:

[0035] Input the target features of the operating condition dimension data when the air conditioner is turned on into the temperature prediction model to obtain the predicted value;

[0036] In cooling mode, determine whether the predicted value is greater than the current set temperature value; if yes, use the predicted value as the target temperature value; if no, use the set temperature value as the target temperature value;

[0037] In the heating mode, it is determined whether the predicted value is less than the current set temperature value; if yes, the predicted value is used as the target temperature value; if no, the set temperature value is used as the target temperature value.

[0038] A second aspect of the present invention provides an air conditioning target temperature prediction system based on historical operating condition data, comprising:

[0039] Data screening module: screening raw data of various dimensions related to air conditioning temperature changes;

[0040] Data processing module: processes the raw data of each dimension to match the operating condition dimension data and the temperature value each time the air conditioner is turned on, and filters out abnormal values; the temperature value is the temperature that the air conditioner can reach;

[0041] Temperature prediction module: perform correlation analysis on the data of each working condition dimension after data processing to obtain the correlation coefficient, and select the target feature from high to low based on the correlation coefficient; train the target feature through the decision tree regression algorithm to obtain the temperature prediction model, and predict the target temperature matched by the working condition dimension data when the air conditioner is turned on based on the temperature prediction model.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] 1. The present invention screens original data of various dimensions related to the temperature change of the air conditioner; processes the original data of various dimensions to match the operating condition dimension data and the temperature value each time the air conditioner is turned on, and filters out abnormal values; performs correlation analysis on the processed operating condition dimension data to obtain correlation coefficients, and selects target features from high to low based on the correlation coefficients; trains the target features through a decision tree regression algorithm to obtain a temperature prediction model, and predicts the target temperature matched by the operating condition dimension data when the air conditioner is turned on based on the temperature prediction model, thereby solving the technical problem that it is difficult to predict the target temperature in different scenarios according to the environmental conditions when the air conditioner is running, and reducing the energy consumption of the air conditioner when the actual achievable temperature deviates from the set temperature value; the present invention can reduce the long-term high-load operation of air-conditioning components and promote the stability and energy saving of air-conditioning operation.

[0044] 2. The present invention can make full use of the historical operation data of different air conditioners, accurately predict the temperature values ​​that networked air conditioners can reach under different working conditions, and achieve temperature control that matches the performance of air conditioners. By reducing the long-term high-load operation of air conditioner components, the fatigue damage of the air conditioner system is effectively reduced, thereby improving the reliability and stability of the system. This preventive maintenance strategy not only reduces maintenance costs, but also greatly extends the service life of air conditioner equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 It is a schematic diagram of the operation steps of the present invention;

[0047] Figure 2 A schematic diagram of the operation steps of filtering outliers in the present invention;

[0048] Figure 3 It is a schematic diagram of the correlation analysis matrix of the present invention. DETAILED DESCRIPTION

[0049] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] See also Figure 1-Figure 2 The first embodiment of the present invention provides an air conditioning target temperature prediction method based on historical operating condition data, comprising:

[0051] S1: Filter the original data of various dimensions related to air conditioning temperature changes;

[0052] S2: Process the original data of each dimension to match the operating condition dimension data and the temperature value each time the air conditioner is turned on, and filter out abnormal values; the temperature value is the temperature that the air conditioner can reach;

[0053] S3: Perform correlation analysis on the data of each working condition dimension after data processing to obtain the correlation coefficient, and select the target feature from high to low based on the correlation coefficient; train the target feature through the decision tree regression algorithm to obtain a temperature prediction model, and predict the target temperature matched by the working condition dimension data when the air conditioner is turned on based on the temperature prediction model.

[0054] In this application, the original data of various dimensions related to the change of air conditioning temperature are screened, including:

[0055] The outdoor ambient temperature, indoor temperature, indoor humidity, and various setting parameters and component status values ​​of the air conditioner space are collected in real time through sensors; the dimensional raw data includes outdoor ambient temperature, indoor temperature, indoor humidity, setting parameters and component status values;

[0056] The original dimensional data of each device is uploaded to the cloud, and the original dimensional data that are preliminarily screened out and strongly related to the cooling and heating capabilities are marked as operating condition data dimensions based on expert experience.

[0057] It should be noted that the setting parameters include setting temperature and setting windshield, etc.

[0058] It should be noted that, in the process of screening data, the present invention obtains historical data recently reported by the target networked air-conditioning equipment, preliminarily screens out data dimensions that are strongly related to cooling and heating capabilities, and can provide data support for subsequent analysis.

[0059] In this application, the original data of each dimension is processed to match the working condition dimension data and the temperature value each time the air conditioner is turned on, including:

[0060] Mark the data between each startup and shutdown of the air conditioner in the historical data as reference data, obtain the lowest indoor temperature value that can be achieved in the cooling mode or the highest indoor temperature value that can be achieved in the heating mode from several reference data, mark the lowest indoor temperature value or the highest indoor temperature value as the temperature value in the corresponding reference data, and match the operating condition dimension data at each startup moment with the corresponding temperature value.

[0061] In this application, outliers are filtered, including:

[0062] A1: prioritize the corresponding matched working condition dimension data in the following order: set temperature, set windshield, and outdoor temperature;

[0063] A2: Circularly judge each working condition dimension data and the other working condition dimension data, mark the working condition dimension data being judged as the target data, and mark the working condition dimension data in the other working condition dimension data being judged as the data to be judged; judge whether the set temperature of the target data and the data to be judged is the same; if yes, jump to A3; if no, do nothing;

[0064] A3: In cooling mode, when the wind speed of the data to be judged is not less than the wind speed of the target data, the outdoor temperature is not greater than the outdoor temperature of the target data, and the temperature value is greater than the temperature value of the target data; or in heating mode, when the wind speed of the data to be judged is not less than the wind speed of the target data, the outdoor temperature is not less than the outdoor temperature of the target data, and the temperature value is less than the temperature value of the target data, the data to be judged is regarded as abnormal data, and the temperature value of the abnormal data is corrected to the temperature value of the current cycle data.

[0065] It is worth noting that the present invention can improve the rationality of the overall data by analyzing the abnormal situation of the current data to be judged by judging the wind speed, outdoor temperature and temperature value of the data to be judged.

[0066] In this application, the correlation coefficients are obtained by performing correlation analysis on the data of each working condition dimension after data processing, including:

[0067] Add two features, day and night conditions and the difference between outdoor temperature and set temperature, to the data of each working condition dimension;

[0068] The Pearson correlation coefficient calculation method is used to calculate the characteristic correlation between the dimension data of each working condition. The Pearson correlation coefficient calculation method is as follows:

[0069]

[0070] Among them, X and Y are the two types of operating condition characteristics to be calculated, Cov is the calculated covariance, σ is the standard deviation of the operating condition variable, and ρ is the correlation coefficient.

[0071] It should be noted that adding the two characteristics of day and night conditions and the difference between outdoor temperature and set temperature to the data of each operating condition dimension can improve the analyzable dimensions and increase the accuracy of the analysis.

[0072] In this application, target features are selected based on the correlation coefficient from high to low, including:

[0073] The correlation coefficients calculated based on the data of each operating condition dimension are arranged in a matrix form, the values ​​in the last column of the matrix are marked as the positive and negative correlation coefficients between the temperature value and each operating condition feature, and the three feature dimensions with the largest absolute values ​​of the positive and negative correlation coefficients are marked as target features.

[0074] It should be noted that if Figure 3 As shown in the figure, the correlation coefficients calculated based on the data of each working condition dimension are arranged in a matrix form. The last column is the positive and negative correlation coefficient between the temperature value and the data of each working condition dimension, which represents the degree of influence on the heating capacity of the air conditioner. The three characteristic dimensions with the largest absolute values ​​of positive and negative correlation are: set temperature, outdoor temperature and set windshield.

[0075] In this application, the target features are trained by decision tree regression algorithm to obtain a temperature prediction model, including:

[0076] B1: A data set is independently established for a single device, with the target features as the sample features of the data set, the sample labels as the corresponding temperature values, and 80% of the data set is randomly sampled as the training set and 20% as the validation set;

[0077] B2: Establish training criteria, and use the training criteria to train the working condition characteristics of the data set to obtain a temperature prediction model; perform input verification and evaluation on the verification set under the temperature prediction model generated by learning the training set, and determine whether the average absolute error between the predicted target temperature value and the temperature value corresponding to the label is not greater than the minimized error; if yes, jump to B3; if no, continue to train and verify the temperature prediction model; the minimized error is obtained through experience;

[0078] B3: Perform grid search optimization training on the temperature prediction model and set the optimized model parameters; the target optimized model parameters include the maximum depth of the tree and the minimum number of samples of leaf nodes.

[0079] Training guidelines are established in this application, including:

[0080] The regression tree algorithm model is used for training. The initial maximum depth parameter of the regression tree is set to 8. Starting from the root node, the threshold of each node feature is calculated based on the objective function criterion of minimizing the square error at each depth of the tree. The objective function is:

[0081]

[0082] Among them, f is the objective function, y i is the label value of the current data, y p is the target temperature mean corresponding to the total data of the current interval divided by the threshold, which is used as the predicted value of this interval; Min means taking the minimum value, x i ∈R n Indicates the categories of all feature points in the current stage.

[0083] In this application, the target temperature matched by the operating condition dimension data when the air conditioner is turned on is predicted based on the temperature prediction model, including:

[0084] Input the target features of the operating condition dimension data when the air conditioner is turned on into the temperature prediction model to obtain the predicted value;

[0085] In cooling mode, determine whether the predicted value is greater than the current set temperature value; if yes, use the predicted value as the target temperature value; if no, use the set temperature value as the target temperature value;

[0086] In the heating mode, it is determined whether the predicted value is less than the current set temperature value; if yes, the predicted value is used as the target temperature value; if no, the set temperature value is used as the target temperature value.

[0087] A second aspect of the present invention provides an air conditioning target temperature prediction system based on historical operating condition data, comprising:

[0088] Data screening module: screening raw data of various dimensions related to air conditioning temperature changes;

[0089] Data processing module: processes the raw data of each dimension to match the operating condition dimension data and the temperature value each time the air conditioner is turned on, and filters out abnormal values; the temperature value is the temperature that the air conditioner can reach;

[0090] Temperature prediction module: perform correlation analysis on the data of each working condition dimension after data processing to obtain the correlation coefficient, and select the target feature from high to low based on the correlation coefficient; train the target feature through the decision tree regression algorithm to obtain the temperature prediction model, and predict the target temperature matched by the working condition dimension data when the air conditioner is turned on based on the temperature prediction model.

[0091] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0092] Working principle of the present invention:

[0093] The present invention first screens the original data of each dimension related to the temperature change of the air conditioner; processes the original data of each dimension to match the operating condition dimension data and the temperature value each time the air conditioner is turned on, and filters the abnormal values; then performs correlation analysis on the processed operating condition dimension data to obtain the correlation coefficient, and selects the target feature from high to low based on the correlation coefficient; finally, trains the target feature through a decision tree regression algorithm to obtain a temperature prediction model, and predicts the target temperature matched by the operating condition dimension data when the air conditioner is turned on based on the temperature prediction model.

[0094] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A method for predicting target temperature of air conditioner based on historical operating condition data, characterized in that: S1: Filter the original data of various dimensions related to air conditioning temperature changes; S2: Process the original data of each dimension to match the operating condition dimension data and the temperature value each time the air conditioner is turned on, and filter out abnormal values; the temperature value is the temperature that the air conditioner can reach; S3: Perform correlation analysis on the data of each working condition dimension after data processing to obtain the correlation coefficient, and select the target feature from high to low based on the correlation coefficient; train the target feature through the decision tree regression algorithm to obtain a temperature prediction model, and predict the target temperature matched by the working condition dimension data when the air conditioner is turned on based on the temperature prediction model.

2. The method for predicting air conditioning target temperature based on historical operating condition data according to claim 1, characterized in that: The screening of raw data of various dimensions related to the change of air-conditioning temperature includes: The outdoor ambient temperature, indoor temperature, indoor humidity, and various setting parameters and component status values ​​of the air conditioner space are collected in real time through sensors; the dimensional raw data includes outdoor ambient temperature, indoor temperature, indoor humidity, setting parameters and component status values; The original dimensional data of each device is uploaded to the cloud, and the original dimensional data that are preliminarily screened out and strongly related to the cooling and heating capabilities are marked as operating condition data dimensions based on expert experience.

3. The method for predicting air conditioning target temperature based on historical operating condition data according to claim 1, characterized in that: The processing of the original data of each dimension to match the working condition dimension data and the temperature value each time the air conditioner is turned on includes: Mark the data between each startup and shutdown of the air conditioner in the historical data as reference data, obtain the lowest indoor temperature value that can be achieved in the cooling mode or the highest indoor temperature value that can be achieved in the heating mode from several reference data, mark the lowest indoor temperature value or the highest indoor temperature value as the temperature value in the corresponding reference data, and match the operating condition dimension data at each startup moment with the corresponding temperature value.

4. The method for predicting air conditioning target temperature based on historical operating condition data according to claim 1, characterized in that: The filtering of outliers includes: A1: prioritize the corresponding matched working condition dimension data in the following order: set temperature, set windshield, and outdoor temperature; A2: Circularly judge each working condition dimension data and the other working condition dimension data, mark the working condition dimension data being judged as the target data, and mark the working condition dimension data in the other working condition dimension data being judged as the data to be judged; judge whether the set temperature of the target data and the data to be judged is the same; if yes, jump to A3; if no, do nothing; A3: In cooling mode, when the wind speed of the data to be judged is not less than the wind speed of the target data, the outdoor temperature is not greater than the outdoor temperature of the target data, and the temperature value is greater than the temperature value of the target data; or in heating mode, when the wind speed of the data to be judged is not less than the wind speed of the target data, the outdoor temperature is not less than the outdoor temperature of the target data, and the temperature value is less than the temperature value of the target data, the data to be judged is regarded as abnormal data, and the temperature value of the abnormal data is corrected to the temperature value of the current cycle data.

5. The method for predicting target temperature of air conditioner based on historical operating condition data according to claim 1, characterized in that: The correlation coefficient is obtained by performing correlation analysis on the data of each working condition dimension after data processing, including: Add two features, day and night conditions and the difference between outdoor temperature and set temperature, to the data of each working condition dimension; The Pearson correlation coefficient calculation method is used to calculate the characteristic correlation between the dimension data of each working condition. The Pearson correlation coefficient calculation method is as follows: Among them, X and Y are the two types of operating condition characteristics to be calculated, Cov is the calculated covariance, σ is the standard deviation of the operating condition variable, and ρ is the correlation coefficient.

6. The method for predicting air conditioning target temperature based on historical operating condition data according to claim 1, characterized in that: The target features are selected from high to low based on the correlation coefficient, including: The correlation coefficients calculated based on the data of each operating condition dimension are arranged in a matrix form, the values ​​in the last column of the matrix are marked as the positive and negative correlation coefficients between the temperature value and each operating condition feature, and the three feature dimensions with the largest absolute values ​​of the positive and negative correlation coefficients are marked as target features.

7. The method for predicting air conditioning target temperature based on historical operating condition data according to claim 1, characterized in that: The temperature prediction model is obtained by training the target features through a decision tree regression algorithm, including: B1: A data set is independently established for a single device, with the target features as the sample features of the data set, the sample labels as the corresponding temperature values, and 80% of the data set is randomly sampled as the training set and 20% as the validation set; B2: Establish training criteria, and use the training criteria to train the working condition characteristics of the data set to obtain a temperature prediction model; perform input verification evaluation on the validation set under the temperature prediction model generated by learning the training set to determine whether the average absolute error between the predicted target temperature value and the temperature value corresponding to the label is not greater than the minimized error; if yes, jump to B3; if no, continue training and verification evaluation of the temperature prediction model; B3: Perform grid search optimization training on the temperature prediction model and set the optimized model parameters; the target optimized model parameters include the maximum depth of the tree and the minimum number of samples of leaf nodes.

8. The method for predicting target temperature of air conditioner based on historical operating condition data according to claim 7, characterized in that: The establishment of training criteria includes: The regression tree algorithm model is used for training. The initial maximum depth parameter of the regression tree is set to 8. Starting from the root node, the threshold of each node feature is calculated based on the objective function criterion of minimizing the square error at each depth of the tree. The objective function is: Among them, f is the objective function, y i is the label value of the current data, y p is the target temperature mean corresponding to the total data of the current interval divided by the threshold, which is used as the predicted value of this interval; Min means taking the minimum value, x i ∈R n Indicates the categories of all feature points in the current stage.

9. The method for predicting air conditioning target temperature based on historical operating condition data according to claim 1, characterized in that: The method of predicting the target temperature matched by the operating condition dimension data when the air conditioner is turned on based on the temperature prediction model includes: Input the target features of the operating condition dimension data when the air conditioner is turned on into the temperature prediction model to obtain the predicted value; In cooling mode, determine whether the predicted value is greater than the current set temperature value; if yes, use the predicted value as the target temperature value; if no, use the set temperature value as the target temperature value; In the heating mode, it is determined whether the predicted value is less than the current set temperature value; if yes, the predicted value is used as the target temperature value; if no, the set temperature value is used as the target temperature value.

10. An air conditioning target temperature prediction system based on historical operating condition data, based on the air conditioning target temperature prediction method based on historical operating condition data according to claims 1 to 9, characterized in that: include: Data screening module: screening raw data of various dimensions related to air conditioning temperature changes; Data processing module: processes the raw data of each dimension to match the operating condition dimension data and the temperature value each time the air conditioner is turned on, and filters out abnormal values; the temperature value is the temperature that the air conditioner can reach; Temperature prediction module: perform correlation analysis on the data of each working condition dimension after data processing to obtain the correlation coefficient, and select the target feature from high to low based on the correlation coefficient; train the target feature through the decision tree regression algorithm to obtain the temperature prediction model, and predict the target temperature matched by the working condition dimension data when the air conditioner is turned on based on the temperature prediction model.

Images