Vehicle air conditioner dynamic performance analysis method and system based on machine learning

Abstract

The invention relates to the field of data analysis, in particular to an automobile air conditioner dynamic performance analysis method and system based on machine learning. The system comprises an automobile air conditioner power influence factor data acquisition module, a final random forest model construction module, a parameter tuning module, a first automobile air conditioner power performance grade judgment module and a second automobile air conditioner power performance grade judgment module. Training a random forest model through the collected automobile air conditioner power influence factor data set, inputting the to-be-detected automobile air conditioner power influence factor data into the final random forest model for classification to obtain a first automobile air conditioner power performance grade, and calculating the cooling speed of the to-be-detected automobile air conditioner; the second automobile air conditioner dynamic performance grade is obtained through analysis, the final to-be-detected automobile air conditioner dynamic performance grade is judged, automobile air conditioner dynamic performance analysis is achieved, and the accuracy and reliability of the analysis result are guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of data analysis, and in particular to a method and system for analyzing the dynamic performance of an automobile air conditioner based on machine learning. Background Art

[0002] With the rapid development of the national economy, people's demand for automobiles is constantly increasing, and at the same time, the requirements for the riding environment are getting higher and higher. Therefore, the various performance of automobiles are gradually being valued by manufacturers.

[0003] Car air conditioners are very different from household air conditioners. Due to the structure of the car body, the requirements for the power performance of car air conditioners are often higher. At the same time, it is also necessary to analyze the power performance of car air conditioners frequently to avoid giving users a bad experience.

[0004] In the prior art, various data of the automobile air conditioner are often collected and directly analyzed to obtain the power performance level of the automobile air conditioner; or weights are set for various data of the automobile air conditioner, and the mean is calculated for analysis to achieve power performance analysis of the automobile air conditioner; it is difficult to avoid errors in the data collection process, and the effect in the actual working state is not considered. Summary of the invention

[0005] 1. Technical issues to be resolved

[0006] In order to solve the shortcomings of the background technology that manual operation is prone to errors and does not take comprehensive consideration into account the actual working conditions, the present invention proposes a method and system for analyzing the dynamic performance of an automobile air conditioner based on machine learning.

[0007] (II) Technical solution

[0008] A method for analyzing the dynamic performance of an automobile air conditioner based on machine learning, the method comprising the following steps:

[0009] S1. Set a category set of factors affecting automobile air conditioning power, collect automobile air conditioning samples, and establish an automobile air conditioning sample set; collect automobile air conditioning power affecting factor data according to the category set of factors affecting automobile air conditioning power and the automobile air conditioning sample set, and obtain an automobile air conditioning power affecting factor data matrix;

[0010] S2, setting a corresponding automobile air-conditioning power performance level set according to the automobile air-conditioning sample set, and setting a corresponding automobile air-conditioning cooling speed interval set according to the automobile air-conditioning power performance level set;

[0011] S3, constructing a final random forest model according to the data matrix of the factors affecting the power of the automobile air conditioner;

[0012] S4. Optimize the parameters of the initial depth threshold and the number of initial decision trees in the random forest model through an intelligent optimization algorithm;

[0013] S5, collecting data on automobile air-conditioning power influencing factors of the automobile air-conditioning to be tested, obtaining a data set of automobile air-conditioning power influencing factors to be tested, and determining the automobile air-conditioning power performance level of the automobile air-conditioning to be tested according to the final random forest model, to obtain a first automobile air-conditioning power performance level;

[0014] S6. Set a unit period, collect cooling data within the unit period of the automobile air conditioner to be tested, calculate the cooling speed of the automobile air conditioner to be tested, and analyze it in combination with the automobile air conditioner cooling speed interval set to obtain the second automobile air conditioner power performance level.

[0015] According to the automobile air-conditioning power influencing factor category set and the automobile air-conditioning sample set, the automobile air-conditioning power influencing factor data is collected to obtain the automobile air-conditioning power influencing factor data matrix, so as to ensure the accuracy of the acquired data and provide a data basis for subsequent operations; the automobile air-conditioning power performance level set and the automobile air-conditioning cooling speed interval set are set to provide a data basis for subsequent operations; the constructed random forest model is trained and tested based on the automobile air-conditioning power influencing factor data matrix to ensure the reliability of the final random forest model; the first automobile air-conditioning power performance level is obtained by inputting the collected automobile air-conditioning power influencing factor data set to be tested into the final random forest model for classification, so as to ensure the accuracy of the classification result; the second automobile air-conditioning power performance level is obtained by calculating the cooling speed of the automobile air-conditioning to be tested and analyzing it in combination with the automobile air-conditioning cooling speed interval set, so as to ensure the reliability of the analysis result.

[0016] Preferably, the specific steps of setting a category set of automobile air-conditioning power influencing factors, collecting automobile air-conditioning samples, and establishing an automobile air-conditioning sample set; collecting automobile air-conditioning power influencing factor data according to the category set of automobile air-conditioning power influencing factors and the automobile air-conditioning sample set, and obtaining an automobile air-conditioning power influencing factor data matrix are as follows:

[0017] S11, set the category set of factors affecting the power of automobile air conditioning A = {a 1 ,a 2 ,…,a i ,…,a k}, where a i represents the i-th category of factors affecting the power of automobile air conditioner, and k represents the total number of categories of factors affecting the power of automobile air conditioner;

[0018] S12, collect automobile air conditioning samples, and establish automobile air conditioning sample set B = {b 1 ,b 2 ,…,b i ,…,b l}, where b i represents the i-th automobile air-conditioning sample, l represents the total number of automobile air-conditioning samples;

[0019] S13, collecting the automobile air conditioning sample set B by data collection equipment = {b 1 ,b 2 ,…,b i ,…,b l The automobile air conditioning power influencing factor data of all automobile air conditioning power influencing factor categories of each automobile air conditioning sample in} are obtained, and the automobile air conditioning power influencing factor data matrix is ​​as follows:

[0020]

[0021] Among them, c ij Represents the automobile air-conditioning power influencing factor data of the i-th automobile air-conditioning power influencing factor category of the j-th automobile air-conditioning sample.

[0022] The data of automobile air-conditioning power influencing factors are collected according to the automobile air-conditioning power influencing factor category set and the automobile air-conditioning sample set to obtain the automobile air-conditioning power influencing factor data matrix to ensure the accuracy of the acquired data and provide a data basis for subsequent operations.

[0023] Preferably, the specific steps of setting a corresponding automobile air-conditioning power performance level set according to the automobile air-conditioning sample set and setting a corresponding automobile air-conditioning cooling speed set according to the automobile air-conditioning power performance level set are as follows:

[0024] S21, according to the automobile air conditioning sample set B = {b 1 ,b 2 ,…,b i ,…,b l} Set the corresponding automobile air conditioning power performance level set D = {d 1 ,d 2 ,…,d i ,…,d l}, where d i represents the automobile air-conditioning power performance level of the i-th automobile air-conditioning sample;

[0025] S22, according to the automobile air conditioning power performance level set D = {d 1 ,d 2 ,…,d i ,…,d l} Set the corresponding car air conditioning cooling speed interval set in, Indicates the cooling speed range corresponding to the i-th automobile air-conditioning power performance level.

[0026] Set the automobile air-conditioning power performance level set and the automobile air-conditioning cooling speed range set to provide a data basis for subsequent operations.

[0027] Preferably, the specific steps of constructing the final random forest model according to the data matrix of the automobile air-conditioning power influencing factors are as follows:

[0028] S31, setting the training data ratio to α and the test data ratio to β, dividing the data matrix of the automobile air-conditioning power influencing factors according to the training data ratio and the test data ratio, and obtaining a training matrix of the automobile air-conditioning power influencing factors data and a test matrix of the automobile air-conditioning power influencing factors data respectively;

[0029] S32, setting the initial depth threshold of the random forest model to L and the initial number of decision trees to N, and training the random forest model based on the training matrix of the data of factors affecting the power of the automobile air conditioner;

[0030] S321, using an unweighted sampling method to perform stratified sampling with replacement on the training matrix of the data of factors affecting the power of automobile air conditioners, and during the sampling process, randomly extracting n columns of sample data from the training matrix of the data of factors affecting the power of automobile air conditioners to obtain a training sample subset;

[0031] S322, constructing an initial decision tree, taking the sample data in the training sample subset as the root node of the initial decision tree, and recording it as the first layer of the initial decision tree;

[0032] S323, the root node of the initial decision tree is divided by the CART algorithm to generate child nodes, and the generated child nodes are used as root nodes for re-division until no new child nodes are generated, and the operation is terminated to obtain a trained decision tree. The division process of the root node is as follows:

[0033] S3231, setting a Gini coefficient threshold, and calculating the Gini coefficient of the sample data of the root node of the initial decision tree by using the CART algorithm;

[0034] If the Gini coefficient is greater than the set Gini coefficient threshold, continue to divide the root node;

[0035] If the Gini coefficient is less than or equal to the set Gini coefficient threshold, the operation ends;

[0036] S324, repeat the operations in S321, S322 and S323 N times to obtain N trained decision tree models, and combine the N trained decision tree models to form a random forest model;

[0037] S33. Set a test accuracy threshold, input the test matrix of automobile air-conditioning power influencing factors into the random forest model for classification test, calculate the accuracy of the classification result, and if the accuracy of the classification result is greater than the set test accuracy threshold, use the random forest model as the final random forest model; otherwise, perform parameter tuning on the initial depth threshold and the number of initial decision trees in the random forest model, and use the random forest model obtained after parameter tuning as the final random forest model.

[0038] The constructed random forest model is trained and tested based on the data matrix of factors affecting automobile air conditioning power to ensure the reliability of the final random forest model.

[0039] Preferably, the specific steps of performing parameter tuning on the initial depth threshold and the number of initial decision trees in the random forest model by using an intelligent optimization algorithm are as follows:

[0040] S41, construct a gray wolf population, set the size of the gray wolf population to M, the current number of iterations to t, and the maximum number of iterations to t max and the dimension of the optimization search space of the gray wolf population is two-dimensional;

[0041] S42, randomly generate the initial position set G of the gray wolf population according to the initial depth threshold and the initial number of decision trees in the random forest model = {g 1 ,g 2 ,…,g i ,…,g M}, where g i represents the initial position of the i-th gray wolf in the gray wolf population;

[0042] S43, calculate the fitness function value of each gray wolf individual in the gray wolf population, and use the three gray wolf individuals with the highest fitness function values ​​as the current optimal solution, optimal solution and suboptimal solution, and record the positions of the three gray wolf individuals at the same time; the fitness function formula is as follows:

[0043]

[0044] Among them, Fit(x i ) represents the fitness function value of the i-th gray wolf individual, x i represents the ability of the i-th gray wolf to improve the classification accuracy of the random forest model, and λ represents the correction value;

[0045] S44, start iterating, during each iteration, the three gray wolf individuals corresponding to the optimal solution, the good solution and the suboptimal solution will identify the location of the prey, after obtaining the location of the prey, the three gray wolf individuals corresponding to the optimal solution, the good solution and the suboptimal solution will lead the remaining gray wolf individuals to surround the prey, and update the locations of all gray wolf individuals in the gray wolf population;

[0046] Calculate the updated fitness function values ​​of all the gray wolf individuals in the gray wolf population, take the three gray wolf individuals with the highest fitness function values ​​as the current optimal solution, optimal solution, and suboptimal solution, and re-record the positions of the three gray wolf individuals;

[0047] S45, determine whether the current number of iterations t is less than the maximum number of iterations t max , if the current number of iterations t is less than the maximum number of iterations t max , the current number of iterations is t plus 1 and returns to S44; otherwise, the current optimal solution is taken as the global optimal solution, and the optimal depth threshold and the optimal number of decision trees corresponding to the global optimal solution are output.

[0048] The gray wolf optimization algorithm is used to optimize the parameters of the initial depth threshold and the number of initial decision trees in the random forest model, simulate the behavior of gray wolves tracking and surrounding prey under the leadership of the alpha wolf in the wolf pack, build a mathematical model, improve the speed and accuracy of the tuning process, and enhance the efficiency of the convergence process.

[0049] Preferably, the specific steps of collecting the data of the automobile air-conditioning power influencing factors of the automobile air-conditioning to be tested, obtaining the data set of the automobile air-conditioning power influencing factors to be tested, and determining the automobile air-conditioning power performance level of the automobile air-conditioning to be tested according to the final random forest model, and obtaining the first automobile air-conditioning power performance level are as follows:

[0050] S51, collecting the automobile air conditioning power influencing factor data of all automobile air conditioning power influencing factor categories in the automobile air conditioning to be tested, and obtaining the automobile air conditioning power influencing factor data set A'={a' 1 ,a' 2 ,…,a' i ,…,a' k}, where a' i The automobile air-conditioning power influencing factor data representing the automobile air-conditioning power influencing factor category of the automobile air-conditioning to be tested;

[0051] S52, inputting the data set of factors affecting the power of the automobile air conditioner to be tested into the final random forest model for classification, and obtaining the first automobile air conditioner power performance level.

[0052] By inputting the collected data set of influencing factors of the automobile air-conditioning power to be tested into the final random forest model for classification, the first automobile air-conditioning power performance level is obtained to ensure the accuracy of the classification results.

[0053] Preferably, the specific steps of setting a unit cycle, collecting cooling data within the unit cycle of the automobile air conditioner to be tested, calculating the cooling speed of the automobile air conditioner to be tested, and analyzing the cooling speed interval set of the automobile air conditioner to obtain the second automobile air conditioner power performance level are as follows:

[0054] S61, setting the unit period to T, collecting the cooling data Q within the unit period of the automobile air conditioner to be tested;

[0055] S62, calculate the cooling speed of the automobile air conditioner to be tested

[0056] S63, from the automobile air conditioning cooling speed range set Finding the automobile air conditioner cooling speed interval in which the cooling speed of the automobile air conditioner to be detected is located, and taking the automobile air conditioner power performance level corresponding to the automobile air conditioner cooling speed interval as the second automobile air conditioner power performance level;

[0057] S64: taking the lower automobile air-conditioning power performance level between the first automobile air-conditioning power performance level and the second automobile air-conditioning power performance level as the final automobile air-conditioning power performance level to be detected.

[0058] By calculating the cooling speed of the automobile air conditioner to be tested and analyzing it in combination with the automobile air conditioner cooling speed interval set, the second automobile air conditioner power performance level is obtained, thereby ensuring the reliability of the analysis result; the final automobile air conditioner power performance level of the automobile air conditioner to be tested is determined according to the first automobile air conditioner power performance level and the second automobile air conditioner power performance level, thereby ensuring the accuracy of the automobile air conditioner power performance analysis result.

[0059] The present invention also discloses a system for analyzing the power performance of an automobile air conditioner based on machine learning, comprising a data collection module for factors affecting the power of an automobile air conditioner, a final random forest model building module, a parameter tuning module, a first automobile air conditioner power performance level determination module, and a second automobile air conditioner power performance level determination module;

[0060] The automobile air-conditioning power influencing factor data collection module collects automobile air-conditioning power influencing factor data according to the automobile air-conditioning power influencing factor category set and the automobile air-conditioning sample set to obtain the automobile air-conditioning power influencing factor data matrix;

[0061] The final random forest model construction module trains and tests the constructed random forest model based on the automobile air conditioning power influencing factor data training matrix and the automobile air conditioning power influencing factor data test matrix obtained by data partitioning the automobile air conditioning power influencing factor data matrix to obtain the final random forest model;

[0062] The parameter tuning module performs parameter tuning on the initial depth threshold and the number of initial decision trees in the random forest model by using the gray wolf optimization algorithm;

[0063] The first automobile air-conditioning power performance level determination module obtains the first automobile air-conditioning power performance level by inputting the collected automobile air-conditioning power influencing factor data set to be detected into the final random forest model for classification;

[0064] The second automobile air-conditioning power performance level determination module calculates the cooling speed of the automobile air-conditioning to be detected and analyzes it in combination with the automobile air-conditioning cooling speed interval set to obtain the second automobile air-conditioning power performance level.

[0065] (III) Beneficial effects

[0066] The present invention provides a method for analyzing the dynamic performance of an automobile air conditioner based on machine learning, which has the following beneficial effects:

[0067] 1. The present invention provides an automobile air-conditioning power influencing factor data acquisition module, a final random forest model construction module, a parameter tuning module, a first automobile air-conditioning power performance level determination module, and a second automobile air-conditioning power performance level determination module; automobile air-conditioning power influencing factor data are collected according to the automobile air-conditioning power influencing factor category set and the automobile air-conditioning sample set to obtain an automobile air-conditioning power influencing factor data matrix to ensure the accuracy of the acquired data and provide a data basis for subsequent operations; the constructed random forest model is trained and tested based on the automobile air-conditioning power influencing factor data matrix to ensure the reliability of the final random forest model; the first automobile air-conditioning power performance level is obtained by inputting the collected automobile air-conditioning power influencing factor data set to be tested into the final random forest model for classification to ensure the accuracy of the classification result; the second automobile air-conditioning power performance level is obtained by calculating the cooling speed of the automobile air-conditioning to be tested and analyzing it in combination with the automobile air-conditioning cooling speed interval set to ensure the reliability of the analysis result.

[0068] 2. The parameter tuning of the initial depth threshold and the initial number of decision trees in the random forest model is achieved through the gray wolf optimization algorithm, which simulates the behavior of gray wolves tracking and surrounding prey under the leadership of the alpha wolf in the wolf pack, constructs a mathematical model, improves the speed and accuracy of the tuning process, and improves the efficiency of the convergence process. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying creative work.

[0070] Figure 1 A flowchart of a method for analyzing the power performance of an automobile air conditioner based on machine learning provided by the present invention;

[0071] Figure 2 A module schematic diagram of a system for analyzing the dynamic performance of an automobile air conditioner based on machine learning provided by the present invention. DETAILED DESCRIPTION

[0072] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not 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.

[0073] In the description of the present invention, it is necessary to understand that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention.

[0074] The first embodiment of the automobile air conditioning power performance analysis method based on machine learning is as follows:

[0075] See also Figure 1 , a method for analyzing the dynamic performance of an automobile air conditioner based on machine learning, comprising the following steps:

[0076] S1. Set a category set of factors affecting automobile air conditioning power, collect automobile air conditioning samples, and establish an automobile air conditioning sample set; collect automobile air conditioning power affecting factor data according to the category set of factors affecting automobile air conditioning power and the automobile air conditioning sample set, and obtain an automobile air conditioning power affecting factor data matrix;

[0077] S11, set the category set of factors affecting the power of automobile air conditioning A = {a 1 ,a 2 ,…,a i ,…,a k}, where a i represents the i-th category of factors affecting the power of automobile air conditioner, and k represents the total number of categories of factors affecting the power of automobile air conditioner;

[0078] S12, collect automobile air conditioning samples, and establish automobile air conditioning sample set B = {b 1 ,b 2 ,…,b i ,…,b l}, where b i represents the i-th automobile air-conditioning sample, l represents the total number of automobile air-conditioning samples;

[0079] S13, collecting the automobile air conditioning sample set B by data collection equipment = {b 1 ,b 2 ,…,b i ,…,b l The automobile air conditioning power influencing factor data of all automobile air conditioning power influencing factor categories of each automobile air conditioning sample in} are obtained, and the automobile air conditioning power influencing factor data matrix is ​​as follows:

[0080]

[0081] Among them, c ij Represents the automobile air-conditioning power influencing factor data of the i-th automobile air-conditioning power influencing factor category of the j-th automobile air-conditioning sample.

[0082] S2, setting a corresponding automobile air-conditioning power performance level set according to the automobile air-conditioning sample set, and setting a corresponding automobile air-conditioning cooling speed interval set according to the automobile air-conditioning power performance level set;

[0083] S21, according to the automobile air conditioning sample set B = {b 1 ,b 2 ,…,b i ,…,b l} Set the corresponding automobile air conditioning power performance level set D = {d 1 ,d 2 ,…,d i ,…,d l}, where d i represents the automobile air-conditioning power performance level of the i-th automobile air-conditioning sample;

[0084] S22, according to the automobile air conditioning power performance level set D = {d 1 ,d 2 ,…,d i ,…,d l} Set the corresponding car air conditioning cooling speed interval set in, Indicates the cooling speed range corresponding to the i-th automobile air-conditioning power performance level.

[0085] S3, constructing a final random forest model according to the data matrix of the factors affecting the power of the automobile air conditioner;

[0086] S31, setting the training data ratio to α and the test data ratio to β, dividing the data matrix of the automobile air-conditioning power influencing factors according to the training data ratio and the test data ratio, and obtaining a training matrix of the automobile air-conditioning power influencing factors data and a test matrix of the automobile air-conditioning power influencing factors data respectively;

[0087] S32, setting the initial depth threshold of the random forest model to L and the initial number of decision trees to N, and training the random forest model based on the training matrix of the data of factors affecting the power of the automobile air conditioner;

[0088] S321, using an unweighted sampling method to perform stratified sampling with replacement on the training matrix of the data of factors affecting the power of automobile air conditioners, and during the sampling process, randomly extracting n columns of sample data from the training matrix of the data of factors affecting the power of automobile air conditioners to obtain a training sample subset;

[0089] S322, constructing an initial decision tree, taking the sample data in the training sample subset as the root node of the initial decision tree, and recording it as the first layer of the initial decision tree;

[0090] S323, the root node of the initial decision tree is divided by the CART algorithm to generate child nodes, and the generated child nodes are used as root nodes for re-division until no new child nodes are generated, and the operation is terminated to obtain a trained decision tree. The division process of the root node is as follows:

[0091] S3231, setting a Gini coefficient threshold, and calculating the Gini coefficient of the sample data of the root node of the initial decision tree by using the CART algorithm;

[0092] If the Gini coefficient is greater than the set Gini coefficient threshold, continue to divide the root node;

[0093] If the Gini coefficient is less than or equal to the set Gini coefficient threshold, the operation ends;

[0094] S324, repeat the operations in S321, S322 and S323 N times to obtain N trained decision tree models, and combine the N trained decision tree models to form a random forest model;

[0095] S33. Set a test accuracy threshold, input the test matrix of automobile air-conditioning power influencing factors into the random forest model for classification test, calculate the accuracy of the classification result, and if the accuracy of the classification result is greater than the set test accuracy threshold, use the random forest model as the final random forest model; otherwise, perform parameter tuning on the initial depth threshold and the number of initial decision trees in the random forest model, and use the random forest model obtained after parameter tuning as the final random forest model.

[0096] S4. Optimize the parameters of the initial depth threshold and the number of initial decision trees in the random forest model through an intelligent optimization algorithm;

[0097] S41, construct a gray wolf population, set the size of the gray wolf population to M, set the current number of iterations to t, and the maximum number of iterations to t max and the dimension of the optimization search space of the gray wolf population is two-dimensional;

[0098] S42, randomly generate the initial position set G of the gray wolf population according to the initial depth threshold and the initial number of decision trees in the random forest model = {g 1 ,g 2 ,…,g i ,…,g M}, where g i represents the initial position of the i-th gray wolf in the gray wolf population;

[0099] S43, calculate the fitness function value of each gray wolf individual in the gray wolf population, and use the three gray wolf individuals with the highest fitness function values ​​as the current optimal solution, optimal solution and suboptimal solution, and record the positions of the three gray wolf individuals at the same time; the fitness function formula is as follows:

[0100]

[0101] Among them, Fit(x i ) represents the fitness function value of the i-th gray wolf individual, x i represents the ability of the i-th gray wolf to improve the classification accuracy of the random forest model, and λ represents the correction value;

[0102] S44, start iterating, during each iteration, the three gray wolf individuals corresponding to the optimal solution, the good solution and the suboptimal solution will identify the location of the prey, after obtaining the location of the prey, the three gray wolf individuals corresponding to the optimal solution, the good solution and the suboptimal solution will lead the remaining gray wolf individuals to surround the prey, and update the locations of all gray wolf individuals in the gray wolf population;

[0103] Calculate the updated fitness function values ​​of all the gray wolf individuals in the gray wolf population, take the three gray wolf individuals with the highest fitness function values ​​as the current optimal solution, optimal solution, and suboptimal solution, and re-record the positions of the three gray wolf individuals;

[0104] S45, determine whether the current number of iterations t is less than the maximum number of iterations t max , if the current number of iterations t is less than the maximum number of iterations t max , the current number of iterations is t plus 1 and returns to S44; otherwise, the current optimal solution is taken as the global optimal solution, and the optimal depth threshold and the optimal number of decision trees corresponding to the global optimal solution are output.

[0105] S5, collecting data on automobile air-conditioning power influencing factors of the automobile air-conditioning to be tested, obtaining a data set of automobile air-conditioning power influencing factors to be tested, and determining the automobile air-conditioning power performance level of the automobile air-conditioning to be tested according to the final random forest model, to obtain a first automobile air-conditioning power performance level;

[0106] S51, collecting the automobile air conditioning power influencing factor data of all automobile air conditioning power influencing factor categories in the automobile air conditioning to be tested, and obtaining the automobile air conditioning power influencing factor data set A'={a' 1 ,a' 2 ,…,a' i ,…,a' k}, where a' i The automobile air-conditioning power influencing factor data representing the automobile air-conditioning power influencing factor category of the automobile air-conditioning to be tested;

[0107] S52, inputting the data set of factors affecting the power of the automobile air conditioner to be tested into the final random forest model for classification, and obtaining the first automobile air conditioner power performance level.

[0108] S6, setting a unit period, collecting cooling data of the automobile air conditioner to be tested within the unit period, calculating the cooling speed of the automobile air conditioner to be tested, and analyzing it in combination with the automobile air conditioner cooling speed interval set to obtain the second automobile air conditioner power performance level;

[0109] S61, setting the unit period to T, collecting the cooling data Q within the unit period of the automobile air conditioner to be tested;

[0110] S62, calculate the cooling speed of the automobile air conditioner to be tested

[0111] S63, from the automobile air conditioning cooling speed range set Finding the automobile air conditioner cooling speed interval in which the cooling speed of the automobile air conditioner to be detected is located, and taking the automobile air conditioner power performance level corresponding to the automobile air conditioner cooling speed interval as the second automobile air conditioner power performance level;

[0112] S64: taking the lower automobile air-conditioning power performance level between the first automobile air-conditioning power performance level and the second automobile air-conditioning power performance level as the final automobile air-conditioning power performance level to be detected.

[0113] The second embodiment of the communication base station data interaction supervision method and system based on data interaction is as follows:

[0114] See also Figure 2 , a system for analyzing the power performance of an automobile air conditioner based on machine learning includes a data acquisition module for factors affecting the power of an automobile air conditioner, a final random forest model building module, a parameter tuning module, a first automobile air conditioner power performance level determination module, and a second automobile air conditioner power performance level determination module;

[0115] The automobile air-conditioning power influencing factor data collection module collects automobile air-conditioning power influencing factor data according to the automobile air-conditioning power influencing factor category set and the automobile air-conditioning sample set to obtain the automobile air-conditioning power influencing factor data matrix;

[0116] The final random forest model construction module trains and tests the constructed random forest model based on the automobile air conditioning power influencing factor data training matrix and the automobile air conditioning power influencing factor data test matrix obtained by data partitioning the automobile air conditioning power influencing factor data matrix to obtain the final random forest model;

[0117] The parameter tuning module performs parameter tuning on the initial depth threshold and the number of initial decision trees in the random forest model by using the gray wolf optimization algorithm;

[0118] The first automobile air-conditioning power performance level determination module obtains the first automobile air-conditioning power performance level by inputting the collected automobile air-conditioning power influencing factor data set to be detected into the final random forest model for classification;

[0119] The second automobile air-conditioning power performance level determination module calculates the cooling speed of the automobile air-conditioning to be detected and analyzes it in combination with the automobile air-conditioning cooling speed interval set to obtain the second automobile air-conditioning power performance level.

[0120] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0121] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can understand and use the invention well.

Claims

1. A method for analyzing the dynamic performance of automobile air conditioners based on machine learning, characterized in that: The steps include: S1. Set a category set of factors affecting automobile air conditioning power, collect automobile air conditioning samples, and establish an automobile air conditioning sample set; collect automobile air conditioning power affecting factor data according to the category set of factors affecting automobile air conditioning power and the automobile air conditioning sample set, and obtain an automobile air conditioning power affecting factor data matrix; S2, setting a corresponding automobile air-conditioning power performance level set according to the automobile air-conditioning sample set, and setting a corresponding automobile air-conditioning cooling speed interval set according to the automobile air-conditioning power performance level set; S3, constructing a final random forest model according to the data matrix of the factors affecting the power of the automobile air conditioner; S4. Optimize the parameters of the initial depth threshold and the number of initial decision trees in the random forest model through an intelligent optimization algorithm; S5, collecting data on automobile air-conditioning power influencing factors of the automobile air-conditioning to be tested, obtaining a data set of automobile air-conditioning power influencing factors to be tested, and determining the automobile air-conditioning power performance level of the automobile air-conditioning to be tested according to the final random forest model, to obtain a first automobile air-conditioning power performance level; S6. Set a unit period, collect cooling data within the unit period of the automobile air conditioner to be tested, calculate the cooling speed of the automobile air conditioner to be tested, and analyze it in combination with the automobile air conditioner cooling speed interval set to obtain the second automobile air conditioner power performance level.

2. The method for analyzing the dynamic performance of an automobile air conditioner based on machine learning according to claim 1, characterized in that: The S1 comprises the following steps: S11. Setting a category set of factors affecting automobile air conditioning power; S12, collect automobile air-conditioning samples and establish an automobile air-conditioning sample set; S13, collecting automobile air-conditioning power influencing factor data of all automobile air-conditioning power influencing factor categories of each automobile air-conditioning sample in the automobile air-conditioning sample set through a data collection device to obtain an automobile air-conditioning power influencing factor data matrix.

3. The method for analyzing the dynamic performance of an automobile air conditioner based on machine learning according to claim 2, characterized in that: The S2 comprises the following steps: S21. Setting a corresponding automobile air-conditioning power performance level set according to the automobile air-conditioning sample set; S22. Setting a corresponding automobile air conditioner cooling speed interval set according to the automobile air conditioner power performance level set.

4. The method for analyzing the dynamic performance of an automobile air conditioner based on machine learning according to claim 3, characterized in that: The S3 comprises the following steps: S31, setting the training data ratio to α and the test data ratio to β, dividing the data matrix of the automobile air-conditioning power influencing factors according to the training data ratio and the test data ratio, and obtaining a training matrix of the automobile air-conditioning power influencing factors data and a test matrix of the automobile air-conditioning power influencing factors data respectively; S32, setting the initial depth threshold of the random forest model to L and the initial number of decision trees to N, and training the random forest model based on the training matrix of the data of factors affecting the power of the automobile air conditioner; S33. Set a test accuracy threshold, input the test matrix of automobile air-conditioning power influencing factors into the random forest model for classification test, calculate the accuracy of the classification result, and if the accuracy of the classification result is greater than the set test accuracy threshold, use the random forest model as the final random forest model; otherwise, perform parameter tuning on the initial depth threshold and the number of initial decision trees in the random forest model, and use the random forest model obtained after parameter tuning as the final random forest model.

5. The method for analyzing the dynamic performance of an automobile air conditioner based on machine learning according to claim 4, characterized in that: The S32 comprises the following steps: S321, using an unweighted sampling method to perform stratified sampling with replacement on the training matrix of the data of factors affecting the power of automobile air conditioners, and during the sampling process, randomly extracting n columns of sample data from the training matrix of the data of factors affecting the power of automobile air conditioners to obtain a training sample subset; S322, constructing an initial decision tree, taking the sample data in the training sample subset as the root node of the initial decision tree, and recording it as the first layer of the initial decision tree; S323, dividing the root node of the initial decision tree by using the CART algorithm to generate child nodes, and then re-dividing the generated child nodes as the root nodes until no new child nodes are generated, and then terminating the operation to obtain a trained decision tree; S324, repeat the operations in S321, S322 and S323 N times to obtain N trained decision tree models, and combine the N trained decision tree models together to form a random forest model.

6. The method for analyzing the dynamic performance of an automobile air conditioner based on machine learning according to claim 5, characterized in that: The root node division process in S323 includes the following steps: S3231, setting a Gini coefficient threshold, and calculating the Gini coefficient of the sample data of the root node of the initial decision tree by using the CART algorithm; If the Gini coefficient is greater than the set Gini coefficient threshold, continue to divide the root node; If the Gini coefficient is less than or equal to the set Gini coefficient threshold, the operation is terminated.

7. The method for analyzing the power performance of an automobile air conditioner based on machine learning according to claim 6, characterized in that: The S4 comprises the following steps: S41, construct a gray wolf population, set the size of the gray wolf population to M, the current number of iterations to t, and the maximum number of iterations to t max and the dimension of the optimization search space of the gray wolf population is two-dimensional; S42, randomly generate the initial position set G of the gray wolf population according to the initial depth threshold and the initial number of decision trees in the random forest model = {g1, g2, ..., g i ,…,g M }, where g i represents the initial position of the i-th gray wolf in the gray wolf population; S43, calculate the fitness function value of each gray wolf individual in the gray wolf population, and use the three gray wolf individuals with the highest fitness function values ​​as the current optimal solution, optimal solution and suboptimal solution, and record the positions of the three gray wolf individuals. The fitness function formula is as follows: Among them, Fit(x i ) represents the fitness function value of the i-th gray wolf individual, x i represents the ability of the i-th gray wolf to improve the classification accuracy of the random forest model, and λ represents the correction value; S44, start iterating, during each iteration, the three gray wolf individuals corresponding to the optimal solution, the good solution and the suboptimal solution will identify the location of the prey, after obtaining the location of the prey, the three gray wolf individuals corresponding to the optimal solution, the good solution and the suboptimal solution will lead the remaining gray wolf individuals to surround the prey, and update the locations of all gray wolf individuals in the gray wolf population; Calculate the updated fitness function values ​​of all the gray wolf individuals in the gray wolf population, take the three gray wolf individuals with the highest fitness function values ​​as the current optimal solution, optimal solution, and suboptimal solution, and re-record the positions of the three gray wolf individuals; S45, determine whether the current number of iterations t is less than the maximum number of iterations t max , if the current number of iterations t is less than the maximum number of iterations t max , the current number of iterations is t plus 1 and returns to S44; otherwise, the current optimal solution is taken as the global optimal solution, and the optimal depth threshold and the optimal number of decision trees corresponding to the global optimal solution are output.

8. The method for analyzing the power performance of an automobile air conditioner based on machine learning according to claim 7, characterized in that: The S5 comprises the following steps: S51, collecting automobile air-conditioning power influencing factor data of all automobile air-conditioning power influencing factor categories of the automobile air-conditioning to be tested, and obtaining a data set of automobile air-conditioning power influencing factors to be tested; S52, inputting the data set of factors affecting the power of the automobile air conditioner to be tested into the final random forest model for classification, and obtaining the first automobile air conditioner power performance level.

9. The method for analyzing the power performance of an automobile air conditioner based on machine learning according to claim 8, characterized in that: The S6 comprises the following steps: S61, setting the unit period to T, collecting the cooling data Q within the unit period of the automobile air conditioner to be tested; S62, calculate the cooling speed of the automobile air conditioner to be tested S63, searching from the automobile air-conditioning cooling speed interval set for the automobile air-conditioning cooling speed interval in which the cooling speed of the automobile air-conditioning to be detected is located, and taking the automobile air-conditioning power performance level corresponding to the automobile air-conditioning cooling speed interval as the second automobile air-conditioning power performance level; S64: taking the lower automobile air-conditioning power performance level between the first automobile air-conditioning power performance level and the second automobile air-conditioning power performance level as the final automobile air-conditioning power performance level to be detected.

10. A system for implementing the automobile air-conditioning power performance analysis method based on machine learning as described in any one of claims 1 to 9, comprising an automobile air-conditioning power influencing factor data acquisition module, a final random forest model construction module, a parameter tuning module, a first automobile air-conditioning power performance level determination module, and a second automobile air-conditioning power performance level determination module; The automobile air-conditioning power influencing factor data collection module collects automobile air-conditioning power influencing factor data according to the automobile air-conditioning power influencing factor category set and the automobile air-conditioning sample set to obtain the automobile air-conditioning power influencing factor data matrix; The final random forest model construction module trains and tests the constructed random forest model based on the automobile air conditioning power influencing factor data training matrix and the automobile air conditioning power influencing factor data test matrix obtained by data partitioning the automobile air conditioning power influencing factor data matrix to obtain the final random forest model; The parameter tuning module performs parameter tuning on the initial depth threshold and the number of initial decision trees in the random forest model by using the gray wolf optimization algorithm; The first automobile air-conditioning power performance level determination module obtains the first automobile air-conditioning power performance level by inputting the collected automobile air-conditioning power influencing factor data set to be detected into the final random forest model for classification; The second automobile air-conditioning power performance level determination module calculates the cooling speed of the automobile air-conditioning to be detected and analyzes it in combination with the automobile air-conditioning cooling speed interval set to obtain the second automobile air-conditioning power performance level.

Images