A vehicle air conditioning intelligent management method and system

Abstract

The present invention discloses a vehicle air conditioning intelligent management method and system, the method includes data collection, data preprocessing, establishing a vehicle air conditioning comfort prediction model, model parameter optimization and vehicle air conditioning intelligent management. The present invention belongs to the field of vehicle air conditioning management technology, specifically refers to a vehicle air conditioning intelligent management method and system, this scheme adopts different optimization strategies for different types of samples by quantifying the credibility of the current prediction of the model, so as to avoid user discomfort caused by misjudgment in the intelligent management of the vehicle air conditioning; by introducing weight factors, the air conditioning system can operate more intelligently and energy-efficiently; by calculating individual fitness values ​​and updating the mutation probability mechanism, global search optimization is realized, so as to avoid the model from falling into the local optimum; based on the mutation probability, position update is realized, and in the dynamically changing in-vehicle environment, the uncertainty and noise data in the model training are effectively dealt with, so that the model is more reliable and the intelligent management of the vehicle air conditioning is more effective.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle air conditioning management, and in particular to a vehicle air conditioning intelligent management method and system. Background Art

[0002] The core of the intelligent management method of vehicle air conditioning is to analyze and predict the comfort needs of passengers in real time through machine learning, evolutionary algorithms and optimization technology, and dynamically adjust the parameters of the air conditioning system according to these prediction results. However, the general intelligent management method of vehicle air conditioning has the problem that the model misjudges the user's discomfort and cannot accurately predict the user's comfort feeling. The misjudgement of the user's comfort leads to a poor user experience of the vehicle air conditioning. The general intelligent management method of vehicle air conditioning has the problem that it cannot achieve global search optimization when optimizing the model parameters, resulting in poor intelligent management effect of the system. The low robustness of the model leads to unreliable models in the dynamically changing in-vehicle environment and poor intelligent management performance. Summary of the invention

[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a method and system for intelligent management of vehicle air conditioning. In view of the problem that the general intelligent management method of vehicle air conditioning has the problem that the model causes user discomfort due to misjudgment, cannot accurately predict the user's comfort feeling, and misjudgment of the user's comfort leads to poor user experience of the vehicle air conditioning, this solution quantifies the credibility of the vehicle air conditioning comfort prediction model for the current prediction, adopts different optimization strategies for different types of samples, and avoids user discomfort due to misjudgment in the intelligent management of vehicle air conditioning; by introducing weight factors, the model has different processing strategies when facing high-confidence and low-confidence samples, effectively reducing the risk of misjudgment, and enabling the air-conditioning system to operate more intelligently and energy-efficiently; the optimization strategy ensures its performance under different environmental conditions Wide applicability, ensuring stability and accuracy in the actual vehicle environment; the general vehicle air-conditioning intelligent management method has the problem that global search optimization cannot be achieved when optimizing model parameters, resulting in poor intelligent management effect of the system, and the low robustness of the model leads to unreliable model in the dynamically changing vehicle environment and poor intelligent management performance. This solution realizes global search optimization through individual fitness value calculation and mutation probability update mechanism, avoiding the model from falling into the local optimum, making the vehicle air-conditioning comfort prediction model closer to the global optimal solution, thereby improving the intelligent management effect of the system; position update is realized based on mutation probability, and in the dynamically changing vehicle environment, the uncertainty and noise data in model training are effectively dealt with, making the model more reliable and making the vehicle air-conditioning intelligent management more effective.

[0004] The technical solution adopted by the present invention is as follows: The present invention provides a vehicle air conditioning intelligent management method, the method comprising the following steps:

[0005] Step S1: data collection;

[0006] Step S2: data preprocessing;

[0007] Step S3: Establishing a vehicle air conditioning comfort prediction model;

[0008] Step S4: model parameter optimization;

[0009] Step S5: Intelligent management of vehicle air conditioning.

[0010] Furthermore, in step S1, the data collection is to collect historical vehicle air-conditioning operation data; the historical vehicle air-conditioning operation data includes in-vehicle sensor data, external environment data, vehicle operation data, vehicle air-conditioning setting data and comfort evaluation level; the comfort evaluation level is used as a data label.

[0011] Furthermore, in step S2, the data preprocessing is to perform data cleaning, data conversion, feature engineering and data set division on the collected data; the data cleaning is to remove missing values, outliers and duplicate data; the data conversion is to standardize and unify the data from different data sources; the feature engineering is to perform feature selection and feature construction; the data set division is to divide the collected original data set into a test set and a training set.

[0012] Furthermore, in step S3, the establishment of the vehicle air conditioning comfort prediction model specifically includes the following steps:

[0013] Step S31: model architecture design; the established vehicle air conditioning comfort prediction model includes: an input layer for receiving a preprocessed data set; a hidden layer for processing input features and extracting nonlinear relationships between features; an output layer for outputting a predicted comfort assessment level; the hidden layer includes: a first fully connected layer, including 64 neurons, using a ReLU activation function, and using L2 regularization; a second fully connected layer, including 128 neurons, using a ReLU activation function, and using L2 regularization; a third fully connected layer, including 64 neurons, using a ReLU activation function, and using Dropout; a fourth fully connected layer, including 32 neurons, using a ReLU activation function, and using Dropout; the model uses the Adam optimizer for parameter adjustment;

[0014] Step S32: define the classification confidence; calculate the definition classification confidence of each sample, the formula used is as follows:

[0015] ;

[0016] In the formula, is the classification confidence of sample s for the true category i2(s); α is the parameter vector of the model; It is the output value of sample s and true category i2(s) in the last layer of the model, i.e., layer M-1; is the maximum output value of other categories;

[0017] Step S33: normalization; determining a normalized scale factor R, normalizing the classification confidence based on the scale factor to obtain a normalized classification confidence; the formula used to calculate the scale factor is as follows:

[0018] ;

[0019] In the formula, is the parameter vector of the i1th layer;

[0020] Step S34: construct a basic loss function; the formula used is as follows:

[0021] ;

[0022] Where L(·) is the basic loss function; K is the total number of categories, and i2 is the category index; is the predicted output for the i2th category;

[0023] Step S35: introducing a weight factor; selecting a scale based on the classification confidence, and constructing a loss function of the vehicle air conditioning comfort prediction model. The formula used is as follows:

[0024] ;

[0025] Where LL(·) is the loss function of the vehicle air conditioning comfort prediction model; R 1 and R 2 is a weight factor used to weight samples in different confidence ranges; is the confidence threshold; and are the normalized predicted output values ​​for the i2th category and category i2(s) respectively; is the normalized classification confidence;

[0026] Step S36: Optimize the loss function: Optimize the dual-scale loss function based on a deterministic gradient descent algorithm, and the formula used is as follows:

[0027] ;

[0028] In the formula, , and They are the parameter vector and weight factor of the model during the n+1th training respectively; , and are the parameter vector and weight factor of the model during the nth training respectively; τ is the learning rate; μ is the momentum coefficient; is the gradient operator; TR is the training set; is the square of the L2 norm; is the regularization coefficient;

[0029] Step S37: model determination; a correctness threshold is set in advance. When the vehicle air-conditioning comfort prediction model converges with the loss of the training set, the vehicle air-conditioning comfort prediction model training is completed, otherwise the model parameters are optimized; when the prediction accuracy of the trained vehicle air-conditioning comfort prediction model for the test set is higher than the correctness threshold, the vehicle air-conditioning comfort prediction model is established, otherwise the data set training model is re-divided.

[0030] Furthermore, in step S4, the model parameter optimization specifically includes the following steps:

[0031] Step S41: initialization; constructing a parameter optimization space based on the learning rate, momentum coefficient, weight factor, regularization coefficient, initial weight and bias of the vehicle air conditioning comfort prediction model; initializing the position of the individual in the optimization population, and taking the prediction accuracy of the vehicle air conditioning comfort prediction model trained based on the individual position and the training set for the test set as the individual fitness value;

[0032] Step S42: Calculate the mutation probability; the formula used is as follows:

[0033] ;

[0034] In the formula, and are the mutation probabilities of the jth dimension of the ith individual at the t+1th iteration and the tth iteration, respectively; is the momentum factor; c 1 and c 2 They are local factors and global factors; and are the historical optimal position of the j-th dimension of the ith individual in the t-th iteration and the global optimal position of the j-th dimension; is the number of times the jth dimension of the i-th individual has not mutated continuously at the t-th iteration; is the mutation ratio of the jth dimension at the tth iteration; is the number of individuals mutated in dimension j at the tth iteration, and pn is the number of individuals in the population; is the position of the i-th individual in the j-th dimension at the t-th iteration;

[0035] Step S43: Update individual positions; update individual positions based on mutation probability; the formula used is as follows:

[0036] ;

[0037] In the formula, is the position of the i-th individual in the j-th dimension at the t+1-th iteration; is the individual fitness value; Indicates mutation; r is a random number between 0 and 1;

[0038] Step S44: optimization judgment; a fitness threshold is set in advance. When there is an individual fitness value higher than the fitness threshold, the vehicle air-conditioning comfort prediction model obtained based on the individual position is established as the vehicle air-conditioning comfort prediction model; if the maximum number of iterations is reached, return to step S41; otherwise, return to step S42.

[0039] Furthermore, in step S5, the vehicle air-conditioning intelligent management is based on the established vehicle air-conditioning comfort prediction model, and collects in-vehicle sensor data, external environment data, vehicle operation data and vehicle air-conditioning setting data to be evaluated in real time, and inputs them into the vehicle air-conditioning comfort prediction model after preprocessing. A comfort evaluation threshold is set in advance, and the vehicle air-conditioning setting data corresponding to the data whose comfort evaluation level predicted by the model is higher than the comfort evaluation threshold is used as the operation setting of the vehicle air-conditioning.

[0040] The present invention provides a vehicle air conditioning intelligent management system, comprising a data acquisition module, a data preprocessing module, a vehicle air conditioning comfort prediction model establishment module, a model parameter optimization module and a vehicle air conditioning intelligent management module;

[0041] The data acquisition module collects historical vehicle air conditioner operation data and sends the data to the data preprocessing module;

[0042] The data preprocessing module performs data cleaning, data conversion, feature engineering processing and data set division on the collected data, and sends the data to the vehicle air conditioning comfort prediction model establishment module;

[0043] The vehicle air conditioning comfort prediction model establishment module quantifies the credibility of the vehicle air conditioning comfort prediction model for the current prediction, adopts different optimization strategies for different types of samples, designs the loss function by introducing weight factors, establishes the vehicle air conditioning comfort prediction model based on the optimization strategy, and sends the data to the model parameter optimization module;

[0044] The model parameter optimization module realizes global search optimization through individual fitness value calculation and mutation probability update mechanism, realizes position update based on mutation probability, and then realizes model parameter optimization; and sends data to the vehicle air conditioning intelligent management module;

[0045] The vehicle air-conditioning intelligent management module realizes the vehicle air-conditioning intelligent management based on the established vehicle air-conditioning comfort prediction model.

[0046] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0047] (1) In view of the problem that the general vehicle air conditioning intelligent management method has the problem that the model causes user discomfort due to misjudgment and cannot accurately predict the user's comfort feeling. Misjudgment of the user's comfort leads to a poor user experience of the vehicle air conditioning. This scheme quantifies the credibility of the vehicle air conditioning comfort prediction model for the current prediction and adopts different optimization strategies for different types of samples to avoid user discomfort caused by misjudgment in the vehicle air conditioning intelligent management; by introducing weight factors, the model has different processing strategies when facing high-confidence and low-confidence samples, effectively reducing the risk of misjudgment and enabling the air conditioning system to operate more intelligently and energy-efficiently; the optimization strategy ensures its wide applicability under different environmental conditions and guarantees stability and accuracy in the actual vehicle environment.

[0048] (2) In view of the problem that general vehicle air-conditioning intelligent management methods cannot achieve global search optimization when optimizing model parameters, resulting in poor intelligent management effect of the system, low robustness of the model, which makes the model unreliable in the dynamically changing in-vehicle environment and poor intelligent management performance, this scheme realizes global search optimization through individual fitness value calculation and mutation probability update mechanism, avoiding the model from falling into the local optimum, making the vehicle air-conditioning comfort prediction model closer to the global optimal solution, thereby improving the intelligent management effect of the system; based on the mutation probability, the position update is realized, and the uncertainty and noise data in the model training are effectively dealt with in the dynamically changing in-vehicle environment, making the model more reliable and making the vehicle air-conditioning intelligent management more effective. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 A schematic diagram of a process flow of a vehicle air conditioning intelligent management method provided by the present invention;

[0050] Figure 2 A schematic diagram of a vehicle air-conditioning intelligent management system provided by the present invention;

[0051] Figure 3 is a schematic diagram of the process of step S3;

[0052] Figure 4 It is a schematic diagram of the process of step S4.

[0053] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0054] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0055] In the description of the present invention, it is necessary to understand that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred system or element must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as a limitation on the present invention.

[0056] Example 1, see Figure 1 The present invention provides a vehicle air conditioning intelligent management method, the method comprising the following steps:

[0057] Step S1: data collection, collecting historical vehicle air conditioning operation data;

[0058] Step S2: Data preprocessing, data cleaning, data conversion, feature engineering and data set division of the collected data;

[0059] Step S3: Establishing a vehicle air conditioning comfort prediction model, by quantifying the credibility of the vehicle air conditioning comfort prediction model for the current prediction, adopting different optimization strategies for different types of samples, and designing a loss function by introducing a weight factor, and establishing the vehicle air conditioning comfort prediction model based on the optimization strategy;

[0060] Step S4: Model parameter optimization, realizing global search optimization through individual fitness value calculation and mutation probability update mechanism, realizing position update based on mutation probability, and then realizing model parameter optimization;

[0061] Step S5: Intelligent management of the vehicle air conditioner, realizing intelligent management of the vehicle air conditioner based on the established vehicle air conditioner comfort prediction model.

[0062] Example 2, see Figure 1 This embodiment is based on the above embodiment. In step S1, the historical vehicle air-conditioning operation data includes in-vehicle sensor data, external environment data, vehicle operation data, vehicle air-conditioning setting data and comfort evaluation level; the comfort evaluation level includes level one comfort, level two comfort and level three comfort; the comfort evaluation level is used as a data label.

[0063] Example 3, see Figure 1This embodiment is based on the above embodiment. In step S2, data cleaning is to remove missing values, outliers, and duplicate data; data conversion is to standardize and unify data from different data sources; feature engineering processing is to perform feature selection and feature construction; and data set division is to divide the collected original data set into a test set and a training set.

[0064] Example 4, see Figure 1 and Figure 3 Based on the above embodiment, in step S3, establishing a vehicle air conditioning comfort prediction model specifically includes the following steps:

[0065] Step S31: model architecture design; the established vehicle air conditioning comfort prediction model includes: an input layer for receiving a preprocessed data set; a hidden layer for processing input features and extracting nonlinear relationships between features; an output layer for outputting a predicted comfort assessment level; the hidden layer includes: a first fully connected layer, including 64 neurons, using a ReLU activation function, and using L2 regularization; a second fully connected layer, including 128 neurons, using a ReLU activation function, and using L2 regularization; a third fully connected layer, including 64 neurons, using a ReLU activation function, and using Dropout; a fourth fully connected layer, including 32 neurons, using a ReLU activation function, and using Dropout; the model uses the Adam optimizer for parameter adjustment;

[0066] Step S32: define the classification confidence; calculate the definition classification confidence of each sample, the formula used is as follows:

[0067] ;

[0068] In the formula, is the classification confidence of sample s for the true category i2(s); α is the parameter vector of the model, ranging from -0.01 to 0.01, obtained by random initialization; It is the output value of the last layer of the model, i.e., the M-1 layer, where M is 6, for the sample s and the true category i2(s); is the maximum output value of other categories;

[0069] Step S33: normalization; determining a normalized scale factor R, normalizing the classification confidence based on the scale factor to obtain a normalized classification confidence; the formula used to calculate the scale factor is as follows:

[0070] ;

[0071] In the formula, is the parameter vector of the i1th layer;

[0072] Step S34: construct a basic loss function; the formula used is as follows:

[0073] ;

[0074] Where L(·) is the basic loss function; K is the total number of categories, which is 3 here, and i2 is the category index; is the predicted output for the i2th category;

[0075] Step S35: introducing a weight factor; selecting a scale based on the classification confidence, and constructing a loss function of the vehicle air conditioning comfort prediction model. The formula used is as follows:

[0076] ;

[0077] Where LL(·) is the loss function of the vehicle air conditioning comfort prediction model; R 1 and R 2 is a weight factor, ranging from 0 to 1, used to weight samples in different confidence ranges; is the confidence threshold, ranging from 0.5 to 0.7; and are the normalized predicted output values ​​for the i2th category and category i2(s) respectively; is the normalized classification confidence;

[0078] Step S36: Optimize the loss function: Optimize the dual-scale loss function based on a deterministic gradient descent algorithm, and the formula used is as follows:

[0079] ;

[0080] In the formula, , and They are the parameter vector and weight factor of the model during the n+1th training respectively; , and are the parameter vector and weight factor of the model during the nth training respectively; τ is the learning rate, and its value range is [0.0001, 0.001]; μ is the momentum coefficient, and its value here is 0.9; is the gradient operator; TR is the training set; is the square of the L2 norm; is the regularization coefficient, and its value range is [0.0001, 0.01];

[0081] Step S37: model determination; a correctness threshold is pre-set, which is set to 0.95 here. When the vehicle air-conditioning comfort prediction model converges to the loss of the training set, the vehicle air-conditioning comfort prediction model training is completed, otherwise the model parameters are optimized; when the prediction accuracy of the trained vehicle air-conditioning comfort prediction model for the test set is higher than the correctness threshold, the vehicle air-conditioning comfort prediction model is established, otherwise the data set training model is re-divided.

[0082] By performing the above operations, the general vehicle air-conditioning intelligent management method has the problem that the model causes user discomfort due to misjudgment, cannot accurately predict the user's comfort feeling, and misjudgment of the user's comfort leads to a poor user experience of the vehicle air-conditioning. This solution quantifies the credibility of the vehicle air-conditioning comfort prediction model for the current prediction, and adopts different optimization strategies for different types of samples to avoid user discomfort due to misjudgment in the vehicle air-conditioning intelligent management; by introducing weight factors, the model has different processing strategies when facing high-confidence and low-confidence samples, effectively reducing the risk of misjudgment, and enabling the air-conditioning system to operate more intelligently and energy-efficiently; the optimization strategy ensures its wide applicability under different environmental conditions and guarantees stability and accuracy in the actual vehicle environment.

[0083] Example 5, see Figure 1 and Figure 4 This embodiment is based on the above embodiment. In step S4, the model parameter optimization specifically includes the following steps:

[0084] Step S41: initialization; constructing a parameter optimization space based on the learning rate, momentum coefficient, weight factor, regularization coefficient, initial weight and bias of the vehicle air conditioning comfort prediction model; initializing the position of the individual in the optimization population, and taking the prediction accuracy of the vehicle air conditioning comfort prediction model trained based on the individual position and the training set for the test set as the individual fitness value;

[0085] Step S42: Calculate the mutation probability; the formula used is as follows:

[0086] ;

[0087] In the formula, and are the mutation probabilities of the jth dimension of the ith individual at the t+1th iteration and the tth iteration, respectively; is the momentum factor, ranging from 0.5 to 0.9; c 1 and c 2 They are local factor and global factor, both ranging from 1.5 to 2.0; and are the historical optimal position of the j-th dimension of the ith individual in the t-th iteration and the global optimal position of the j-th dimension; is the number of times the jth dimension of the i-th individual has not mutated continuously at the t-th iteration; is the mutation ratio of the jth dimension at the tth iteration; is the number of individuals mutated in dimension j at the tth iteration, pn is the number of individuals in the population, which is 50 here; is the position of the i-th individual at the j-th dimension at the t-th iteration;

[0088] Step S43: Update individual positions; update individual positions based on mutation probability; the formula used is as follows:

[0089] ;

[0090] In the formula, is the position of the i-th individual in the j-th dimension at the t+1-th iteration; is the individual fitness value; Indicates mutation; r is a random number between 0 and 1;

[0091] Step S44: optimization judgment; a fitness threshold is set in advance. When there is an individual fitness value higher than the fitness threshold, the vehicle air-conditioning comfort prediction model obtained based on the individual position is established as the completed vehicle air-conditioning comfort prediction model; if the maximum number of iterations is reached, which is set to 100 here, return to step S41; otherwise, return to step S42.

[0092] By performing the above operations, the general vehicle air-conditioning intelligent management method has the problem that it is impossible to achieve global search optimization when optimizing the model parameters, resulting in poor intelligent management effect of the system, low robustness of the model, resulting in unreliable model in the dynamically changing vehicle environment, and poor intelligent management performance. This scheme realizes global search optimization through individual fitness value calculation and mutation probability update mechanism, avoids the model from falling into the local optimum, makes the vehicle air-conditioning comfort prediction model closer to the global optimal solution, and thus improves the intelligent management effect of the system; realizes position update based on mutation probability, effectively copes with the uncertainty and noise data in model training in the dynamically changing vehicle environment, makes the model more reliable, and makes the vehicle air-conditioning intelligent management more effective.

[0093] Example 6, see Figure 1 This embodiment is based on the above embodiment. In step S5, the intelligent management of the vehicle air conditioner is based on the established vehicle air conditioner comfort prediction model, and the in-vehicle sensor data, external environment data, vehicle operation data and vehicle air conditioner setting data to be evaluated are collected in real time, and are input into the vehicle air conditioner comfort prediction model after preprocessing. A comfort evaluation threshold is set in advance, and the vehicle air conditioner setting data corresponding to the data whose comfort evaluation level predicted by the model is higher than the comfort evaluation threshold is used as the operation setting of the vehicle air conditioner.

[0094] Embodiment 7, see Figure 2 , this embodiment is based on the above embodiment, and the present invention provides a vehicle air conditioning intelligent management system, including a data acquisition module, a data preprocessing module, a vehicle air conditioning comfort prediction model establishment module, a model parameter optimization module and a vehicle air conditioning intelligent management module;

[0095] The data acquisition module collects historical vehicle air conditioner operation data and sends the data to the data preprocessing module;

[0096] The data preprocessing module performs data cleaning, data conversion, feature engineering processing and data set division on the collected data, and sends the data to the vehicle air conditioning comfort prediction model establishment module;

[0097] The vehicle air conditioning comfort prediction model establishment module quantifies the credibility of the vehicle air conditioning comfort prediction model for the current prediction, adopts different optimization strategies for different types of samples, designs the loss function by introducing weight factors, establishes the vehicle air conditioning comfort prediction model based on the optimization strategy, and sends the data to the model parameter optimization module;

[0098] The model parameter optimization module realizes global search optimization through individual fitness value calculation and mutation probability update mechanism, realizes position update based on mutation probability, and then realizes model parameter optimization; and sends data to the vehicle air conditioning intelligent management module;

[0099] The vehicle air-conditioning intelligent management module realizes the vehicle air-conditioning intelligent management based on the established vehicle air-conditioning comfort prediction model.

[0100] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0101] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0102] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A vehicle air conditioning intelligent management method, characterized in that: The method comprises the following steps: Step S1: data collection, collecting historical vehicle air conditioning operation data; Step S2: Data preprocessing, data cleaning, data conversion, feature engineering and data set division of the collected data; Step S3: Establishing a vehicle air conditioning comfort prediction model, by quantifying the credibility of the vehicle air conditioning comfort prediction model for the current prediction, adopting different optimization strategies for different types of samples, and designing a loss function by introducing a weight factor, and establishing the vehicle air conditioning comfort prediction model based on the optimization strategy; Step S4: Model parameter optimization, realizing global search optimization through individual fitness value calculation and mutation probability update mechanism, realizing position update based on mutation probability, and then realizing model parameter optimization; Step S5: Intelligent management of vehicle air conditioning, realizing intelligent management of vehicle air conditioning based on the established vehicle air conditioning comfort prediction model; Step S3 includes step S35: introducing a weight factor; selecting a scale based on classification confidence, and constructing a vehicle air conditioning comfort prediction model loss function, the formula used is as follows: ； Where LL(·) is the loss function of the vehicle air conditioning comfort prediction model; R1 and R2 are weight factors used to weight samples in different confidence ranges; is the confidence threshold; α is the parameter vector of the model; K is the total number of categories, and i2 is the category index; and are the normalized predicted output values ​​for the i2th category and category i2(s) respectively; is the normalized classification confidence; i2(s) is the true category of sample s; M is the number of model layers; s is the sample.

2. The vehicle air conditioning intelligent management method according to claim 1, characterized in that: In step S3, the establishment of the vehicle air conditioning comfort prediction model specifically includes the following steps: Step S31: model architecture design; the established vehicle air conditioning comfort prediction model includes: an input layer for receiving a preprocessed data set; a hidden layer for processing input features and extracting nonlinear relationships between features; an output layer for outputting a predicted comfort assessment level; the hidden layer includes: a first fully connected layer, including 64 neurons, using a ReLU activation function, and using L2 regularization; a second fully connected layer, including 128 neurons, using a ReLU activation function, and using L2 regularization; a third fully connected layer, including 64 neurons, using a ReLU activation function, and using Dropout; a fourth fully connected layer, including 32 neurons, using a ReLU activation function, and using Dropout; the model uses the Adam optimizer for parameter adjustment; Step S32: define the classification confidence; calculate the definition classification confidence of each sample, the formula used is as follows: ； In the formula, is the classification confidence of sample s for the true category i2(s); α is the parameter vector of the model; It is the output value of sample s and true category i2(s) in the last layer of the model, i.e., layer M-1; is the maximum output value of other categories; Step S33: normalization; determining a normalized scale factor R, normalizing the classification confidence based on the scale factor to obtain a normalized classification confidence; the formula used to calculate the scale factor is as follows: ； In the formula, is the parameter vector of the i1th layer; Step S34: construct a basic loss function; the formula used is as follows: ； Where L(·) is the basic loss function; K is the total number of categories, and i2 is the category index; is the predicted output for the i2th category; Step S35: introducing weight factors; Step S36: Optimize the loss function: Optimize the dual-scale loss function based on a deterministic gradient descent algorithm, and the formula used is as follows: ； In the formula, , and They are the parameter vector and weight factor of the model during the n+1th training respectively; , and are the parameter vector and weight factor of the model during the nth training respectively; τ is the learning rate; μ is the momentum coefficient; is the gradient operator; TR is the training set; is the square of the L2 norm; is the regularization coefficient; Step S37: model determination; a correctness threshold is set in advance. When the vehicle air-conditioning comfort prediction model converges with the loss of the training set, the vehicle air-conditioning comfort prediction model training is completed, otherwise the model parameters are optimized; when the prediction accuracy of the trained vehicle air-conditioning comfort prediction model for the test set is higher than the correctness threshold, the vehicle air-conditioning comfort prediction model is established, otherwise the data set training model is re-divided.

3. The vehicle air conditioning intelligent management method according to claim 1, characterized in that: In step S4, the model parameter optimization specifically includes the following steps: Step S41: initialization; constructing a parameter optimization space based on the learning rate, momentum coefficient, weight factor, regularization coefficient, initial weight and bias of the vehicle air conditioning comfort prediction model; initializing the position of the individual in the optimization population, and taking the prediction accuracy of the vehicle air conditioning comfort prediction model trained based on the individual position and the training set for the test set as the individual fitness value; Step S42: Calculate the mutation probability; the formula used is as follows: ； In the formula, and are the mutation probabilities of the jth dimension of the ith individual at the t+1th iteration and the tth iteration, respectively; is the momentum factor; c1 and c2 are local and global factors respectively; and are the historical optimal position of the j-th dimension of the ith individual in the t-th iteration and the global optimal position of the j-th dimension; is the number of times the jth dimension of the i-th individual has not mutated continuously at the t-th iteration; is the mutation ratio of the jth dimension at the tth iteration; is the number of individuals mutated in dimension j at the tth iteration, and pn is the number of individuals in the population; is the position of the i-th individual in the j-th dimension at the t-th iteration; Step S43: Update individual positions; update individual positions based on mutation probability; the formula used is as follows: ； In the formula, is the position of the i-th individual in the j-th dimension at the t+1-th iteration; is the individual fitness value; Indicates mutation; r is a random number between 0 and 1; Step S44: optimization judgment; a fitness threshold is set in advance. When there is an individual fitness value higher than the fitness threshold, the vehicle air-conditioning comfort prediction model obtained based on the individual position is established as the vehicle air-conditioning comfort prediction model; if the maximum number of iterations is reached, return to step S41; otherwise, return to step S42.

4. The vehicle air conditioning intelligent management method according to claim 1, characterized in that: In step S1, the historical vehicle air-conditioning operation data includes in-vehicle sensor data, external environment data, vehicle operation data, vehicle air-conditioning setting data and comfort assessment level; the vehicle air-conditioning setting data includes the air-conditioning wind speed, air outlet direction and temperature setting; the comfort assessment level is used as a data label.

5. The vehicle air conditioning intelligent management method according to claim 1, characterized in that: In step S5, the vehicle air-conditioning intelligent management is based on the established vehicle air-conditioning comfort prediction model, and collects in-vehicle sensor data, external environment data, vehicle operation data and vehicle air-conditioning setting data to be evaluated in real time, and inputs them into the vehicle air-conditioning comfort prediction model after preprocessing. A comfort evaluation threshold is set in advance, and the vehicle air-conditioning setting data corresponding to the data whose comfort evaluation level predicted by the model is higher than the comfort evaluation threshold is used as the operation setting of the vehicle air-conditioning.

6. The vehicle air conditioning intelligent management method according to claim 1, characterized in that: In step S2, the data cleaning is to remove missing values, outliers, and duplicate data; the data conversion is to standardize and unify data from different data sources; the feature engineering processing is to perform feature selection and feature construction; the data set division is to divide the collected original data set into a test set and a training set.

7. A vehicle air conditioning intelligent management system, used to implement a vehicle air conditioning intelligent management method as claimed in any one of claims 1 to 6, characterized in that: It includes data acquisition module, data preprocessing module, vehicle air conditioning comfort prediction model building module, model parameter optimization module and vehicle air conditioning intelligent management module; The data acquisition module collects historical vehicle air conditioner operation data and sends the data to the data preprocessing module; The data preprocessing module performs data cleaning, data conversion, feature engineering processing and data set division on the collected data, and sends the data to the vehicle air conditioning comfort prediction model establishment module; The vehicle air conditioning comfort prediction model establishment module quantifies the credibility of the vehicle air conditioning comfort prediction model for the current prediction, adopts different optimization strategies for different types of samples, designs the loss function by introducing weight factors, establishes the vehicle air conditioning comfort prediction model based on the optimization strategy, and sends the data to the model parameter optimization module; The model parameter optimization module realizes global search optimization through individual fitness value calculation and mutation probability update mechanism, realizes position update based on mutation probability, and then realizes model parameter optimization; and sends data to the vehicle air conditioning intelligent management module; The vehicle air-conditioning intelligent management module realizes the vehicle air-conditioning intelligent management based on the established vehicle air-conditioning comfort prediction model.

Images