Intelligent monitoring method and system for automobile data based on machine learning

Abstract

The present invention discloses a method and system for intelligent monitoring of automobile data based on machine learning, including data collection, data preprocessing, data scale transformation, establishment of an intelligent monitoring model for automobile data, optimization of the intelligent monitoring model for automobile data, and intelligent monitoring. The present invention relates to the field of automobile data monitoring technology, and specifically refers to a method and system for intelligent monitoring of automobile data based on machine learning. This solution transforms the data scale through data reconstruction, generation of simulation data, and data verification, thereby improving data quality, enriching data diversity, and improving the accuracy of the model; by designing a correlation function, a kernel function, and a loss function and introducing adaptive gradient acceleration and dynamic dimensionality adjustment, the processing capability of the model data is improved, and more effective training is performed; by designing a search acceleration factor, conducting preliminary random exploration, and high-density exploration, the diversity and comprehensiveness of the search are increased, which helps to find the optimal parameter combination.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile data monitoring, and in particular to an automobile data intelligent monitoring method and system based on machine learning. Background Art

[0002] The intelligent monitoring method and system for automobile data based on machine learning utilizes machine learning technology and data processing technology to collect automobile operation data in real time, analyze and process it, build an accurate monitoring model, and intelligently monitor automobile data. However, the traditional intelligent monitoring model for automobile data has problems of low adaptability, flexibility, and accuracy due to insufficient data quality and richness; the traditional intelligent monitoring model for automobile data has problems of poor data relationship capture and data processing capabilities, and poor model training effects; the traditional model parameter optimization algorithm has problems of low search flexibility, efficiency, diversity, and comprehensiveness. 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 monitoring of automobile data based on machine learning. In view of the problems of low adaptability, flexibility and accuracy of the traditional intelligent monitoring model of automobile data due to insufficient data quality and richness, this solution transforms the data scale through data reconstruction, generation of simulation data, data verification, definition of loss and joint training, improves data quality, enriches data diversity, better adapts to data of different types and scales, makes the model more flexible in processing data, and improves the accuracy of the model; in view of the problems of poor data relationship capture and data processing capabilities of the traditional intelligent monitoring model of automobile data, and the low model training efficiency In order to solve the problem of poor results, this solution designs a correlation function to more accurately capture the relationship between data; designs kernel functions to better adapt to the characteristics of different data distributions and improve the model's ability to process complex data; designs a loss function and introduces adaptive gradient acceleration and dynamic dimensionality adjustment to better measure the model's error and guide the model to perform more effective training; in order to solve the problems of low search flexibility, efficiency, diversity and comprehensiveness in traditional model parameter optimization algorithms, this solution designs a search acceleration factor and dynamically adjusts the search speed to make the search more flexible and efficient; designs preliminary random exploration and high-density exploration to increase the diversity and comprehensiveness of the search and help find the optimal parameter combination.

[0004] The technical solution adopted by the present invention is as follows: a method for intelligent monitoring of automobile data based on machine learning, the method comprising the following steps:

[0005] Step S1: data collection;

[0006] Step S2: data preprocessing;

[0007] Step S3: data scale transformation;

[0008] Step S4: Establishing an intelligent monitoring model for automobile data;

[0009] Step S5: optimizing the intelligent monitoring model of automobile data;

[0010] Step S6: Intelligent monitoring.

[0011] Furthermore, in step S1, the data collection is to collect vehicle operating status data, engine performance parameters, vehicle fault code data, vehicle interior environment data, vehicle surrounding environment data and vehicle data status; the vehicle operating status data includes vehicle speed, braking status, steering angle, and tire status; the engine performance parameters include engine speed, engine temperature, fuel consumption data, intake pressure and temperature, and ignition advance angle; the vehicle interior environment data includes temperature and humidity inside the vehicle; the vehicle surrounding environment data includes temperature, humidity, air pressure, road conditions, geographical location and altitude around the vehicle; the vehicle data status includes normal status and abnormal status.

[0012] Furthermore, in step S2, the data preprocessing specifically includes the following steps:

[0013] Step S21: perform data cleaning, correct outliers, and fill in missing values; perform data standardization, and unify the data range and format;

[0014] Step S22: Setting the tag, setting the vehicle data status to the tag data.

[0015] Furthermore, in step S3, the data scale transformation specifically includes the following steps:

[0016] Step S31: Data input, receiving input data and condition information, and outputting the probability distribution of latent variables as follows:

[0017] ;

[0018] in, represents input data, c represents conditional information, represents the output latent variable, represents the probability distribution of latent variables generated according to input data and conditional information, represents the data mean when receiving input data and condition information as input, represents the data variance when receiving input data and condition information as input, O represents the unit matrix, Indicates is the mean value, is the normal distribution of the covariance matrix;

[0019] Step S32: Data reconstruction, receiving latent variables and condition information, generating reconstructed data, which is expressed as follows:

[0020] ;

[0021] in, Reconstructed data, represents the probability distribution of reconstructed data generated based on latent variables and conditional information, represents the data mean when receiving latent variables and conditional information as input, represents the data variance when receiving latent variables and conditional information as input, Indicates is the mean value, is the normal distribution of the covariance matrix;

[0022] Step S33: Generate simulation data, receive hidden variables, condition information and reconstruction data, and generate simulation data, which is expressed as follows:

[0023] ;

[0024] in, represents simulation data, k1, k2 and k3 represent simulation parameters;

[0025] Step S34: Data verification, data verification is performed based on real data, simulation data and condition information, as shown below:

[0026] ;

[0027] in, represents the data test result, mn(·) represents the mean value, Indicates taking the absolute value, Represents real data, Represents the sigmoid function;

[0028] Step S35: Define the loss, set the mean square error of the data scale transformation process as the loss, including the reconstruction loss of the data reconstruction process , the generation loss of the simulation data generation process and the inspection loss of the data inspection process ;

[0029] Step S36: Joint training, design two training strategies: the first is to fix the simulation data generation process, update the data verification process and the data reconstruction process, and minimize the verification loss and the reconstruction loss; the second is to fix the data verification process, update the simulation data generation process and the data reconstruction process, and minimize the reconstruction loss and the generation loss, which are expressed as follows:

[0030] ;

[0031] ;

[0032] in, and denote the first joint loss and the second joint loss respectively, and represents the adjustment parameter;

[0033] Step S37: Scale transformation, generate simulation data, and expand the original data.

[0034] Furthermore, in step S4, the establishment of the automobile data intelligent monitoring model specifically includes the following steps:

[0035] Step S41: Design a correlation function, which is expressed as follows:

[0036] ;

[0037] Among them, i represents the index of the feature dimension, and Represents the values ​​of two data points on the i-th feature dimension, Represents the correlation between two data points on the i-th feature dimension, Represents the sensitivity factor of the i-th feature dimension;

[0038] Step S42: design a kernel function, which is expressed as follows:

[0039] ;

[0040] in, and represents two data points in the dataset, represents the kernel function value of two data points, d represents the total number of data dimensions, represents the association weight of the i-th data dimension, e represents a natural constant, represents the secondary weight, and represents the scale parameter;

[0041] Step S43: Find the support vector, substitute the data points into the kernel function calculation, determine the support vector, learn to divide the data in the feature space, and distinguish samples of different categories;

[0042] Step S44: Design a loss function, which is expressed as follows:

[0043] ;

[0044] Among them, Lf represents the loss value of the model, N represents the total number of sample data, j represents the index of sample data, and max(·) represents the maximum value. represents the true value of the jth sample, k represents the index of the support vector, represents the support coefficient of the kth support vector, represents the kth support vector, represents the kernel function value of the jth sample data and the kth support vector, represents decision bias;

[0045] Step S45: Adaptive gradient acceleration, taking into account the size and direction of the gradient and the changes in the historical gradient to adjust the current gradient, as shown below:

[0046] ;

[0047] Where f represents the number of model training times. represents the gradient of the model during the f+1th training. represents the gradient of the model during the f-th training, represents the learning rate, represents the adaptation parameter, w represents the index of the number of training times, represents the gradient of the model during the w-th training, represents the historical average gradient;

[0048] Step S46: Dynamic dimension shrinkage, using the gradient change rate to measure the impact of feature dimensions on model performance and adjust the training precision of the model;

[0049] Step S47: Model training: training the model based on adaptive gradient acceleration and dynamic dimension shrinkage.

[0050] Furthermore, in step S5, the automobile data intelligent monitoring model is optimized, specifically comprising the following steps:

[0051] Step S51: search initialization, creating a search space for optimization parameters, the search parameters include sensitivity factors, secondary weights and scale parameters, setting the model loss as parameter optimization intensity, and generating an initial optimization search point group;

[0052] Step S52: Design a search acceleration factor, expressed as follows:

[0053] ;

[0054] Where t represents the number of search iterations, represents the search acceleration factor at the tth search iteration, T represents the maximum search iteration number, represents sensitivity, r0 represents a random number between 0 and 1;

[0055] Step S53: Preliminary random exploration, expressed as follows:

[0056] ;

[0057] in, represents the position of the parameter point searched during the t+1th initial random exploration, represents the position of the parameter point searched during the tth initial random exploration, Indicates the position of the current optimal parameter, ul indicates the upper boundary of the search space, dl indicates the lower boundary of the search space, and r1 indicates a random number between 0 and 1;

[0058] Step S54: high-density exploration, expressed as follows:

[0059] ;

[0060] in, represents the position of the parameter point searched during the t+1th high-density exploration, r2 represents a random number between 0 and 1, Indicates the position of the parameter point searched during the t-th high-density exploration;

[0061] Step S55: Parameter optimization search, set the optimization strength threshold of the parameters, first perform preliminary random exploration, calculate the optimization strength of the searched parameter points, if the optimization strength of more than 50% of the search points is improved, perform two high-density explorations; otherwise, continue to perform preliminary random exploration; if five consecutive preliminary random explorations do not meet the conditions for high-density exploration, force a high-density exploration, and then perform the next preliminary random exploration; if during the search process, more than 80% of the search points have an optimization strength greater than the optimization strength threshold, stop the search, and output the parameters of the parameter point with the largest optimization strength at this time as the optimal parameter; if the maximum number of search iterations is reached, search again; otherwise, continue the search.

[0062] Furthermore, in step S6, the intelligent monitoring is carried out by collecting data of the vehicle in real time during operation and inputting it into the vehicle data intelligent monitoring model. The model analyzes the input data, monitors abnormal conditions of the vehicle data in real time, and continuously accumulates and analyzes historical data to continuously update and improve the model.

[0063] The automobile data intelligent monitoring system based on machine learning provided by the present invention includes a data acquisition module, a data preprocessing module, a data scale transformation module, a module for establishing an automobile data intelligent monitoring model, a module for optimizing the automobile data intelligent monitoring model and an intelligent monitoring module;

[0064] The data acquisition module collects vehicle operation status data, engine performance parameters, vehicle fault code data, vehicle interior environment data, vehicle surrounding environment data and vehicle data status, and sends the data to the data preprocessing module;

[0065] The data preprocessing module receives the data sent by the data acquisition module, performs data preprocessing on the received data, and sends the preprocessed data to the data scale conversion module;

[0066] The data scale conversion module receives the data sent by the data preprocessing module, performs scale conversion on the data, expands and enriches the original data, and sends the data to the module for establishing the automobile data intelligent monitoring model;

[0067] The automobile data intelligent monitoring model establishment module receives the data sent by the data scale conversion module, establishes the automobile data intelligent monitoring model, and sends the data to the automobile data intelligent monitoring model optimization module;

[0068] The automobile data intelligent monitoring model optimization module receives data sent by the automobile data intelligent monitoring model establishment module, optimizes parameters of the model, and sends the data to the intelligent monitoring module;

[0069] The intelligent monitoring module receives data from the automobile data intelligent monitoring model optimization module, collects data of the automobile in real time during operation, and inputs the data into the automobile data intelligent monitoring model. The model monitors the status of the automobile data in real time.

[0070] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0071] (1) In view of the problems of low adaptability, flexibility and accuracy of traditional automobile data intelligent monitoring models due to insufficient data quality and richness, this solution transforms the data scale through data reconstruction, generation of simulation data, data verification, definition of loss and joint training, improves data quality, enriches data diversity, and better adapts to data of different types and sizes, making the model more flexible in processing data and improving the accuracy of the model.

[0072] (2) In view of the problems that traditional automobile data intelligent monitoring models have poor data relationship capture and data processing capabilities, and poor model training effects, this solution designs a correlation function to more accurately capture the relationship between data; designs a kernel function to better adapt to the characteristics of different data distributions and improve the model's ability to process complex data; designs a loss function and introduces adaptive gradient acceleration and dynamic dimensionality adjustment to better measure the model's error and guide the model to be trained more effectively.

[0073] (3) To address the problems of low search flexibility, efficiency, diversity and comprehensiveness in traditional model parameter optimization algorithms, this solution designs a search acceleration factor to dynamically adjust the search speed to make the search more flexible and efficient. It also designs preliminary random exploration and high-density exploration to increase the diversity and comprehensiveness of the search, which helps to find the optimal parameter combination. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 A schematic diagram of the process of the intelligent monitoring method of automobile data based on machine learning provided by the present invention;

[0075] Figure 2 A schematic diagram of an automobile data intelligent monitoring system based on machine learning provided by the present invention;

[0076] Figure 3 is a schematic flow chart of step S3;

[0077] Figure 4 is a schematic flow chart of step S4;

[0078] Figure 5 It is a flowchart diagram of step S5.

[0079] 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

[0080] 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.

[0081] In the description of the present invention, it should be understood 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 device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as limiting the present invention.

[0082] Example 1, see Figure 1 The present invention provides a method for intelligently monitoring automobile data based on machine learning, the method comprising the following steps:

[0083] Step S1: data collection, collecting vehicle operation status data, engine performance parameters, vehicle fault code data, vehicle interior environment data, vehicle surrounding environment data and vehicle data status;

[0084] Step S2: Data preprocessing, data cleaning and data standardization, and label setting;

[0085] Step S3: transform the data scale, determine the probability distribution of latent variables, perform data reconstruction, generate simulation data, verify data, define loss and jointly train to transform the data scale and expand the original data;

[0086] Step S4: Establishing an intelligent monitoring model for automobile data, by designing a correlation function, designing a kernel function, finding a support vector, designing a loss function, adaptive gradient acceleration, and dynamic dimension shrinkage;

[0087] Step S5: optimizing the intelligent monitoring model of automobile data, designing a search acceleration factor, and designing preliminary random exploration and high-density exploration strategies to perform parameter optimization search;

[0088] Step S6: Intelligent monitoring, by collecting data of the vehicle in real time during operation and inputting it into the vehicle data intelligent monitoring model, real-time monitoring of abnormal conditions in the vehicle data.

[0089] Example 2, see Figure 1 This embodiment is based on the above embodiment. In step S1, the data collection is to collect vehicle running status data, engine performance parameters, vehicle fault code data, vehicle interior environment data, vehicle surrounding environment data and vehicle data status; the vehicle running status data includes vehicle speed, braking status, steering angle, and tire status; the engine performance parameters include engine speed, engine temperature, fuel consumption data, intake pressure and temperature, and ignition advance angle; the vehicle interior environment data includes temperature and humidity in the vehicle; the vehicle surrounding environment data includes temperature, humidity, air pressure, road conditions, geographical location and altitude around the vehicle; the vehicle data status includes normal status and abnormal status.

[0090] Example 3, see Figure 1 This embodiment is based on the above embodiment. In step S2, the data preprocessing specifically includes the following steps:

[0091] Step S21: perform data cleaning, correct outliers, and fill in missing values; perform data standardization, and unify the data range and format;

[0092] Step S22: Setting the tag, setting the vehicle data status to the tag data.

[0093] Example 4, see Figure 1 and Figure 3 This embodiment is based on the above embodiment. In step S3, the data scale transformation specifically includes the following steps:

[0094] Step S31: Data input, receiving input data and condition information, and outputting the probability distribution of latent variables as follows:

[0095] ;

[0096] in, represents input data, c represents conditional information, represents the output latent variable, represents the probability distribution of latent variables generated according to input data and conditional information, represents the data mean when receiving input data and condition information as input, represents the data variance when receiving input data and condition information as input, O represents the unit matrix, Indicates is the mean value, is the normal distribution of the covariance matrix;

[0097] Step S32: Data reconstruction, receiving latent variables and condition information, generating reconstructed data, which is expressed as follows:

[0098] ;

[0099] in, Reconstructed data, represents the probability distribution of reconstructed data generated based on latent variables and conditional information, represents the data mean when receiving latent variables and conditional information as input, represents the data variance when receiving latent variables and conditional information as input, Indicates is the mean value, is the normal distribution of the covariance matrix;

[0100] Step S33: Generate simulation data, receive hidden variables, condition information and reconstruction data, and generate simulation data, which is expressed as follows:

[0101] ;

[0102] in, represents simulation data, k1, k2 and k3 represent simulation parameters;

[0103] Step S34: Data verification, data verification is performed based on real data, simulation data and condition information, as shown below:

[0104] ;

[0105] in, represents the data test result, mn(·) represents the mean value, Indicates taking the absolute value, Represents real data, Represents the sigmoid function;

[0106] Step S35: Define the loss, set the mean square error of the data scale transformation process as the loss, including the reconstruction loss of the data reconstruction process , the generation loss of the simulation data generation process and the inspection loss of the data inspection process ;

[0107] Step S36: Joint training, design two training strategies: the first is to fix the simulation data generation process, update the data verification process and the data reconstruction process, and minimize the verification loss and the reconstruction loss; the second is to fix the data verification process, update the simulation data generation process and the data reconstruction process, and minimize the reconstruction loss and the generation loss, which are expressed as follows:

[0108] ;

[0109] ;

[0110] in, and denote the first joint loss and the second joint loss respectively, and represents the adjustment parameter;

[0111] Step S37: Scale transformation, generate simulation data, and expand the original data.

[0112] By performing the above operations, in order to address the problems of low model adaptability, flexibility and accuracy caused by insufficient data quality and richness in traditional automobile data intelligent monitoring models, this solution transforms the data scale through data reconstruction, generation of simulation data, data verification, definition of loss and joint training, improves data quality, enriches data diversity, better adapts to data of different types and sizes, makes the model more flexible in processing data, and improves the accuracy of the model.

[0113] Example 5, see Figure 1 and Figure 4 This embodiment is based on the above embodiment. In step S4, the establishment of the automobile data intelligent monitoring model specifically includes the following steps:

[0114] Step S41: Design a correlation function, which is expressed as follows:

[0115] ;

[0116] Among them, i represents the index of the feature dimension, and Represents the values ​​of two data points on the i-th feature dimension, Represents the correlation between two data points on the i-th feature dimension, Represents the sensitivity factor of the i-th feature dimension;

[0117] Step S42: design a kernel function, which is expressed as follows:

[0118] ;

[0119] in, and represents two data points in the dataset, represents the kernel function value of two data points, d represents the total number of data dimensions, represents the association weight of the i-th data dimension, e represents a natural constant, represents the secondary weight, and represents the scale parameter;

[0120] Step S43: Find the support vector, substitute the data points into the kernel function calculation, determine the support vector, learn to divide the data in the feature space, and distinguish samples of different categories;

[0121] Step S44: Design a loss function, which is expressed as follows:

[0122] ;

[0123] Among them, Lf represents the loss value of the model, N represents the total number of sample data, j represents the index of sample data, and max(·) represents the maximum value. represents the true value of the jth sample, k represents the index of the support vector, represents the support coefficient of the kth support vector, represents the kth support vector, represents the kernel function value of the jth sample data and the kth support vector, represents decision bias;

[0124] Step S45: Adaptive gradient acceleration, taking into account the size and direction of the gradient and the changes in the historical gradient to adjust the current gradient, as shown below:

[0125] ;

[0126] Where f represents the number of model training times. represents the gradient of the model during the f+1th training. represents the gradient of the model during the f-th training, represents the learning rate, represents the adaptation parameter, w represents the index of the number of training times, represents the gradient of the model during the w-th training, represents the historical average gradient;

[0127] Step S46: Dynamic dimension shrinkage, using the gradient change rate to measure the impact of feature dimensions on model performance and adjust the training precision of the model;

[0128] Step S47: Model training: training the model based on adaptive gradient acceleration and dynamic dimension shrinkage.

[0129] By performing the above operations, in view of the problems of traditional automobile data intelligent monitoring models in data relationship capture and data processing capabilities, and poor model training effects, this solution designs a correlation function to more accurately capture the relationship between data; designs a kernel function to better adapt to the characteristics of different data distributions and improve the model's ability to process complex data; designs a loss function and introduces adaptive gradient acceleration and dynamic dimensionality adjustment to better measure the model's error and guide the model to be trained more effectively.

[0130] Example 6, see Figure 1 and Figure 5 This embodiment is based on the above embodiment. In step S5, the automobile data intelligent monitoring model is optimized, specifically including the following steps:

[0131] Step S51: search initialization, creating a search space for optimization parameters, the search parameters include sensitivity factors, secondary weights and scale parameters, setting the model loss as parameter optimization intensity, and generating an initial optimization search point group;

[0132] Step S52: Design a search acceleration factor, expressed as follows:

[0133] ;

[0134] Where t represents the number of search iterations, represents the search acceleration factor at the tth search iteration, T represents the maximum search iteration number, represents sensitivity, r0 represents a random number between 0 and 1;

[0135] Step S53: Preliminary random exploration, expressed as follows:

[0136] ;

[0137] in, represents the position of the parameter point searched during the t+1th initial random exploration, represents the position of the parameter point searched during the tth initial random exploration, Indicates the position of the current optimal parameter, ul indicates the upper boundary of the search space, dl indicates the lower boundary of the search space, and r1 indicates a random number between 0 and 1;

[0138] Step S54: high-density exploration, expressed as follows:

[0139] ;

[0140] in, represents the position of the parameter point searched during the t+1th high-density exploration, r2 represents a random number between 0 and 1, Indicates the position of the parameter point searched during the t-th high-density exploration;

[0141] Step S55: Parameter optimization search, set the optimization strength threshold of the parameters, first perform preliminary random exploration, calculate the optimization strength of the searched parameter points, if the optimization strength of more than 50% of the search points is improved, perform two high-density explorations; otherwise, continue to perform preliminary random exploration; if five consecutive preliminary random explorations do not meet the conditions for high-density exploration, force a high-density exploration, and then perform the next preliminary random exploration; if during the search process, more than 80% of the search points have an optimization strength greater than the optimization strength threshold, stop the search, and output the parameters of the parameter point with the largest optimization strength at this time as the optimal parameter; if the maximum number of search iterations is reached, search again; otherwise, continue the search.

[0142] By performing the above operations, in order to address the problems of low search flexibility, efficiency, diversity and comprehensiveness in the traditional model parameter optimization algorithm, this solution designs a search acceleration factor, dynamically adjusts the search speed, makes the search more flexible and efficient, designs preliminary random exploration and high-density exploration, increases the diversity and comprehensiveness of the search, and helps to find the optimal parameter combination.

[0143] Embodiment 7, see Figure 1 This embodiment is based on the above embodiment. In step S6, the intelligent monitoring is carried out by collecting data of the vehicle in real time during operation and inputting it into the vehicle data intelligent monitoring model. The model analyzes the input data and monitors abnormal conditions of the vehicle data in real time. At the same time, it continuously accumulates and analyzes historical data to continuously update and improve the model.

[0144] Embodiment 8, see Figure 2, this embodiment is based on the above embodiment, the automobile data intelligent monitoring system based on machine learning provided by the present invention includes a data acquisition module, a data preprocessing module, a data scale transformation module, a module for establishing an automobile data intelligent monitoring model, an automobile data intelligent monitoring model optimization module and an intelligent monitoring module;

[0145] The data acquisition module collects vehicle operation status data, engine performance parameters, vehicle fault code data, vehicle interior environment data, vehicle surrounding environment data and vehicle data status, and sends the data to the data preprocessing module;

[0146] The data preprocessing module receives the data sent by the data acquisition module, performs data preprocessing on the received data, and sends the preprocessed data to the data scale conversion module;

[0147] The data scale conversion module receives the data sent by the data preprocessing module, performs scale conversion on the data, expands and enriches the original data, and sends the data to the module for establishing the automobile data intelligent monitoring model;

[0148] The automobile data intelligent monitoring model establishment module receives the data sent by the data scale conversion module, establishes the automobile data intelligent monitoring model, and sends the data to the automobile data intelligent monitoring model optimization module;

[0149] The automobile data intelligent monitoring model optimization module receives data sent by the automobile data intelligent monitoring model establishment module, optimizes parameters of the model, and sends the data to the intelligent monitoring module;

[0150] The intelligent monitoring module receives data from the automobile data intelligent monitoring model optimization module, collects data of the automobile in real time during operation, and inputs the data into the automobile data intelligent monitoring model. The model monitors the status of the automobile data in real time.

[0151] 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.

[0152] 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.

[0153] 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. An intelligent automobile data monitoring method based on machine learning, characterized in that: The method comprises the following steps: Step S1: data collection, collecting vehicle operation status data, engine performance parameters, vehicle fault code data, vehicle interior environment data, vehicle surrounding environment data and vehicle data status; Step S2: Data preprocessing, data cleaning and data standardization, and label setting; Step S3: transform the data scale, determine the probability distribution of latent variables, perform data reconstruction, generate simulation data, verify data, define loss and jointly train to transform the data scale and expand the original data; Step S4: Establishing an intelligent monitoring model for automobile data, by designing a correlation function, designing a kernel function, finding a support vector, designing a loss function, adaptive gradient acceleration, and dynamic dimension shrinkage; Step S5: Optimizing the intelligent monitoring model of automobile data by designing a search acceleration factor, designing preliminary random exploration and high-density exploration strategies to perform parameter optimization search; Step S6: Intelligent monitoring, real-time collection of data during the operation of the vehicle, and inputting the data into the vehicle data intelligent monitoring model to monitor the vehicle data; In step S3, the data scale transformation specifically includes the following steps: Step S31: Data input, receiving input data and condition information, and outputting the probability distribution of latent variables as follows: ； in, represents input data, c represents conditional information, represents the output latent variable, represents the probability distribution of latent variables generated according to input data and conditional information, represents the data mean when receiving input data and condition information as input, represents the data variance when receiving input data and condition information as input, O represents the unit matrix, Indicates is the mean value, is the normal distribution of the covariance matrix; Step S32: Data reconstruction, receiving latent variables and condition information, generating reconstructed data, which is expressed as follows: ； in, Represents the reconstructed data, represents the probability distribution of reconstructed data generated based on latent variables and conditional information, represents the data mean when receiving latent variables and conditional information as input, represents the data variance when receiving latent variables and conditional information as input, Indicates is the mean value, is the normal distribution of the covariance matrix; Step S33: Generate simulation data, receive hidden variables, condition information and reconstruction data, and generate simulation data, which is expressed as follows: ； in, represents simulation data, k1, k2 and k3 represent simulation parameters; Step S34: Data verification, data verification is performed based on real data, simulation data and condition information, as shown below: ； in, represents the data test result, mn(·) represents the mean value, Indicates taking the absolute value, Represents real data, Represents the sigmoid function; Step S35: Define the loss, set the mean square error of the data scale transformation process as the loss, including the reconstruction loss of the data reconstruction process , the generation loss of the simulation data generation process and the inspection loss of the data inspection process ; Step S36: Joint training, design two training strategies: the first is to fix the simulation data generation process, update the data verification process and the data reconstruction process, and minimize the verification loss and the reconstruction loss; the second is to fix the data verification process, update the simulation data generation process and the data reconstruction process, and minimize the reconstruction loss and the generation loss, which are expressed as follows: ； ； in, and denote the first joint loss and the second joint loss respectively, and represents the adjustment parameter; Step S37: scale transformation, generating simulation data, and expanding original data; In step S4, the establishment of the automobile data intelligent monitoring model specifically includes the following steps: Step S41: Design a correlation function, which is expressed as follows: ； Among them, i represents the index of the feature dimension, and Represents the values ​​of two data points on the i-th feature dimension, Represents the correlation between two data points on the i-th feature dimension, Represents the sensitivity factor of the i-th feature dimension; Step S42: design a kernel function, which is expressed as follows: ； in, and represents two data points in the dataset, represents the kernel function value of two data points, d represents the total number of data dimensions, represents the association weight of the i-th data dimension, e represents a natural constant, represents the secondary weight, and represents the scale parameter; Step S43: Find the support vector, substitute the data points into the kernel function calculation, and determine the support vector; Step S44: Design a loss function, which is expressed as follows: ； Among them, Lf represents the loss value of the model, N represents the total number of sample data, j represents the index of sample data, and max(·) represents the maximum value. represents the true value of the jth sample, k represents the index of the support vector, represents the support coefficient of the kth support vector, represents the kth support vector, represents the kernel function value of the jth sample data and the kth support vector, represents decision bias; Step S45: Adaptive gradient acceleration, taking into account the size and direction of the gradient and the changes in the historical gradient to adjust the current gradient, as shown below: ； Where f represents the number of model training times. represents the gradient of the model during the f+1th training. represents the gradient of the model during the f-th training, represents the learning rate, represents the adaptation parameter, w represents the index of the number of training times, represents the gradient of the model during the w-th training, represents the historical average gradient; Step S46: Dynamic dimension shrinkage, using the gradient change rate to measure the impact of feature dimensions on model performance and adjust the training precision of the model; Step S47: Model training: training the model based on adaptive gradient acceleration and dynamic dimension shrinkage.

2. The method for intelligent monitoring of automobile data based on machine learning according to claim 1, characterized in that: In step S5, the vehicle data intelligent monitoring model is optimized, specifically including the following steps: Step S51: search initialization, creating a search space for optimization parameters, the search parameters include sensitivity factors, secondary weights and scale parameters, setting the model loss as parameter optimization intensity, and generating an initial optimization search point group; Step S52: Design a search acceleration factor, expressed as follows: ； Where t represents the number of search iterations, represents the search acceleration factor at the tth search iteration, T represents the maximum search iteration number, represents sensitivity, r0 represents a random number between 0 and 1; Step S53: Preliminary random exploration, expressed as follows: ； in, represents the position of the parameter point searched during the t+1th initial random exploration, represents the position of the parameter point searched during the tth initial random exploration, Indicates the position of the current optimal parameter, ul indicates the upper boundary of the search space, dl indicates the lower boundary of the search space, and r1 indicates a random number between 0 and 1; Step S54: high-density exploration, expressed as follows: ； in, represents the position of the parameter point searched during the t+1th high-density exploration, r2 represents a random number between 0 and 1, Indicates the position of the parameter point searched during the t-th high-density exploration; Step S55: Parameter optimization search, set the optimization strength threshold of the parameters, first perform preliminary random exploration, calculate the optimization strength of the searched parameter points, if the optimization strength of more than 50% of the search points is improved, perform two high-density explorations; otherwise, continue to perform preliminary random exploration; if five consecutive preliminary random explorations do not meet the conditions for high-density exploration, force a high-density exploration, and then perform the next preliminary random exploration; if during the search process, more than 80% of the search points have an optimization strength greater than the optimization strength threshold, stop the search, and output the parameters of the parameter point with the largest optimization strength at this time as the optimal parameter; if the maximum number of search iterations is reached, search again; otherwise, continue the search.

3. The method for intelligent monitoring of automobile data based on machine learning according to claim 1, characterized in that: In step S2, the data preprocessing specifically includes the following steps: Step S21: perform data cleaning, correct outliers, and fill in missing values; perform data standardization, and unify the data range and format; Step S22: Setting the tag, setting the vehicle data status to the tag data.

4. The method for intelligent monitoring of automobile data based on machine learning according to claim 1, characterized in that: In step S1, the data collection is to collect vehicle operation status data, engine performance parameters, vehicle fault code data, vehicle interior environment data, vehicle surrounding environment data and vehicle data status; the vehicle operation status data includes vehicle speed, braking status, steering angle, tire status; the engine performance parameters include engine speed, engine temperature, fuel consumption data, intake pressure and temperature and ignition advance angle; the vehicle interior environment data includes temperature and humidity in the vehicle; the vehicle surrounding environment data includes temperature, humidity, air pressure, road conditions, geographical location and altitude around the vehicle; the vehicle data status includes normal status and abnormal status; In step S6, the intelligent monitoring is carried out by collecting data of the vehicle in real time during operation and inputting it into the vehicle data intelligent monitoring model. The model analyzes the input data, monitors abnormal conditions in the vehicle data in real time, and continuously accumulates and analyzes historical data to update and improve the model.

5. An intelligent vehicle data monitoring system based on machine learning, used to implement the intelligent vehicle data monitoring method based on machine learning as described in any one of claims 1 to 4, characterized in that: It includes data acquisition module, data preprocessing module, data scale transformation module, automobile data intelligent monitoring model establishment module, automobile data intelligent monitoring model optimization module and intelligent monitoring module; The data acquisition module collects vehicle operation status data, engine performance parameters, vehicle fault code data, vehicle interior environment data, vehicle surrounding environment data and vehicle data status, and sends the data to the data preprocessing module; The data preprocessing module receives the data sent by the data acquisition module, performs data preprocessing on the received data, and sends the preprocessed data to the data scale conversion module; The data scale conversion module receives the data sent by the data preprocessing module, performs scale conversion on the data, expands the original data, and sends the data to the automobile data intelligent monitoring model establishment module; The automobile data intelligent monitoring model establishment module receives the data sent by the data scale conversion module, establishes the automobile data intelligent monitoring model, and sends the data to the automobile data intelligent monitoring model optimization module; The automobile data intelligent monitoring model optimization module receives data sent by the automobile data intelligent monitoring model establishment module, optimizes parameters of the model, and sends the data to the intelligent monitoring module; The intelligent monitoring module receives data from the automobile data intelligent monitoring model optimization module, collects data of the automobile in real time during operation, and inputs the data into the automobile data intelligent monitoring model. The model monitors the status of the automobile data in real time.

Images