A method and system for identifying vehicle operation faults based on simulator

By establishing a database of vehicle operation fault abnormalities and using analysis and abnormal detection algorithms, fault characteristic data is extracted, and combined with particle swarm optimization algorithms, the problem of low efficiency in automobile fault diagnosis in the existing technology is solved, efficient fault identification and prediction is achieved, and improved The reliability and safety of the car.

CN118656583BActive Publication Date: 2024-12-03NANTONG ZHONGQIANG AUTOMOBILE SERVICE CO LTD
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202411138777.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-19
Publication Date
2024-12-03
Estimated Expiration
2044-08-19

AI Technical Summary

Technical Problem

In the prior art, it is not convenient to compare abnormal data with the vehicle operation fault abnormality database and abnormal detection algorithm, resulting in low efficiency of automobile fault diagnosis, low accuracy and efficiency, and the inability to predict potential problems in advance, which increases the sudden failure to operate the automobile. Influence.

Method used

By establishing a database of vehicle operation fault abnormalities, obtaining and preprocessing the normal and fault data of the car, using analytical algorithms to extract fault characteristic data, combining anomaly detection algorithm and particle swarm optimization algorithm for diagnosis, establishing a fault prediction model, and evaluating the future operating performance of the car in real time. And formulate maintenance measures.

Benefits of technology

It realizes more accurate fault identification and diagnosis, predicts potential problems in advance, improves the reliability and safety of the car, reduces fault frequency and maintenance costs, and improves operating efficiency and overall performance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118656583B_ABST
    Figure CN118656583B_ABST
Patent Text Reader

Abstract

The present invention discloses a method and system for identifying automobile operation faults based on a simulator, and relates to the technical field of automobile fault diagnosis. The method for identifying automobile operation faults based on a simulator includes the following steps: setting the corresponding relationship between parameters in a fault anomaly database; obtaining automobile operation fault characteristic data; obtaining key parameters affecting automobile operation faults; comparing the diagnosis results with the parameters in the fault anomaly database to determine whether they conform to known fault modes and determine the fault type. The present invention can more accurately identify and diagnose automobile faults by combining simulator data and anomaly detection algorithms, and can predict potential problems before faults occur by using prediction models, thereby achieving preventive maintenance, and can avoid large-scale faults, reduce maintenance costs and time, and thus reduce the frequency and severity of faults, and improve the operating efficiency and safety of the automobile.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of automobile fault diagnosis, and in particular to a method and system for identifying automobile operation faults based on a simulator. Background Art

[0002] From the perspective of automobile structure, automobiles are composed of multiple parts, which are then combined into assemblies or systems. Each system operates and cooperates according to the designed functions to meet the overall operation requirements of the automobile. During the driving process of the automobile, the parts, components, and assemblies of the vehicle will continue to wear out, resulting in continuous automobile failures. Serious automobile failures are very likely to cause traffic accidents, resulting in casualties and property losses. In order to reduce the occurrence of automobile failures and ensure the driving safety of the automobile, a simulator is used to simulate the operation of the automobile, determine the possible failures of the automobile during operation, and repair the automobile according to the failures obtained by the simulator to reduce the probability of traffic accidents caused by automobile failures. In the simulator simulating the operation scenario of the automobile, when the monitoring data of the vehicle operation is abnormal, it is necessary to accurately judge the abnormal data points and adjust the vehicle parameters according to the abnormal data points to ensure the overall stability of the vehicle body and thus protect the personal safety of the occupants.

[0003] In the prior art, it is not convenient to combine the automobile operation fault anomaly database and the anomaly detection algorithm to perform abnormal data comparison, so it is not convenient to determine the fault type based on the known fault mode, which reduces the efficiency of diagnosing automobile faults, and it is not convenient to effectively classify a large amount of fault feature data and identify key fault features, which reduces the accuracy and efficiency of fault identification. At the same time, it is not convenient to predict potential problems in advance when a fault occurs, and thus intervention and maintenance cannot be performed in advance, which increases the impact of sudden faults on automobile operation.

[0004] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention proposes a simulator-based automobile operation fault identification method and system, which solves the problem mentioned in the above background technology that it is not convenient to combine the automobile operation fault anomaly database and the anomaly detection algorithm to compare the abnormal data, and thus it is not convenient to determine the fault type according to the known fault mode, which reduces the efficiency of diagnosing automobile faults, and it is not convenient to effectively classify a large amount of fault feature data and identify key fault features, which reduces the accuracy and efficiency of fault identification. At the same time, it is not convenient to predict potential problems in advance when a fault occurs, and thus it is impossible to intervene and repair in advance, which increases the impact of sudden faults on automobile operation.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0007] According to one aspect of the present invention, a method for identifying vehicle operation faults based on a simulator is provided, and the method for identifying vehicle operation faults based on a simulator comprises the following steps:

[0008] S1. Establishing a database of abnormal vehicle operation failures and setting the corresponding relationship between various parameters in the abnormal failure database;

[0009] S2. Run the car model in the simulator, obtain the normal operation data and fault data of the car, and perform preprocessing to obtain the car operation fault characteristic data;

[0010] S3, analyzing the obtained vehicle operation fault characteristic data through an analysis algorithm to obtain key parameters that affect the vehicle operation fault;

[0011] S4. Diagnose the key parameters that affect the vehicle operation fault by using an anomaly detection algorithm, and compare the diagnosis results with the parameters in the fault anomaly database to determine whether they conform to the known fault mode and determine the fault type;

[0012] S5. According to the determined fault type, a vehicle operation fault prediction model is established through a classification algorithm, and the vehicle operation fault prediction model is used to evaluate the future operation performance of the vehicle in real time;

[0013] S6. Develop and implement corresponding maintenance or optimization measures based on the evaluation results of the vehicle operation fault prediction model.

[0014] Furthermore, running the car model in the simulator, obtaining the normal operation data and fault data of the car, and performing preprocessing to obtain the car operation fault characteristic data includes the following steps:

[0015] S21, obtaining duplicate data corresponding to normal operation data and fault data of the vehicle in the simulator, and performing denoising, filtering and smoothing on the duplicate data;

[0016] S22, connecting the unprocessed data rows in the normal operation data and fault data of the vehicle in the simulator to generate a new data table, and associating different data tables through external key values ​​to generate a complete data table, and obtain an accurate data set;

[0017] S23. Extract features from the accurate data set and obtain corresponding vehicle operation fault feature data.

[0018] Furthermore, the key parameters affecting the vehicle operation failure are obtained by analyzing the characteristic data of the vehicle operation failure through the analysis algorithm:

[0019] S31, using the t-SNE method to map the obtained automobile operation fault feature data to the feature space, and select a number of feature vectors to form an automobile fault feature point set;

[0020] S32, taking a number of cluster centers as initial solutions, setting a number of particle numbers and an optimal number of iterations, and randomly generating a number of initial solutions;

[0021] S33, according to the current cluster center position, the fitness value of each particle is calculated through the fitness function, and the current fitness value is updated as the individual optimal value and the current position is updated as the individual optimal position, and the global optimal value and the global optimal position are found through the individual optimal values ​​of all particles;

[0022] S34, judging whether the set optimal number of iterations is reached, if so, going to step S38; if not, continuing to update the speed and position of the particle;

[0023] S35, according to the updated particle positions, using the minimum distance principle to assign each automobile fault sample in the automobile fault feature point set to a corresponding number of cluster centers;

[0024] S36, recalculating the fitness value of each particle. If the fitness value of a particle is less than its individual best fitness value, then updating the individual best fitness value of the particle to the current fitness value, and updating its individual best position;

[0025] S37, compare the individual best fitness values ​​of all particles, find the minimum value as the global extreme value, update the global extreme value position, and return to step S34 to continue iteration;

[0026] S38. After the iteration is completed, the final clustering result is used as the key parameter affecting the vehicle operation failure.

[0027] Furthermore, the calculation formula of the fitness function is:

[0028] ;

[0029] in, S Expressed as the fitness value of the particle;

[0030] T Represented as a set of feature points;

[0031] M Represented as cluster center;

[0032] F Expressed as fitness function;

[0033] Expressed as x The cluster centers and jThe distance between sample points.

[0034] Furthermore, the key parameters that affect the vehicle operation fault are diagnosed by an anomaly detection algorithm, and the diagnosis results are compared with the parameters in the fault anomaly database to determine whether they conform to the known fault mode and determine the fault type, including the following steps:

[0035] S41, obtaining a key parameter data set that affects vehicle operation failures, and using the key parameter data set as a training sample for the SVM;

[0036] S42, obtaining optimal parameters of SVM parameters through an optimization algorithm;

[0037] S43, using the obtained optimal SVM parameters to train the training set, and establishing an abnormal point detection model for automobile fault diagnosis;

[0038] S44, inputting key parameters that affect vehicle operation failure into an abnormal point detection model, and performing real-time fault diagnosis;

[0039] S45, comparing the fault results detected by the anomaly detection model with the known fault modes in the fault anomaly database, and confirming whether they are consistent;

[0040] S46. Determine the specific fault type that affects the operation of the vehicle based on the comparison results.

[0041] Furthermore, obtaining the optimal parameters of the SVM parameters through the optimization algorithm includes the following steps:

[0042] S421, initializing the candidate solution set and related parameters of the optimization algorithm;

[0043] S422, calculating the fitness value of each candidate solution, selecting the best position and the best fitness value in the current iteration number, and selecting the current worst position and the worst fitness value;

[0044] S423, selecting a candidate solution with the best fitness value from the candidate solution set as the initial finder, and updating the initial finder position according to the finder position update formula;

[0045] S424, generating a refraction reverse learning candidate solution of the initial discoverer according to the refraction reverse learning solution formula;

[0046] S425, comparing the fitness values ​​of the initial discoverer and its reverse learning candidate solution, and selecting the one with the best fitness value as the new discoverer;

[0047] S426, taking the candidate solutions other than the new discoverer as joiners, and updating the positions of the joiners according to the variable logarithmic spiral formula;

[0048] S427, randomly selecting a candidate solution from new discoverers and joiners as a sentinel, and updating the position of the sentinel according to the sentinel position update formula;

[0049] S428, determine whether the optimization algorithm has reached the optimal number of iterations, if so, the loop ends, if not, return to step S423;

[0050] S429: Output the global optimal position and the optimal fitness value to obtain the optimal parameters of the SVM parameters.

[0051] Furthermore, the refraction reverse learning solution formula is:

[0052] ;

[0053] in, is expressed as an infinitesimal constant;

[0054] u Represents the current iteration number;

[0055] U Expressed as the total number of iterations;

[0056] Indicates the number of iterations in the current z The original discoverer was c The location of the dimension;

[0057] Expressed as Candidate solutions for reverse learning of ;

[0058] a c Expressed as c The upper boundary in the dimensional space;

[0059] b c Expressed as c The lower boundary in dimensional space;

[0060] k Indicates adjustment parameters;

[0061] ω is represented as a hyperparameter.

[0062] Furthermore, according to the determined fault type, a vehicle operation fault prediction model is established through a classification algorithm, and the future operation performance of the vehicle is evaluated in real time using the vehicle operation fault prediction model, including the following steps:

[0063] S51, inputting an initial sample set of automobile fault characteristic data under different operating conditions;

[0064] S52, classifying all automobile fault feature data in the initial sample set to obtain a number of sample subsets with different key features;

[0065] S53, performing loop processing on each sample subset, and obtaining a coefficient matrix related to the vehicle fault type through analysis;

[0066] S54, rearranging the coefficient matrix, and transforming the initial sample set based on the coefficient matrix to generate a new automobile fault feature data sample set;

[0067] S55, training the vehicle operation fault prediction model based on the new vehicle fault feature data sample set to obtain a sub-model of the vehicle operation fault prediction model, returning to step S52, and looping a preset number of times to obtain an integrated vehicle operation fault prediction model;

[0068] S56, acquiring real-time automobile fault characteristic data during automobile operation, and inputting the real-time automobile fault characteristic data into an integrated automobile operation fault prediction model;

[0069] S57. Analyze and predict the real-time automobile fault characteristic data through the automobile operation fault prediction model, and obtain an evaluation of the future operation performance of the automobile.

[0070] Further, loop processing is performed on each sample subset, and a coefficient matrix related to the vehicle fault type is obtained by analysis, including the following steps:

[0071] S531, resampling the automobile fault feature data sample subset by using a stratified sampling method to generate a new automobile fault feature data sample subset;

[0072] S532, performing principal component analysis on a new subset of automobile fault feature data samples, and obtaining a principal component coefficient vector related to the automobile fault type;

[0073] S533, repeatedly executing step S531 to step S532, processing different test feature data sample subsets, obtaining several groups of principal component coefficient vectors, and merging the principal component coefficient vectors into a principal component coefficient matrix.

[0074] According to another aspect of the present invention, a vehicle operation fault identification system based on a simulator is also provided, and the vehicle operation fault identification system based on a simulator comprises:

[0075] A fault anomaly database establishment module is used to establish a vehicle operation fault anomaly database and set the corresponding relationship between various parameters in the fault anomaly database;

[0076] The data acquisition and processing module is used to run the car model in the simulator, obtain the normal operation data and fault data of the car, and perform preprocessing to obtain the characteristic data of the car operation fault;

[0077] A data analysis module is used to analyze the characteristic data of vehicle operation failures through an analysis algorithm to obtain key parameters that affect the vehicle operation failures;

[0078] The fault diagnosis and discrimination module is used to diagnose the key parameters that affect the vehicle operation faults through the anomaly detection algorithm, and compare the diagnosis results with the parameters in the fault anomaly database to determine whether they conform to the known fault mode and determine the fault type;

[0079] A model building and evaluation module is used to build a vehicle operation fault prediction model according to the determined fault type, and use the vehicle operation fault prediction model to evaluate the future operation performance of the vehicle in real time;

[0080] Fault repair and decision-making module, used to formulate and implement corresponding repair or optimization measures according to the evaluation results of the vehicle operation fault prediction model;

[0081] Among them, the fault anomaly database establishment module is connected with the data analysis module through the data acquisition and processing module, the data analysis module is connected with the model establishment and evaluation module through the fault diagnosis and discrimination module, and the model establishment and evaluation module is connected with the fault maintenance and decision-making module.

[0082] The beneficial effects of the present invention are:

[0083] 1. The present invention can more accurately identify and diagnose automobile faults by establishing an abnormal database of automobile operation faults and combining simulator data and abnormality detection algorithms. It can also predict potential problems before faults occur by utilizing prediction models, thereby achieving preventive maintenance, avoiding sudden faults, and increasing automobile reliability and safety. At the same time, by identifying problems in advance, large-scale faults can be avoided, maintenance costs and time can be reduced, and the frequency and severity of faults can be reduced, thereby improving the overall performance and reliability of the automobile, avoiding potential faults, and improving the operating efficiency and safety of the automobile.

[0084] 2. The present invention analyzes the characteristic data of automobile operation faults through an analysis algorithm, so that through precise feature extraction and cluster analysis, the key fault parameters affecting the operation of the automobile can be more accurately identified, thereby improving the accuracy of fault diagnosis, and the particle swarm optimization can efficiently process a large amount of data, quickly find the optimal solution or the approximate optimal solution, thereby improving the efficiency of data processing, and by continuously iteratively updating the speed and position of the particles, the algorithm can dynamically adapt to different data sets and fault characteristics, improve the generalization ability of the model, and at the same time, by accurately identifying the cause of the fault and key parameters, it can reduce the diagnosis time, quickly locate the fault, and thus reduce the maintenance time and related costs.

[0085] 3. The present invention diagnoses the key parameters that affect automobile operation failures through an anomaly detection algorithm, so that the anomaly point detection model established by SVM can more accurately distinguish between normal operation and fault status, thereby improving the accuracy of fault diagnosis, and using an optimization algorithm to obtain the best parameters of SVM, which can effectively improve the efficiency and accuracy of model training and ensure that the model has the best performance. At the same time, by comparing the fault detection results with the known fault modes in the fault anomaly database, the fault type can be quickly and accurately located, shortening the fault diagnosis time, and then through accurate models and optimized parameters, misdiagnosis and missed diagnosis can be effectively reduced, thereby improving the reliability of fault detection.

[0086] 4. The present invention establishes a vehicle operation fault prediction model, so that through stratified sampling and principal component analysis, the model can more accurately identify fault characteristics and improve the accuracy of fault diagnosis. It can also analyze the fault characteristic data in vehicle operation in real time and predict potential faults, so as to take maintenance measures in advance, reduce the occurrence of sudden faults, and improve the operation reliability of the vehicle. At the same time, because the model includes the steps of loop processing and model training, it can continuously optimize and update itself through new data, adapt to different operating conditions and fault types, maintain its prediction performance, and thus enable fault prediction to reduce unnecessary inspections and repairs, save maintenance costs and time, and improve maintenance efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0088] Figure 1 is a flow chart of a method for identifying vehicle operation faults based on a simulator according to an embodiment of the present invention;

[0089] Figure 2 The present invention is a block diagram of a simulator-based automobile operation fault identification system according to an embodiment of the present invention.

[0090] In the figure:

[0091] 1. Fault anomaly database establishment module; 2. Data acquisition and processing module; 3. Data analysis module; 4. Fault diagnosis and identification module; 5. Model establishment and evaluation module; 6. Fault maintenance and decision-making module. DETAILED DESCRIPTION

[0092] The technical solutions in the embodiments of the present invention will be described clearly and completely 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.

[0093] In the description of the present invention, unless otherwise specified, the meaning of "plurality" is two or more. In addition, the terms "first", "second", "third", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0094] According to an embodiment of the present invention, a simulator-based automobile operation fault identification method and system are provided.

[0095] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to the vehicle operation fault identification method based on the simulator according to an embodiment of the present invention, the vehicle operation fault identification method based on the simulator includes the following steps:

[0096] S1. Establishing a database of abnormal vehicle operation failures and setting the corresponding relationship between various parameters in the abnormal failure database;

[0097] Specifically, the parameters include engine performance parameters (such as engine temperature, speed, fuel consumption rate, emission level, etc.), electronic system parameters (including battery status, generator output, readings of various sensors, etc.), mechanical component conditions (such as brake system, steering system, transmission system wear or fault records), vehicle dynamic parameters (such as speed, acceleration, steering angle, etc.), historical maintenance records (historical maintenance type, frequency, fault diagnosis results, etc.), driving behavior data (such as frequency of sudden acceleration and sudden braking, etc.) and environmental factors (such as temperature, humidity, road conditions and other external factors that may affect vehicle performance).

[0098] Specifically, the correspondence between parameters refers to the correlation or causal relationship established between these parameters. For example, analyzing the relationship between engine temperature and fuel consumption rate to determine whether there is a cooling system problem. Predicting the possible failure time of certain components through correlation analysis of historical maintenance records and vehicle dynamic parameters. Combining driving behavior data with the condition of mechanical components to evaluate the impact of driving habits on vehicle wear, etc.

[0099] S2. Run the car model in the simulator, obtain the normal operation data and fault data of the car, and perform preprocessing to obtain the car operation fault characteristic data;

[0100] Specifically, a simulator is a software or system that can simulate the operation of a car under various conditions. This simulator can reproduce various situations that a car may encounter in a real environment, including normal operation and various fault conditions.

[0101] Specifically, normal operating data includes engine performance data (such as temperature, speed, fuel consumption, etc.), vehicle dynamics (such as speed, acceleration, steering angle, etc.), electronic system readings (such as battery status, sensor readings, etc.), data under normal environment (such as stable temperature, humidity, etc.), etc.

[0102] Specifically, fault data includes engine fault data (such as overheating, abnormal speed), electronic system fault (such as sensor failure, battery problems, etc.), mechanical component fault (such as brake system problems, transmission system abnormalities, etc.), vehicle dynamic changes under fault conditions (such as speed changes and steering response when responding to faults), etc.

[0103] S3, analyzing the obtained vehicle operation fault characteristic data through an analysis algorithm to obtain key parameters that affect the vehicle operation fault;

[0104] Specifically, key parameters include engine performance indicators (such as engine temperature, speed, fuel consumption rate, etc. These data can indicate the working status of the engine, and abnormal values ​​​​indicate a fault), vehicle dynamic data (such as speed, acceleration, steering response, etc. These data are very important in analyzing the overall dynamic behavior of the vehicle, especially in the event of a mechanical failure), electronic system data (including battery voltage, generator output, sensor fault code, etc. Abnormalities in these parameters indicate that there may be problems with the electronic system), mechanical component status (such as brake wear, tightness of the steering system, etc. Mechanical failures are usually reflected in these parameters), environmental factors (such as external temperature, humidity, road conditions, etc., these factors sometimes affect vehicle performance, especially under extreme conditions) and driving behavior data (such as the driver's acceleration and braking habits, these data help identify potential vehicle problems caused by driving behavior), etc.

[0105] S4. Diagnose the key parameters that affect the vehicle operation fault by using an anomaly detection algorithm, and compare the diagnosis results with the parameters in the fault anomaly database to determine whether they conform to the known fault mode and determine the fault type;

[0106] S5. According to the determined fault type, a vehicle operation fault prediction model is established through a classification algorithm, and the vehicle operation fault prediction model is used to evaluate the future operation performance of the vehicle in real time;

[0107] S6. Develop and implement corresponding maintenance or optimization measures based on the evaluation results of the vehicle operation fault prediction model.

[0108] Preferably, running the automobile model in the simulator, obtaining the normal operation data and fault data of the automobile, and performing preprocessing to obtain the automobile operation fault characteristic data comprises the following steps:

[0109] S21, obtaining duplicate data corresponding to normal operation data and fault data of the vehicle in the simulator, and performing denoising, filtering and smoothing on the duplicate data;

[0110] S22, connecting the unprocessed data rows in the normal operation data and fault data of the vehicle in the simulator to generate a new data table, and associating different data tables through external key values ​​to generate a complete data table, and obtain an accurate data set;

[0111] S23. Extract features from the accurate data set and obtain corresponding vehicle operation fault feature data.

[0112] Preferably, the obtained vehicle operation fault characteristic data is analyzed by an analysis algorithm to obtain key parameters affecting the vehicle operation fault:

[0113] S31, using the t-SNE method to map the obtained automobile operation fault feature data to the feature space, and select a number of feature vectors to form an automobile fault feature point set;

[0114] Specifically, t-SNE (t-Distributed Stochastic Neighbor Embedding) is a powerful machine learning algorithm that is particularly suitable for dimensionality reduction and visualization of high-dimensional data. It can help display the intrinsic structure of data in two-dimensional or three-dimensional space, and is particularly effective in exploring the clustering and distribution of data sets.

[0115] S32, taking a number of cluster centers as initial solutions, setting a number of particle numbers and an optimal number of iterations, and randomly generating a number of initial solutions;

[0116] S33, according to the current cluster center position, the fitness value of each particle is calculated through the fitness function, and the current fitness value is updated as the individual optimal value and the current position is updated as the individual optimal position, and the global optimal value and the global optimal position are found through the individual optimal values ​​of all particles;

[0117] S34, judging whether the set optimal number of iterations is reached, if so, going to step S38; if not, continuing to update the speed and position of the particle;

[0118] S35, according to the updated particle positions, using the minimum distance principle to assign each automobile fault sample in the automobile fault feature point set to a corresponding number of cluster centers;

[0119] Specifically, the working principle of the minimum distance principle is:

[0120] Define the distance metric: First decide on the method used to calculate the distance, usually Euclidean distance, but it could be Manhattan distance, cosine similarity, or another suitable metric.

[0121] Determine cluster centers: At the beginning of the clustering process, determine a set of initial cluster centers.

[0122] Calculate distance and assign data points: For each sample point in the feature point set, calculate its distance from all cluster centers and assign it to the nearest cluster center.

[0123] Update cluster centers: After all data points have been assigned, update the center position of each cluster based on the average position of the data points to which they belong (or other criteria).

[0124] Repeated iteration: Repeat the process of calculating distance and updating cluster centers until certain convergence conditions are met.

[0125] S36, recalculating the fitness value of each particle. If the fitness value of a particle is less than its individual best fitness value, then updating the individual best fitness value of the particle to the current fitness value, and updating its individual best position;

[0126] S37, compare the individual best fitness values ​​of all particles, find the minimum value as the global extreme value, update the global extreme value position, and return to step S34 to continue iteration;

[0127] S38. After the iteration is completed, the final clustering result is used as the key parameter affecting the vehicle operation failure.

[0128] Specifically, clustering refers to identifying and classifying various key parameters that affect vehicle operation failures.

[0129] Specifically, the analysis algorithm is a particle swarm optimization clustering algorithm based on feature analysis. The algorithm is mainly divided into two steps: first, the sample points to be divided are mapped to the feature space using the t-SNE method to form a feature point set, and then the PSO algorithm is used to cluster in the feature space.

[0130] Specifically, the PSO algorithm first initializes a group of random particles (random solutions), and then finds the optimal solution through iteration. In each iteration, the particles update themselves by tracking two extreme values, namely the individual extreme value of the particle itself and the global extreme value.

[0131] Preferably, the calculation formula of the fitness function is:

[0132] ;

[0133] in, S Expressed as the fitness value of the particle;

[0134] T Represented as a set of feature points;

[0135] M Represented as cluster center;

[0136] F Expressed as fitness function;

[0137] Expressed as x The cluster centers and j The distance between sample points.

[0138] Preferably, the key parameters that affect the vehicle operation fault are diagnosed by an abnormality detection algorithm, and the diagnosis results are compared with the parameters in the fault anomaly database to determine whether they conform to the known fault mode and determine the fault type, including the following steps:

[0139] S41, obtaining a key parameter data set that affects vehicle operation failures, and using the key parameter data set as a training sample for the SVM;

[0140] S42, obtaining optimal parameters of SVM parameters through an optimization algorithm;

[0141] Specifically, the optimization algorithm is the ISSA algorithm (Improved Sparrow Search Algorithm), which is an improvement on the traditional sparrow search algorithm (SSA). It solves the problem that the SSA is prone to fall into the local optimal solution in some cases, and improves the efficiency of the algorithm and the quality of the solution. In the traditional sparrow search algorithm, the number of initial sparrows is small. After a certain number of iterations, the sparrows in the group will gradually approach a certain optimal solution. If the position of the sparrows in the last iteration is not ideal, then continuing to update the individual position of the sparrow will affect the final result, making the algorithm prone to fall into the local optimal solution. In response to these problems, the present invention introduces refractive reverse learning, and after each iteration, the candidate solution of the refractive reverse learning is obtained according to the position of the discoverer, and then the fitness value of the discoverer is compared with the fitness value of the candidate solution, and the sparrow with a better fitness value is taken as the discoverer under the current number of iterations, which effectively improves the local search capability, the efficiency of the algorithm and the diversity of the population.

[0142] S43, using the obtained optimal SVM parameters to train the training set, and establishing an outlier detection model for automobile fault diagnosis (i.e., ISSA-SVM outlier detection model);

[0143] S44, inputting key parameters that affect vehicle operation failure into an abnormal point detection model, and performing real-time fault diagnosis;

[0144] S45, comparing the fault results detected by the anomaly detection model with the known fault modes in the fault anomaly database, and confirming whether they are consistent;

[0145] S46. Determine the specific fault type that affects the operation of the vehicle based on the comparison results.

[0146] Specifically, the anomaly detection algorithm is a support vector machine (SVM) algorithm, which is a method for detecting anomalies. However, it is still difficult to efficiently obtain the optimal parameters, resulting in low detection efficiency and poor stability. Therefore, the present invention uses an improved sparrow search algorithm ISSA and uses it to optimize the support vector machine parameters. First, the traditional sparrow search algorithm is improved by refraction reverse learning and variable logarithmic spiral. Then, the support vector machine parameters are optimized using the improved sparrow search algorithm ISSA, and the optimal parameters of the SVM are obtained through the optimal sparrow position. The improved sparrow search algorithm is used to optimize the support vector machine parameters, enhance the self-learning and self-adaptation capabilities of the anomaly detection algorithm, and improve the accuracy of anomaly detection. Finally, the optimized support vector machine is used to detect and diagnose the key parameters that affect the vehicle operation failure.

[0147] Preferably, obtaining the optimal parameters of the SVM parameters by the optimization algorithm comprises the following steps:

[0148] S421, initializing the candidate solution set (sparrow population) and related parameters of the optimization algorithm;

[0149] S422, calculating the fitness value of each candidate solution (each sparrow), selecting the best position and the best fitness value in the current iteration number, and selecting the current worst position and the worst fitness value;

[0150] S423, selecting a candidate solution with the best fitness value from the candidate solution set (in the sparrow population) as an initial discoverer, and updating the initial discoverer position according to the discoverer position update formula;

[0151] Specifically, the discoverer position update formula is:

[0152] ;

[0153] in, Expressed as u The first iteration z ( z =1, 2, …, n ) positions of candidate solutions (sparrows);

[0154] Expressed as the maximum number of iterations;

[0155] α Represented as a random number (0, 1);

[0156] R 2 It is indicated as the warning value;

[0157] Specifically, R 2 ∈[0,1]。

[0158] SU It is indicated as a safe value;

[0159] Specifically, SU ∈[0.5,1]。

[0160] Q Represented as random numbers that follow a standard normal distribution;

[0161] L Represented as elements in a matrix.

[0162] S424, generating a refraction reverse learning candidate solution of the initial discoverer according to the refraction reverse learning solution formula;

[0163] S425, comparing the fitness values ​​of the initial discoverer and its reverse learning candidate solution, and selecting the one with the best fitness value as the new discoverer;

[0164] S426, taking the candidate solutions other than the new discoverer as joiners, and updating the positions of the joiners according to the variable logarithmic spiral formula;

[0165] ;

[0166] in, Expressed as u +The position of the best candidate solution at iteration 1;

[0167] Expressed as u The global worst position at iteration ;

[0168] A Represented as a matrix;

[0169] A + Expressed as the pseudo-inverse of the matrix;

[0170] s Expressed as a scaling factor.

[0171] S427, randomly selecting a candidate solution from new discoverers and joiners as a sentinel, and updating the position of the sentinel according to the sentinel position update formula;

[0172] Specifically, the sentinel position update formula is:

[0173] ;

[0174] in, Represents the current global best position;

[0175] β Represented as a random number that follows a standard normal distribution;

[0176] G It is expressed as the moving direction and step size control parameters of the candidate solution (sparrow);

[0177] Specifically, G is a random number between -1 and 1.

[0178] f z It is represented by the fitness value of the individual of the current candidate solution (sparrow);

[0179] f g Represents the current global optimal fitness value;

[0180] f w Represents the current global worst fitness value.

[0181] S428, determine whether the optimization algorithm has reached the optimal number of iterations, if so, the loop ends, if not, return to step S423;

[0182] S429: Output the global optimal position and the optimal fitness value to obtain the optimal parameters of the SVM parameters.

[0183] Specifically, in the present invention, the discoverer, joiner and guard are key concepts in the Sparrow Search Algorithm (SSA), which respectively represent the sparrows (solutions or individuals) of different roles in the algorithm, and these roles are used to find the optimal parameters of the SVM.

[0184] Specifically, the discoverer: In the process of finding the best parameters for SVM, the discoverer represents the best parameter combination currently found, that is, those parameters that perform best on the validation data.

[0185] Specifically, joiners are the remaining parameter combinations besides the discoverer, which improve the performance by imitating the discoverer or exploring new possibilities.

[0186] Specifically, the sentinel is a sparrow randomly selected from the discoverers and joiners. In this process, the sentinel is responsible for exploring the neglected parameter space to prevent the algorithm from falling into the local optimum.

[0187] Preferably, the refraction reverse learning solution formula is:

[0188] ;

[0189] in, is expressed as an infinitesimal constant;

[0190] Specifically, ∈[0,1], to avoid the denominator being zero.

[0191] u Represents the current iteration number;

[0192] U Expressed as the total number of iterations;

[0193] Indicates the number of iterations in the current z The candidate solution is c The location of the dimension;

[0194] Expressed as Candidate solutions for reverse learning of ;

[0195] a c Expressed as c The upper boundary in the dimensional space;

[0196] b c Expressed as c The lower boundary in dimensional space;

[0197] k It is represented as adjustment parameter;

[0198] ω is represented as a hyperparameter.

[0199] Specifically, the refraction reverse learning solution formula in the present invention is an improved refraction reverse learning solution formula based on the refraction reverse learning solution formula. The candidate solution obtained by refraction reverse learning can effectively make the algorithm jump out of the local optimal solution. However, due to the fitness in the later stage of iteration, the present invention introduces a hyperparameter ω , according to the different number of iterations ω The value is adjusted to increase the randomness of the solution, so that the candidate solution has a better ability to jump out of the local extreme value and can find the optimal solution more efficiently.

[0200] Preferably, according to the determined fault type, establishing a vehicle operation fault prediction model through a classification algorithm, and using the vehicle operation fault prediction model to evaluate the future operation performance of the vehicle in real time includes the following steps:

[0201] S51, inputting an initial sample set of automobile fault characteristic data under different operating conditions;

[0202] S52, classifying all automobile fault feature data in the initial sample set to obtain a number of sample subsets with different key features;

[0203] S53, performing loop processing on each sample subset, and obtaining a coefficient matrix (principal component coefficient matrix) related to the automobile fault type through analysis (principal component analysis);

[0204] S54, rearranging the coefficient matrix, and transforming the initial sample set based on the coefficient matrix (principal component coefficient matrix) to generate a new automobile fault feature data sample set;

[0205] S55, training the vehicle operation fault prediction model based on the new vehicle fault feature data sample set to obtain a sub-model of the vehicle operation fault prediction model, returning to step S52, and looping a preset number of times to obtain an integrated vehicle operation fault prediction model;

[0206] S56, acquiring real-time automobile fault characteristic data during automobile operation, and inputting the real-time automobile fault characteristic data into an integrated automobile operation fault prediction model;

[0207] S57. Analyze and predict the real-time automobile fault characteristic data through the automobile operation fault prediction model, and obtain an evaluation of the future operation performance of the automobile.

[0208] Specifically, the classification algorithm is a heterogeneous multi-classifier integration algorithm. First, the rotation forest is used to transform and divide the original sample set to obtain a new sample set. Then, a support vector machine with high classification accuracy or a kernel matching pursuit with fast classification speed is selected as the basic integrated individual classifier through a specific proportion, and the new sample set is classified to obtain its prediction label. Finally, the prediction labels under the two models are combined. By combining two different classifier models, the algorithm achieves the complementarity of accuracy and speed. After the mixed integration of the two, the generalization error of the integrated system is improved.

[0209] Preferably, performing a loop process on each sample subset to obtain a coefficient matrix related to the vehicle fault type through analysis includes the following steps:

[0210] S531, resampling the automobile fault feature data sample subset by using a stratified sampling method to generate a new automobile fault feature data sample subset;

[0211] Specifically, the stratified sampling method is used to resample the subset of automobile fault feature data samples. When extracting samples, the data is divided into different layers or subsets, and each layer represents a specific subgroup in the data. The purpose of this method is to ensure that the sample is more representative statistically, because it can ensure that data from each important subgroup in the data set are included in the sample.

[0212] S532, performing principal component analysis on a new subset of automobile fault feature data samples, and obtaining a principal component coefficient vector related to the automobile fault type;

[0213] S533, repeatedly executing step S531 to step S532, processing different test feature data sample subsets, obtaining several groups of principal component coefficient vectors, and merging the principal component coefficient vectors into a principal component coefficient matrix.

[0214] According to another embodiment of the present invention, Figure 2 As shown, a vehicle operation fault identification system based on a simulator is also provided, and the vehicle operation fault identification system based on a simulator includes:

[0215] The fault abnormality database establishment module 1 is used to establish a vehicle operation fault abnormality database and set the corresponding relationship between various parameters in the fault abnormality database;

[0216] The data acquisition and processing module 2 is used to run the car model in the simulator, obtain the normal operation data and fault data of the car, and perform preprocessing to obtain the characteristic data of the car operation fault;

[0217] The data analysis module 3 is used to analyze the obtained vehicle operation fault characteristic data through an analysis algorithm to obtain key parameters that affect the vehicle operation fault;

[0218] The fault diagnosis and identification module 4 is used to diagnose the key parameters that affect the vehicle operation fault through an abnormality detection algorithm, and compare the diagnosis results with the parameters in the fault anomaly database to determine whether they meet the known fault mode and determine the fault type;

[0219] The model building and evaluation module 5 is used to build a vehicle operation fault prediction model according to the determined fault type, and use the vehicle operation fault prediction model to evaluate the future operation performance of the vehicle in real time;

[0220] Fault repair and decision module 6, used to formulate and implement corresponding repair or optimization measures according to the evaluation results of the vehicle operation fault prediction model;

[0221] Among them, the fault abnormality database establishment module 1 is connected to the data analysis module 3 through the data acquisition and processing module 2, the data analysis module 3 is connected to the model establishment and evaluation module 5 through the fault diagnosis and discrimination module 4, and the model establishment and evaluation module 5 is connected to the fault maintenance and decision-making module 6.

[0222] In summary, with the aid of the above-mentioned technical scheme of the present invention, the present invention analyzes the obtained automobile operation fault characteristic data through an analysis algorithm, so that through precise feature extraction and cluster analysis, the key fault parameters affecting the operation of the automobile can be more accurately identified, thereby improving the accuracy of fault diagnosis, and the particle swarm optimization can be used to efficiently process a large amount of data, quickly find the optimal solution or the approximate optimal solution, thereby improving the efficiency of data processing, and by continuously iteratively updating the speed and position of the particles, the algorithm can dynamically adapt to different data sets and fault characteristics, and improve the generalization ability of the model, and at the same time, by accurately identifying the cause of the fault and key parameters, the diagnosis time can be reduced, the fault can be quickly located, and the maintenance time and related costs can be reduced. The present invention diagnoses the key parameters that affect automobile operation failures through an anomaly detection algorithm, so that the anomaly detection model established by SVM can more accurately distinguish between normal operation and fault status, thereby improving the accuracy of fault diagnosis, and uses an optimization algorithm to obtain the best parameters of SVM, which can effectively improve the efficiency and accuracy of model training and ensure that the model has the best performance. At the same time, by comparing the fault detection results with the known fault modes in the fault anomaly database, the fault type can be quickly and accurately located, shortening the fault diagnosis time, and then through accurate models and optimized parameters, misdiagnosis and missed diagnosis can be effectively reduced, and the reliability of fault detection can be improved. The present invention establishes an automobile operation fault prediction model, so that through stratified sampling and principal component analysis, the model can more accurately identify fault characteristics and improve the accuracy of fault diagnosis, and can analyze the fault characteristic data in automobile operation in real time to predict potential faults, so as to take maintenance measures in advance, reduce the occurrence of sudden faults, and improve the operation reliability of the automobile. At the same time, because the model includes the steps of loop processing and model training, it can continuously optimize and update itself through new data, adapt to different operating conditions and fault types, maintain its prediction performance, and thus enable fault prediction to reduce unnecessary inspections and repairs, save maintenance costs and time, and improve maintenance efficiency.

[0223] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A vehicle operation fault identification method based on a simulator, characterized in that: The vehicle operation fault identification method based on the simulator comprises the following steps: S1. Establishing a database of abnormal vehicle operation failures and setting the corresponding relationship between various parameters in the abnormal failure database; S2. Run the car model in the simulator, obtain the normal operation data and fault data of the car, and perform preprocessing to obtain the car operation fault characteristic data; S3, analyzing the obtained vehicle operation fault characteristic data through an analysis algorithm to obtain key parameters that affect the vehicle operation fault; S4. Diagnose the key parameters that affect the vehicle operation fault by using an anomaly detection algorithm, and compare the diagnosis results with the parameters in the fault anomaly database to determine whether they conform to the known fault mode and determine the fault type; S5. According to the determined fault type, a vehicle operation fault prediction model is established through a classification algorithm, and the vehicle operation fault prediction model is used to evaluate the future operation performance of the vehicle in real time; S6. Formulate and implement corresponding maintenance or optimization measures based on the evaluation results of the vehicle operation fault prediction model; The method of diagnosing the key parameters that affect the vehicle operation fault obtained by the anomaly detection algorithm, and comparing the diagnosis results with the parameters in the fault anomaly database to determine whether they conform to the known fault mode and determine the fault type includes the following steps: S41, obtaining a key parameter data set that affects vehicle operation failures, and using the key parameter data set as a training sample for the SVM; S42, obtaining optimal parameters of SVM parameters through an optimization algorithm; S43, using the obtained optimal SVM parameters to train the training set, and establishing an abnormal point detection model for automobile fault diagnosis; S44, inputting key parameters that affect vehicle operation failure into an abnormal point detection model, and performing real-time fault diagnosis; S45, comparing the fault results detected by the anomaly detection model with the known fault modes in the fault anomaly database, and confirming whether they are consistent; S46. Determine the specific fault type that affects the operation of the vehicle based on the comparison result; The method for obtaining the optimal parameters of the SVM parameters by the optimization algorithm comprises the following steps: S421, initializing the candidate solution set and related parameters of the optimization algorithm; S422, calculating the fitness value of each candidate solution, selecting the best position and the best fitness value in the current iteration number, and selecting the current worst position and the worst fitness value; S423, selecting a candidate solution with the best fitness value from the candidate solution set as the initial finder, and updating the initial finder position according to the finder position update formula; S424, generating a refraction reverse learning candidate solution of the initial discoverer according to the refraction reverse learning solution formula; S425, comparing the fitness values ​​of the initial discoverer and its reverse learning candidate solution, and selecting the one with the best fitness value as the new discoverer; S426, taking the candidate solutions other than the new discoverer as joiners, and updating the positions of the joiners according to the variable logarithmic spiral formula; S427, randomly selecting a candidate solution from new discoverers and joiners as a sentinel, and updating the position of the sentinel according to the sentinel position update formula; S428, determine whether the optimization algorithm has reached the optimal number of iterations, if so, the loop ends, if not, return to step S423; S429: Output the global optimal position and the optimal fitness value to obtain the optimal parameters of the SVM parameters.

2. The method for identifying vehicle operation faults based on a simulator according to claim 1, characterized in that: The step of running the automobile model in the simulator, obtaining normal operation data and fault data of the automobile, and performing preprocessing to obtain the characteristic data of automobile operation faults comprises the following steps: S21, obtaining duplicate data corresponding to normal operation data and fault data of the vehicle in the simulator, and performing denoising, filtering and smoothing on the duplicate data; S22, connecting the unprocessed data rows in the normal operation data and fault data of the vehicle in the simulator to generate a new data table, and associating different data tables through external key values ​​to generate a complete data table, and obtain an accurate data set; S23. Extract features from the accurate data set and obtain corresponding vehicle operation fault feature data.

3. The method for identifying vehicle operation faults based on a simulator according to claim 1, characterized in that: The analysis algorithm is used to analyze the characteristic data of the vehicle operation failure to obtain the key parameters that affect the vehicle operation failure: S31, using the t-SNE method to map the obtained automobile operation fault feature data to the feature space, and select a number of feature vectors to form an automobile fault feature point set; S32, taking a number of cluster centers as initial solutions, setting a number of particle numbers and an optimal number of iterations, and randomly generating a number of initial solutions; S33, according to the current cluster center position, the fitness value of each particle is calculated through the fitness function, and the current fitness value is updated as the individual optimal value and the current position is updated as the individual optimal position, and the global optimal value and the global optimal position are found through the individual optimal values ​​of all particles; S34, judging whether the set optimal number of iterations is reached, if so, going to step S38; if not, continuing to update the speed and position of the particle; S35, according to the updated particle positions, using the minimum distance principle to assign each automobile fault sample in the automobile fault feature point set to a corresponding number of cluster centers; S36, recalculating the fitness value of each particle. If the fitness value of a particle is less than its individual best fitness value, then updating the individual best fitness value of the particle to the current fitness value, and updating its individual best position; S37, compare the individual best fitness values ​​of all particles, find the minimum value as the global extreme value, update the global extreme value position, and return to step S34 to continue iteration; S38. After the iteration is completed, the final clustering result is used as the key parameter affecting the vehicle operation failure.

4. The method for identifying vehicle operation faults based on a simulator according to claim 3, characterized in that: The calculation formula of the fitness function is: ; in, S Expressed as the fitness value of the particle; T Represented as a set of feature points; M Represented as cluster center; F Expressed as fitness function; Expressed as x The cluster centers and j The distance between sample points.

5. The method for identifying vehicle operation faults based on a simulator according to claim 1, characterized in that: The refraction reverse learning solution formula is: ; in, is expressed as an infinitesimal constant; u Represents the current iteration number; U Expressed as the total number of iterations; Indicates the number of iterations in the current z The original discoverer was c The location of the dimension; Expressed as Candidate solutions for reverse learning of ; a c Expressed as c The upper boundary in the dimensional space; b c Expressed as c The lower boundary in dimensional space; k Indicates adjustment parameters; ω is represented as a hyperparameter.

6. The method for identifying vehicle operation faults based on a simulator according to claim 1, characterized in that: The method of establishing a vehicle operation fault prediction model by a classification algorithm according to the determined fault type, and using the vehicle operation fault prediction model to evaluate the future operation performance of the vehicle in real time includes the following steps: S51, inputting an initial sample set of automobile fault characteristic data under different operating conditions; S52, classifying all automobile fault feature data in the initial sample set to obtain a number of sample subsets with different key features; S53, performing loop processing on each sample subset, and obtaining a coefficient matrix related to the vehicle fault type through analysis; S54, rearranging the coefficient matrix, and transforming the initial sample set based on the coefficient matrix to generate a new automobile fault feature data sample set; S55, training the vehicle operation fault prediction model based on the new vehicle fault feature data sample set to obtain a sub-model of the vehicle operation fault prediction model, returning to step S52, and looping a preset number of times to obtain an integrated vehicle operation fault prediction model; S56, acquiring real-time automobile fault characteristic data during automobile operation, and inputting the real-time automobile fault characteristic data into an integrated automobile operation fault prediction model; S57. Analyze and predict the real-time automobile fault characteristic data through the automobile operation fault prediction model, and obtain an evaluation of the future operation performance of the automobile.

7. The method for identifying vehicle operation faults based on a simulator according to claim 6, characterized in that: The loop processing is performed on each sample subset to obtain a coefficient matrix related to the vehicle fault type through analysis, comprising the following steps: S531, resampling the automobile fault feature data sample subset by using a stratified sampling method to generate a new automobile fault feature data sample subset; S532, performing principal component analysis on a new subset of automobile fault feature data samples, and obtaining a principal component coefficient vector related to the automobile fault type; S533, repeatedly executing step S531 to step S532, processing different test feature data sample subsets, obtaining several groups of principal component coefficient vectors, and merging the principal component coefficient vectors into a principal component coefficient matrix.

8. A simulator-based vehicle operation fault identification system, used to implement the simulator-based vehicle operation fault identification method according to any one of claims 1 to 7, characterized in that: The simulator-based vehicle operation fault identification system includes: A fault anomaly database establishment module is used to establish a vehicle operation fault anomaly database and set the corresponding relationship between various parameters in the fault anomaly database; The data acquisition and processing module is used to run the car model in the simulator, obtain the normal operation data and fault data of the car, and perform preprocessing to obtain the characteristic data of the car operation fault; A data analysis module is used to analyze the characteristic data of vehicle operation failures through an analysis algorithm to obtain key parameters that affect the vehicle operation failures; The fault diagnosis and discrimination module is used to diagnose the key parameters that affect the vehicle operation faults through the anomaly detection algorithm, and compare the diagnosis results with the parameters in the fault anomaly database to determine whether they conform to the known fault mode and determine the fault type; A model building and evaluation module is used to build a vehicle operation fault prediction model according to the determined fault type, and use the vehicle operation fault prediction model to evaluate the future operation performance of the vehicle in real time; Fault repair and decision-making module, used to formulate and implement corresponding repair or optimization measures according to the evaluation results of the vehicle operation fault prediction model; Among them, the fault anomaly database establishment module is connected with the data analysis module through the data acquisition and processing module, the data analysis module is connected with the model establishment and evaluation module through the fault diagnosis and discrimination module, and the model establishment and evaluation module is connected with the fault maintenance and decision-making module.

Citation Information

Patent Citations

  • Pedestrian target identification method

    CN114973329A

  • GIS fault mode identification method based on SSA-SVM

    CN115407140A

  • District user-transformer relation identification method based on multivariate analysis model

    CN117764463A

  • Remote monitoring and fault prediction system for water chiller

    CN118194150A

  • Digestive tract hemorrhage risk assessment method and system based on neural network

    CN118280577A