New energy automobile battery fault detection method

By combining traditional battery detection tools and random forest machine learning algorithms, efficient and accurate detection of battery failures in new energy vehicles is achieved, and the problem of insufficient relying on manual experience and data analysis capabilities in traditional methods is solved, reducing maintenance costs and extending the battery life.

CN119024168AInactive Publication Date: 2024-11-26四川吉利学院
View PDF 0 Cites 3 Cited by

Patent Information

Application Number
CN202411134434.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-19
Publication Date
2024-11-26
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Traditional new energy vehicle battery failure detection methods rely on manual experience, and lack subjectivity and data analysis capabilities, resulting in high maintenance costs and may only be maintained after a failure occurs, affecting battery performance.

Method used

The multimeter, battery internal resistance tester, infrared thermal imager, insulation resistance tester, BMS diagnostic instrument and other tools are used to combine random forest machine learning algorithms to detect and analyze battery faults, predict the fault types in real time and perform corresponding repairs or replacements.

Benefits of technology

It significantly improves the efficiency and accuracy of fault detection, reduces human error, and can promptly detect and warn in the early stages of faults, reduce maintenance costs, extend battery life, and improve the reliability and economicality of new energy vehicles.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119024168A_ABST
    Figure CN119024168A_ABST
Patent Text Reader

Abstract

The invention discloses a new energy automobile battery fault detection method, belongs to the technical field of new energy battery detection, and remarkably improves the fault detection efficiency and accuracy by combining the traditional battery detection steps with a machine learning algorithm. In the step S7, machine learning algorithms such as random forest and the like are used for deeply analyzing the charging and discharging performance data of the battery, and the abnormal behavior and the potential fault mode of the single battery can be quickly identified. Compared with the traditional judgment only depending on artificial experience, the machine learning model can process a large amount of data and learn complex modes and relationships from the data, so that the battery fault can be predicted more accurately. By means of the combination mode, personal errors are reduced, faults can be found and early warned in time at the initial stage, and therefore more serious damage and potential safety accidents are avoided.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of new energy battery detection, and specifically relates to a new energy vehicle battery fault detection method. Background Art

[0002] New energy vehicle batteries are the core components of new energy vehicles, and their performance is directly related to the vehicle's cruising range, safety and service life. New energy vehicle batteries mainly use lithium-ion battery technology and have the following characteristics: High energy density: The energy stored per unit volume or weight of the battery is high, providing new energy vehicles with longer endurance. Long cycle life: The number of charge and discharge cycles of lithium-ion batteries can reach thousands of times, ensuring the service life of the battery. High charging efficiency: The battery charges quickly, and some models can be charged to more than 80% of the power in half an hour. Environmental protection and energy saving: New energy vehicle batteries are pollution-free and in line with the concept of green travel. New energy vehicle battery fault detection is an important means to ensure battery performance and safety, mainly including the following: Voltage detection: By detecting the voltage of the battery cell, it is determined whether the battery has overcharge, over-discharge and other faults. Temperature detection: Real-time monitoring of battery temperature to prevent battery overheating and ensure safe use. Internal resistance detection: By measuring the internal resistance of the battery, the degree of battery aging and performance status are evaluated. Insulation detection: Detect the insulation performance of the battery to prevent leakage accidents. Fault diagnosis and analysis: Use advanced diagnostic algorithms to quickly locate and classify battery faults.

[0003] However, traditional methods often rely on the experience and expertise of technicians, which may lead to subjective judgment of battery failure and inability to process large amounts of data for analysis. At the same time, maintenance may be performed only after a failure occurs, resulting in high maintenance costs and may cause rapid degradation of battery performance. Summary of the invention

[0004] The purpose of the present invention is to provide a new energy vehicle battery fault detection method in order to solve the above-mentioned problem.

[0005] The technical solution adopted by the present invention is as follows: a method for detecting battery failure of a new energy vehicle, the method comprising the following steps:

[0006] S1: Prepare for battery testing first, and prepare multimeter, battery internal resistance tester, infrared thermal imager, insulation resistance tester, battery charge and discharge tester, BMS diagnostic tool and equipment;

[0007] S2: Check the appearance of the battery to ensure that there is no damage, deformation or corrosion on the outside of the battery;

[0008] S3: After disconnecting the battery pack from the vehicle, perform a battery cell voltage test;

[0009] S4: Detect the internal resistance of the battery using a battery internal resistance tester;

[0010] S5: Use an infrared thermal imager to detect battery temperature;

[0011] S6: Use an insulation resistance tester to test the insulation resistance;

[0012] S7: Perform battery charge and discharge performance testing, combine with the random forest machine learning algorithm, analyze the model prediction results, determine whether the battery has a fault and its type, and perform corresponding repairs or replacements according to the fault type indicated by the model;

[0013] S8: Use the BMS diagnostic instrument to read the battery management system data. Check the working status of the BMS system to ensure that all parameters are normal. Analyze the BMS system fault code and find potential fault causes;

[0014] S9: Based on the test results, the faulty battery cells are evenly charged, repaired or replaced, the battery pack cooling system and insulation materials are inspected and maintained, damaged connectors and plug-in components are replaced, the battery pack performance is retested, and the fault is confirmed to be eliminated, after which the entire new energy vehicle battery fault detection process is completed.

[0015] In a preferred embodiment, in step S1, the detection environment must ensure that the detection site is well ventilated and avoid fire and high temperature safety hazards to prevent accidents during the detection process;

[0016] In step S2, it is also necessary to check whether the connectors and plug-in components of the battery pack are firm, loose, detached or damaged. The stability of the fixing bracket is also important, and it is necessary to ensure that there is no looseness or deformation, because these problems may cause the battery to shift during vehicle operation, thereby causing more serious failures.

[0017] In a preferred embodiment, in step S3, a multimeter is used to measure the voltage of each battery cell one by one, and the data is recorded and compared with the standard voltage range (usually 2.8V-4.2V). If the voltage is abnormal, the faulty cell number should be recorded immediately. At the same time, the voltage of each cell is compared. If the voltage difference exceeds 0.3V, it indicates that the voltage of the battery cells is unbalanced and balanced charging is required to restore battery performance.

[0018] In a preferred embodiment, in step S4, when measuring, it is necessary to ensure that the internal resistance of each battery cell is within the normal range. If an abnormal internal resistance is detected, the faulty cell number should be recorded. In addition, the internal resistance between cells is compared. If the internal resistance difference exceeds 10mΩ, it indicates that the internal resistance between battery cells is unbalanced and balanced charging is required.

[0019] In a preferred embodiment, in step S5, if there is a local overheating phenomenon, the fault location should be recorded. The reasons for the abnormal temperature may include poor heat dissipation, battery cell aging or internal short circuit, all of which may lead to battery performance degradation or safety hazards;

[0020] In step S6, an insulation resistance tester is used to measure the insulation resistance of the battery pack to ensure that it meets national standards. If the insulation resistance is detected to be low, the fault location should be recorded and the cause should be analyzed.

[0021] In a preferred embodiment, the step S7 specifically includes the following steps:

[0022] Step 1: Data collection: During the battery charge and discharge test, a large amount of battery performance data is collected, including voltage, current, temperature, internal resistance, charge / discharge time, capacity, and power. For known faulty batteries, their performance data is collected and marked as faulty data.

[0023] Step 2: Data preprocessing: Data cleaning: Remove outliers, missing values, and irrelevant data. Feature engineering: Extract features related to battery performance.

[0024] Step 3: Model training, select the random forest machine learning algorithm to divide the preprocessed data into a training set and a validation set, and then use the training set data to train the machine learning model.

[0025] Step 4: Model validation and optimization, use validation set data to evaluate model performance and adjust model parameters. Perform cross validation to ensure the generalization ability of the model. Based on the validation results, optimize the model structure and parameters.

[0026] Step 5: Fault detection: The real-time collected battery performance data is input into the trained machine learning model. The model will output the prediction result of whether the battery is normal.

[0027] Threshold setting: Set a threshold based on the probability or decision score output by the model to determine whether the battery is faulty.

[0028] Step 6: Analyze the model prediction results to determine whether the battery has a fault and its type. Perform corresponding repairs or replacements based on the fault type indicated by the model.

[0029] In a preferred embodiment, the data cleaning is performed using the z-score method, and the calculation formula is:

[0030] z=(X-μ) / σ, where X is the observed value, μ is the mean, σ is the standard deviation, and data with |z|>3 are removed.

[0031] In a preferred embodiment, the model training selects random forest, and its decision function is:

[0032] Where wi is the weight of the i-th tree, and hi(x) is the prediction result of the i-th tree.

[0033] Divide the training set Dtrain = {(Xi, Yi)}, where Xi is the feature vector and Yi is the label;

[0034] The mean square error optimization algorithm is used to minimize the loss function during model training. The calculation formula is:

[0035]

[0036] Where Y^i is the model prediction value.

[0037] In a preferred embodiment, the model validation uses a validation set Dval={(Xi, Yi)} to calculate the loss or accuracy on the validation set; the cross-validation of the model uses k-fold cross-validation to evaluate the model stability: the data set is divided into k subsets, k-1 subsets are used for training each time, and the remaining subset is used for validation;

[0038] The fault detection real-time data X real-time Input into the model, the formula for model output prediction is: Then set the threshold θ, if Y ^real-time >θ, it is judged as a fault.

[0039] In a preferred embodiment, in step S8, using the BMS diagnostic instrument generally follows the following steps:

[0040] Connection: Connect the BMS diagnostic tool to the vehicle's OBD port or BMS communication interface.

[0041] Power on and initialization: Turn on the diagnostic instrument, wait for it to initialize and establish a communication connection with the BMS.

[0042] Read data: Select the corresponding function to read the real-time data or historical fault codes of the BMS.

[0043] Analyze data: Analyze the status and potential problems of the battery system based on the read data.

[0044] Perform actions: Perform necessary actions based on the diagnosis results.

[0045] Recording and reporting: Record the diagnostic process and results, and generate reports for subsequent analysis and archiving.

[0046] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0047] 1. In the present invention, the efficiency and accuracy of fault detection are significantly improved by combining the traditional battery detection steps with machine learning algorithms. In step S7, the battery charge and discharge performance data is deeply analyzed using machine learning algorithms such as random forests, which can quickly identify abnormal behaviors and potential failure modes of battery cells. Compared with traditional judgments that rely solely on manual experience, machine learning models can process large amounts of data and learn complex patterns and relationships from them, thereby more accurately predicting battery failures. This combination not only reduces human errors, but also can detect and warn in time at the early stages of a fault, thereby avoiding more serious damage and potential safety accidents.

[0048] 2. In the present invention, through the data preprocessing step, the z-score method is used to clean the data and feature engineering is used to extract key indicators, thereby ensuring the quality of the data input into the machine learning model. High-quality data input is the key to ensuring the accuracy of model prediction, and the effective processing of this link by the method further improves the accuracy of detection. Cross-validation and parameter optimization during model training and verification ensure that the model has good generalization ability and robustness, enabling it to maintain a high detection accuracy under different working conditions.

[0049] 3. In the present invention, the use of the BMS diagnostic instrument provides comprehensive monitoring and diagnosis for the battery management system, ensuring that the battery operates in the best condition. Through firmware updates and system calibration, the BMS diagnostic instrument can further improve the performance of the battery management system and optimize the battery charging and discharging strategy, thereby extending the battery life. Compared with post-repair, this preventive maintenance measure can more effectively reduce long-term maintenance costs. In summary, this method not only improves detection efficiency and accuracy, but also reduces maintenance costs and extends battery life, which is of great significance to the reliability and economy of new energy vehicles. It provides an efficient and intelligent solution for battery maintenance of new energy vehicles, which helps to promote the healthy development of the new energy vehicle industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a schematic diagram of the process principle of the present invention. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0052] Reference Figure 1 ,

[0053] A method for detecting battery failure of a new energy vehicle comprises the following steps:

[0054] S1: Prepare for battery testing first, and prepare multimeter, battery internal resistance tester, infrared thermal imager, insulation resistance tester, battery charge and discharge tester, BMS diagnostic tool and equipment;

[0055] S2: Check the appearance of the battery to ensure that there is no damage, deformation or corrosion on the outside of the battery;

[0056] S3: After disconnecting the battery pack from the vehicle, perform a battery cell voltage test;

[0057] S4: Detect the battery internal resistance by using a battery internal resistance tester;

[0058] S5: Use an infrared thermal imager to detect battery temperature;

[0059] S6: Use an insulation resistance tester to test the insulation resistance;

[0060] S7: Perform battery charge and discharge performance testing, combine with the random forest machine learning algorithm, analyze the model prediction results, determine whether the battery has a fault and its type, and perform corresponding repairs or replacements according to the fault type indicated by the model;

[0061] S8: Use the BMS diagnostic instrument to read the battery management system data. Check the working status of the BMS system to ensure that all parameters are normal. Analyze the BMS system fault code and find potential fault causes;

[0062] S9: Based on the test results, the faulty battery cells are evenly charged, repaired or replaced, the battery pack cooling system, insulation materials, etc. are inspected and maintained, damaged connectors, plug-ins and other components are replaced, the battery pack performance is retested, and the fault is confirmed to be eliminated, after which the entire new energy vehicle battery fault detection process is completed.

[0063] In step S1, the testing environment must ensure that the testing site is well ventilated and avoid potential safety hazards such as fire and high temperature to prevent accidents during the testing process;

[0064] In step S2, it is also necessary to check whether the connectors, plug-ins and other components of the battery pack are firm, loose, detached or damaged. The stability of the fixing bracket is also important, and it is necessary to ensure that there is no looseness or deformation, because these problems may cause the battery to shift during vehicle operation, which may lead to more serious failures.

[0065] In step S3, use a multimeter to measure the voltage of each battery cell one by one, record the data and compare it with the standard voltage range (usually 2.8V-4.2V). If the voltage is abnormal, the faulty cell number should be recorded immediately. At the same time, compare the voltage of each cell. If the voltage difference exceeds 0.3V, it indicates that the battery cell voltage is unbalanced and balanced charging is required to restore battery performance;

[0066] In step S4, when measuring, it is necessary to ensure that the internal resistance of each battery cell is within the normal range (generally 20mΩ-80mΩ). If an abnormal internal resistance is detected, the faulty cell number should be recorded. In addition, the internal resistance between cells is compared. If the internal resistance difference exceeds 10mΩ, it indicates that the internal resistance between battery cells is unbalanced and balanced charging is required.

[0067] In step S5, if there is local overheating, the fault location should be recorded. The causes of abnormal temperature should be analyzed, which may include poor heat dissipation, battery cell aging or internal short circuit, etc. These factors may lead to battery performance degradation or safety hazards;

[0068] In step S6, an insulation resistance tester is used to measure the insulation resistance of the battery pack to ensure that it meets the national standard (generally above 500Ω / V). If the insulation resistance is detected to be low, the fault location should be recorded and the cause should be analyzed, such as water ingress into the battery pack, aging or damage of the insulation material, etc., which may cause battery leakage or short circuit.

[0069] Step S7 specifically includes the following steps:

[0070] Step 1: Data collection: During the battery charge and discharge test, a large amount of battery performance data is collected, including voltage, current, temperature, internal resistance, charge / discharge time, capacity, power, etc. For known faulty batteries, their performance data is collected and marked as faulty data.

[0071] Step 2: Data preprocessing: Data cleaning: Remove outliers, missing values, and irrelevant data. Feature engineering: Extract features related to battery performance, such as time series features, statistical features, etc.

[0072] Step 3: Model training, select the random forest machine learning algorithm to divide the preprocessed data into a training set and a validation set, and then use the training set data to train the machine learning model.

[0073] Step 4: Model validation and optimization, use validation set data to evaluate model performance and adjust model parameters. Perform cross validation to ensure the generalization ability of the model. Based on the validation results, optimize the model structure and parameters.

[0074] Step 5: Fault detection: The real-time collected battery performance data is input into the trained machine learning model. The model will output the prediction result of whether the battery is normal.

[0075] Threshold setting: Set a threshold based on the probability or decision score output by the model to determine whether the battery is faulty.

[0076] Step 6: Analyze the model prediction results to determine whether the battery has a fault and its type. Perform corresponding repairs or replacements based on the fault type indicated by the model.

[0077] Data cleaning: Use the z-score method for cleaning, and the calculation formula is:

[0078] z=(X-μ) / σ, where X is the observed value, μ is the mean, σ is the standard deviation, and data with |z|>3 are removed.

[0079] The model training selects random forest, and its decision function is:

[0080] Where wi is the weight of the i-th tree, and hi(x) is the prediction result of the i-th tree.

[0081] Divide the training set Dtrain = {(Xi, Yi)}, where Xi is the feature vector and Yi is the label;

[0082] The mean square error optimization algorithm is used to minimize the loss function during model training. The calculation formula is:

[0083]

[0084] Where Y^i is the model prediction value

[0085] The model validation uses the validation set Dval = {(Xi, Yi)} to calculate the loss or accuracy on the validation set; the cross-validation of the model uses k-fold cross-validation to evaluate the model stability: the data set is divided into k subsets, k-1 subsets are used for training each time, and the remaining subset is used for validation;

[0086] Fault Detection Real-Time DataX real-time Input into the model, the formula for model output prediction is: Then set the threshold θ, if Y ^real-time >θ, it is judged as a fault.

[0087] In step S8, the BMS diagnostic tool is used to perform the following steps:

[0088] Connection: Connect the BMS diagnostic tool to the vehicle's OBD port or BMS communication interface.

[0089] Power on and initialization: Turn on the diagnostic instrument, wait for it to initialize and establish a communication connection with the BMS.

[0090] Read data: Select the corresponding function to read the real-time data or historical fault codes of the BMS.

[0091] Analyze data: Analyze the status and potential problems of the battery system based on the read data.

[0092] Execute Action: Perform necessary actions based on the diagnosis results, such as equalization charging, system calibration, or firmware update.

[0093] Recording and reporting: Record the diagnostic process and results, and generate reports for subsequent analysis and archiving.

[0094] In the present invention, the efficiency and accuracy of fault detection are significantly improved by combining the traditional battery detection steps with machine learning algorithms. In step S7, the battery charge and discharge performance data is deeply analyzed using machine learning algorithms such as random forests, which can quickly identify abnormal behaviors and potential failure modes of battery cells. Compared with traditional judgments that rely solely on manual experience, machine learning models can process large amounts of data and learn complex patterns and relationships from them, thereby more accurately predicting battery failures. This combination not only reduces human errors, but also can detect and warn in a timely manner at the early stages of a fault, thereby avoiding more serious damage and potential safety accidents.

[0095] In the present invention, the quality of data input into the machine learning model is ensured through data preprocessing steps, such as the z-score method to clean data and feature engineering to extract key indicators. High-quality data input is the key to ensuring the accuracy of model prediction, and the effective processing of this link in this method further improves the accuracy of detection. Cross-validation and parameter optimization during model training and verification ensure that the model has good generalization ability and robustness, enabling it to maintain a high detection accuracy under different working conditions.

[0096] In the present invention, the use of the BMS diagnostic instrument provides comprehensive monitoring and diagnosis for the battery management system, ensuring that the battery operates in the best condition. Through firmware updates and system calibration, the BMS diagnostic instrument can further improve the performance of the battery management system and optimize the battery charging and discharging strategy, thereby extending the service life of the battery. This preventive maintenance measure can more effectively reduce long-term maintenance costs compared to post-repairs. In summary, this method not only improves detection efficiency and accuracy, but also reduces maintenance costs and extends battery life, which is of great significance to the reliability and economy of new energy vehicles. It provides an efficient and intelligent solution for battery maintenance of new energy vehicles, which helps to promote the healthy development of the new energy vehicle industry.

[0097] 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 term "comprises" or any other variant thereof is 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. In the absence of further restrictions, the elements defined by the sentence "comprises a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0098] The above description enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A new energy vehicle battery fault detection method, characterized in that: The method comprises the following steps: S1: Prepare for battery testing first, and prepare multimeter, battery internal resistance tester, infrared thermal imager, insulation resistance tester, battery charge and discharge tester, BMS diagnostic tool and equipment; S2: Check the appearance of the battery to ensure that there is no damage, deformation or corrosion on the outside of the battery; S3: After disconnecting the battery pack from the vehicle, perform battery cell voltage detection; S4: Detect the internal resistance of the battery using a battery internal resistance tester; S5: Use an infrared thermal imager to detect battery temperature; S6: Use an insulation resistance tester to test the insulation resistance; S7: Perform battery charge and discharge performance testing, combine with the random forest machine learning algorithm, analyze the model prediction results, determine whether the battery has a fault and its type, and perform corresponding repairs or replacements according to the fault type indicated by the model; S8: Use the BMS diagnostic instrument to read the battery management system data; check the working status of the BMS system to ensure that all parameters are normal; analyze the BMS system fault code and find potential fault causes; S9: Based on the test results, the faulty battery cells are evenly charged, repaired or replaced, the battery pack cooling system and insulation materials are inspected and maintained, damaged connectors and plug-in components are replaced, the battery pack performance is retested, and the fault is confirmed to be eliminated, after which the entire new energy vehicle battery fault detection process is completed.

2. A new energy vehicle battery fault detection method as claimed in claim 1, characterized in that: In step S1, the detection environment must ensure that the detection site is well ventilated and avoid fire and high temperature safety hazards to prevent accidents during the detection process; In step S2, it is also necessary to check whether the connectors and plug-in components of the battery pack are firm and whether they are loose, detached or damaged. The stability of the fixing bracket is also important, and it is necessary to ensure that there is no looseness or deformation, because these problems cause the battery to shift during vehicle operation, which in turn causes more serious failures.

3. A new energy vehicle battery fault detection method as claimed in claim 1, characterized in that: In step S3, a multimeter is used to measure the voltage of each battery cell one by one, and the data is recorded and compared with the standard voltage range. If an abnormal voltage is found, the faulty cell number should be recorded immediately. At the same time, the voltages of each cell are compared. If the voltage difference exceeds 0.3V, it indicates that the voltage of the battery cells is unbalanced and balanced charging is required to restore battery performance.

4. A new energy vehicle battery fault detection method as claimed in claim 1, characterized in that: In step S4, during measurement, it is necessary to ensure that the internal resistance of each battery cell is within the normal range; if an abnormal internal resistance is detected, the faulty cell number should be recorded; in addition, the internal resistance between cells is compared. If the internal resistance difference exceeds 10mΩ, it indicates that the internal resistance between battery cells is unbalanced and balanced charging is required.

5. A new energy vehicle battery fault detection method as claimed in claim 1, characterized in that: In step S5, if there is local overheating, the fault location should be recorded; the causes of abnormal temperature should be analyzed, including poor heat dissipation, battery cell aging or internal short circuit, all of which lead to battery performance degradation or safety hazards; In step S6, an insulation resistance tester is used to measure the insulation resistance of the battery pack to ensure that it complies with national standards; if the insulation resistance is detected to be low, the fault location should be recorded and the cause should be analyzed.

6. A new energy vehicle battery fault detection method as claimed in claim 1, characterized in that: The step S7 specifically includes the following steps: Step 1: Data collection: During the battery charge and discharge test, a large amount of battery performance data is collected, including voltage, current, temperature, internal resistance, charge / discharge time, capacity, and power; for known faulty batteries, their performance data is collected and marked as faulty data; Step 2: Data preprocessing: Data cleaning: remove outliers, missing values ​​and irrelevant data; Feature engineering: extract features related to battery performance; Step 3: Model training: select the random forest machine learning algorithm to divide the preprocessed data into a training set and a validation set, and then use the training set data to train the machine learning model; Step 4: Model validation and optimization: Use validation set data to evaluate model performance and adjust model parameters; perform cross-validation to ensure the generalization ability of the model; optimize model structure and parameters based on validation results; Step 5: Fault detection: The real-time collected battery performance data is input into the trained machine learning model; the model outputs the prediction result of whether the battery is normal; Threshold setting: Set a threshold based on the probability or decision parameter output by the model to determine whether the battery is faulty; Step 6: Analyze the model prediction results to determine whether the battery has a fault and its type; perform corresponding repairs or replacements based on the fault type indicated by the model.

7. A new energy vehicle battery fault detection method as claimed in claim 6, characterized in that: The data cleaning is performed using the z-score method, and the calculation formula is: z=(X-μ) / σ, where X is the observed value, μ is the mean, σ is the standard deviation, and data with |z|>3 are removed.

8. A new energy vehicle battery fault detection method as claimed in claim 6, characterized in that: The model training selects random forest, and its decision function is: Where wi is the weight of the i-th tree, hi(x) is the prediction result of the i-th tree; Divide the training set Dtrain = {(Xi, Yi)}, where Xi is the feature vector and Yi is the label; The mean square error optimization algorithm is used to minimize the loss function during model training. The calculation formula is: Where Y^i is the model prediction value.

9. A new energy vehicle battery fault detection method as claimed in claim 6, characterized in that: The model validation uses a validation set Dval={(Xi, Yi)} to calculate the loss or accuracy on the validation set; the cross-validation of the model uses k-fold cross-validation to evaluate the model stability: the data set is divided into k subsets, k-1 subsets are used for training each time, and the remaining subset is used for validation; The fault detection real-time data X real-time Input into the model, the formula for model output prediction is: Then set the threshold θ, if Y ^real-time >θ, it is judged as a fault.

10. A new energy vehicle battery fault detection method as claimed in claim 1, characterized in that: In step S8, the following steps are usually followed when using the BMS diagnostic instrument: Connection: Connect the BMS diagnostic instrument to the vehicle's OBD port or BMS communication interface; Power on and initialization: Turn on the diagnostic instrument, wait for it to initialize and establish a communication connection with the BMS; Read data: Select the corresponding function to read the real-time data or historical fault codes of BMS; Analyze data: Analyze the status and potential problems of the battery system based on the read data; Execute operations: perform necessary operations based on the diagnosis results; Recording and reporting: Record the diagnostic process and results, and generate reports for subsequent analysis and archiving.

Citation Information

Cited By

  • New energy automobile power supply fault detection method

    CN119619868A

  • Power battery fault detection method based on improved isolated forest algorithm

    CN120802040A

  • Unmanned aerial vehicle lithium battery anomaly detection system and method based on machine learning model

    CN120831596A