A method for locating overvoltage / undervoltage faults in a battery system

Abstract

This invention relates to a method for locating overvoltage / undervoltage faults in battery systems, belonging to the field of new energy vehicle technology. The method includes: real-time acquisition of battery system SOC, maximum cell voltage, minimum cell voltage, and total current; when an overvoltage alarm occurs in the battery system, fault location is performed: based on characteristic indicators and their corresponding set thresholds, it is determined whether the fault is due to individual cell damage, unreasonable control strategy, slow current limiting by the charging pile, secondary charging, sampling line problems, current feedback, circulating current, or active balancing; the process for determining each set threshold is as follows: based on a failure case library, a fault tree containing all causes of overvoltage faults is established to obtain training set data; the training set data is input into a decision tree classifier to obtain the characteristic indicators of each overvoltage fault cause and the corresponding set thresholds. This invention provides timely fault location in the event of an alarm, improving battery safety.

Description

Technical Field

[0001] This invention relates to a method for locating overvoltage / undervoltage faults in a battery system, belonging to the field of new energy vehicle technology. Background Technology

[0002] New energy vehicles possess the advantages of being clean and pollution-free, and from the perspective of the clean and low-carbon energy transition strategy, their mainstream development has become an inevitable trend. The main power system of new energy vehicles includes the power battery, a device that converts chemical energy into electrical energy. Its main parameters include voltage, internal resistance, temperature, capacity, power, and lifespan. The Battery Management System (BMS) is primarily used for real-time monitoring of the electric vehicle's power battery parameters, fault diagnosis, SOC estimation, mileage estimation, short-circuit protection, leakage detection, display and alarm functions, and charging / discharging mode selection. It also interacts with the vehicle's integrated controller or charger via a CAN bus to ensure the efficient, reliable, and safe operation of the electric vehicle.

[0003] The power systems of new energy vehicles have different normal operating voltage ranges depending on the type of battery material. When the highest voltage of a single cell exceeds the upper limit of the normal operating voltage range, the battery management system will issue an overvoltage alarm; when the lowest voltage of a single cell is lower than the lower limit of the normal operating voltage range, the battery management system will issue an undervoltage alarm.

[0004] However, simply issuing an alarm in the power system is insufficient for fault location. In actual vehicle operation, there are many reasons for overvoltage and undervoltage, and the current method of simply issuing an alarm based on exceeding a threshold is inadequate for practical problem-solving. To determine the exact cause of overvoltage or undervoltage, after-sales personnel need to go to the site to investigate each possible cause, resulting in low efficiency in fault location. Summary of the Invention

[0005] The purpose of this application is to provide a method for locating overvoltage / undervoltage faults in a battery system, in order to solve the problem of low efficiency in existing fault location methods.

[0006] To achieve the above objectives, this application proposes a technical solution for locating overvoltage faults in a battery system, comprising the following steps:

[0007] 1) Real-time acquisition of battery system characteristic indicators; characteristic indicators include battery system SOC, maximum voltage of individual cells, minimum voltage of individual cells, and total current;

[0008] 2) When an overvoltage alarm occurs in the battery system, perform fault location:

[0009] Select battery data whose alarm time is longer than the set time, extract the highest battery system SOC and minimum total current within that period, and compare the highest battery system SOC, minimum total current, and highest single cell voltage with the corresponding set thresholds to determine whether a single cell is damaged.

[0010] When the vehicle is charging and is a hybrid vehicle, the battery system SOC is compared with the corresponding set threshold to determine whether the control strategy is unreasonable.

[0011] When the vehicle is charging and is a hybrid / pure electric vehicle, the number of records with negative total current before the alarm is compared with the corresponding set threshold to determine whether it is due to slow current limiting of the charging pile or secondary charging.

[0012] When the vehicle is in motion, the changes in the highest and lowest voltages of individual cells are compared with the corresponding set thresholds to determine whether it is a sampling line problem.

[0013] When the vehicle is in motion, the number of records of negative total current before the alarm and the battery system SOC are compared with the corresponding set threshold to determine whether it is current feedback, circulating current or active balancing.

[0014] The process for determining each set threshold is as follows:

[0015] Establish a fault tree containing all causes of overvoltage failure;

[0016] Training set data is built based on the fault tree; the training set data includes faulty vehicle data and normal vehicle data. Faulty vehicle data includes overvoltage fault cause identification and all characteristic index values; normal vehicle data includes normal vehicle identification and all characteristic index values.

[0017] The training set data is input into the decision tree classifier for training and learning, thereby establishing a decision tree for each overvoltage fault cause and obtaining the feature indicators and corresponding threshold values ​​for each overvoltage fault cause.

[0018] The beneficial effects of the technical solution of the battery system overvoltage fault location method of the present invention are as follows: Based on the analysis of faulty vehicles in the failure database, the present invention establishes a complete fault tree, and uses a machine learning method—decision tree algorithm—to obtain the characteristic indicators of various fault causes and their corresponding set thresholds. Then, under the real-time collected characteristic indicators, once an alarm occurs, the fault can be located. This not only accurately locates the specific problem and identifies the problem in advance, but also significantly improves the efficiency of fault repair, comprehensively improves the safety of the battery, ensures that all faulty vehicles can be relieved of danger in a timely manner, and avoids major safety accidents.

[0019] Furthermore, if the highest battery system SOC is less than or equal to the first set threshold, or if the highest battery system SOC is greater than the first set threshold and the minimum total current is greater than the second set threshold and the highest voltage of a single cell is greater than the third set threshold, then the single cell is damaged; otherwise, the single cell is not damaged.

[0020] Furthermore, when the vehicle is charging and is a hybrid vehicle, if the battery system SOC is less than the fourth set threshold, it is determined that the control strategy is unreasonable; when the vehicle is charging and is a hybrid / pure electric vehicle, if the total current is negative and the number of records before the alarm is the fifth set threshold, it is determined that the charging pile is limiting the current too slowly, otherwise it is determined that it is recharging.

[0021] Furthermore, when the vehicle is in operation, if the change in the highest voltage of a single cell is greater than the high voltage jump-off setting threshold and the change in the lowest voltage of a single cell is greater than the low voltage jump-off setting threshold, it is determined to be a sampling line problem; when the vehicle is in operation, if the total current before the alarm is negative, the number of records is the fifth setting threshold, and the battery system SOC is less than the sixth setting threshold, it is determined to be current feedback; otherwise, it is determined to be circulating current or active balancing.

[0022] Furthermore, it also includes a step of pruning the decision tree for each cause of overvoltage fault.

[0023] In addition, this application proposes a technical solution for a battery system undervoltage fault location method, including the following steps:

[0024] 1) Real-time acquisition of battery system characteristic indicators; characteristic indicators include battery system SOC, maximum cell voltage and its corresponding cell number, minimum cell voltage and its corresponding cell number, total current, SOC corresponding to the maximum cell voltage, and SOC corresponding to the minimum cell voltage;

[0025] 2) When the battery system experiences an intermittent undervoltage alarm, compare the changes in the highest and lowest voltage of a single cell with the corresponding set thresholds to determine if it is a sampling line problem; compare the battery system SOC and total current with the corresponding set thresholds to determine if voltage polarization is caused by increased current.

[0026] When the battery system experiences a continuous low voltage alarm, the system compares the difference between the SOC corresponding to the highest voltage and the SOC corresponding to the lowest voltage of a single cell, the highest voltage of the single cell at the end of charging, and the lowest voltage of the single cell at the end of discharging with the corresponding set thresholds to determine if the cell internal resistance is too high; it also compares the difference between the SOC corresponding to the highest voltage and the SOC corresponding to the lowest voltage of a single cell, the lowest voltage of the single cell at the end of charging, and the lowest voltage of the single cell at the end of discharging with the corresponding set thresholds to determine if the cell capacity is abnormal; and it further compares the lowest voltage of the single cell and the total current with the corresponding set thresholds to determine if the discharge relay is stuck.

[0027] The process for determining each set threshold is as follows:

[0028] Build a fault tree that includes all causes of undervoltage faults;

[0029] Training set data is built based on the fault tree; the training set data includes faulty vehicle data and normal vehicle data. Faulty vehicle data includes undervoltage fault cause identifiers and all characteristic index values; normal vehicle data includes normal vehicle identifiers and all characteristic index values.

[0030] The training set data is input into the decision tree classifier for training and learning, thereby establishing a decision tree for each undervoltage fault cause and obtaining the feature indicators and the corresponding set thresholds for each undervoltage fault cause.

[0031] The beneficial effects of the technical solution of the battery system undervoltage fault location method of the present invention are as follows: Based on the analysis of faulty vehicles in the failure database, the present invention establishes a complete fault tree, and uses a machine learning method—decision tree algorithm—to obtain the characteristic indicators of various fault causes and their corresponding set thresholds. Then, under the real-time collection of characteristic indicators, once an alarm occurs, the fault can be located. This not only accurately locates the specific problem and identifies the problem in advance, but also significantly improves the efficiency of fault repair, comprehensively improves the safety of the battery, ensures that all faulty vehicles can be relieved of danger in a timely manner, and avoids major safety accidents.

[0032] Furthermore, if the difference between the SOC corresponding to the highest voltage of a single cell and the SOC corresponding to the lowest voltage of a single cell is greater than the seventh set threshold, and the highest voltage of a single cell at the charging end is greater than the high voltage set threshold, and the lowest voltage of a single cell at the discharging end is less than the first low voltage set threshold, and the highest voltage and the lowest voltage of a single cell are from the same single cell, then the single cell is judged to have a high internal resistance; if the difference between the SOC corresponding to the highest voltage of a single cell and the SOC corresponding to the lowest voltage of a single cell is greater than the eighth set threshold and shows an upward trend, and the lowest voltage of a single cell at the charging end is less than the second low voltage set threshold, and the lowest voltage of a single cell at the discharging end is less than the third low voltage set threshold, and the highest voltage and the lowest voltage of a single cell are from the same single cell, then the single cell is judged to have an abnormal capacity.

[0033] Furthermore, if the lowest voltage of a single cell is less than the ninth set threshold and continues to decrease, and the total current is greater than the tenth set threshold, then the discharge relay is determined to be stuck.

[0034] Furthermore, if the battery system SOC is less than the thirteenth set threshold and the total current is greater than the fourteenth set threshold, and the current reduction alarm is canceled, then the voltage polarization is determined to be caused by increased current; if the change in the highest voltage of a single cell is greater than the high voltage jump setting threshold and the change in the lowest voltage of a single cell is greater than the low voltage jump setting threshold, then the sampling line problem is determined to be the problem.

[0035] Furthermore, it also includes a step of pruning the decision tree for each cause of undervoltage fault. Attached Figure Description

[0036] Figure 1 This is a flowchart of the battery system overvoltage fault location method of the present invention;

[0037] Figure 2 This is the fault tree of all causes of overvoltage faults in this invention;

[0038] Figure 3 This is the decision tree for individual cell damage under overvoltage fault conditions according to the present invention;

[0039] Figure 4 This is a flowchart of the battery system undervoltage fault location method of the present invention;

[0040] Figure 5 This is the fault tree of all causes of undervoltage faults in this invention. Detailed Implementation

[0041] Example of a method for locating overvoltage faults in a battery system:

[0042] The main concept of this invention is to establish a fault tree for each cause of overvoltage failure in the battery system, and use a decision tree algorithm to determine the characteristic indicators of each fault cause and the corresponding set thresholds. Then, by using the characteristic indicators of the battery system, the fault can be quickly located when an overvoltage alarm occurs, and the cause of the overvoltage failure can be found, thereby improving the efficiency of fault location.

[0043] Specifically, the battery system overvoltage fault location method is as follows: Figure 1 As shown, it includes the following steps:

[0044] 1) Establish a fault tree containing the causes of overvoltage failures based on the corresponding failure case library, and obtain the characteristic indicators of each overvoltage failure cause and the set threshold corresponding to the characteristic indicators based on the decision tree algorithm.

[0045] The specific process for this step is as follows:

[0046] a. Create a fault tree containing all causes of overvoltage failure.

[0047] Based on the existing failure case database, the specific causes of overvoltage were identified by investigating the faulty vehicles. Factors affecting battery overvoltage include individual cell damage, abnormal sampling lines, and circulating current. A system was established to address these issues. Figure 2 The fault tree shown represents all the causes of overvoltage faults.

[0048] b. Establish training set data based on the fault tree; the training set data includes faulty vehicle data and normal vehicle data (the ratio of faulty vehicle data to normal vehicle data is 1:1). The faulty vehicle data includes the overvoltage fault cause identifier and all characteristic index values; the normal vehicle data includes the normal vehicle identifier and all characteristic index values. Here, the faulty vehicle data corresponds to a specific overvoltage fault cause.

[0049] c. Input the training set data into the decision tree classifier for training and learning, and then build a decision tree for each cause of overvoltage fault.

[0050] d. Decision tree pruning: Decision tree pruning is the process of verifying and correcting the decision tree generated in the previous stage. It mainly uses data from a new sample dataset (called the test dataset, which has the same data type as the training set) to verify the preliminary rules generated during the decision tree generation process, and prunes those branches that affect the accuracy of the prediction, and finally obtains the characteristic indicators of each overvoltage fault cause and the set thresholds corresponding to the characteristic indicators.

[0051] Taking the cause of damage to a single battery cell as an example, this paper provides a detailed explanation of the process for determining the threshold for setting a decision tree.

[0052] (1) The core influencing factors of abnormal battery cells are SOC, current, voltage and time. The characteristic indicators are the battery system SOC, total current, lowest cell maximum voltage, highest cell maximum voltage and overvoltage duration during the overvoltage alarm stage.

[0053] (2) Using the Gini index as the splitting criterion, the attribute with the smallest Gini index is used as the optimal splitting attribute to construct the decision tree. Then, a post-pruning strategy is used to prune the decision tree to obtain the final result. Figure 3 The decision tree shown represents the damage to a single battery cell.

[0054] Based on the post-pruning strategy, the final identification logic and feature values ​​are determined by three characteristic indicators: Select battery data with an overvoltage alarm duration greater than 1 minute (here, the duration is set to 1 minute), and extract the highest battery system SOC, minimum total current, and highest single-cell voltage within this time period. If the highest battery system SOC ≤ X1, and an overvoltage alarm is reported for a single cell, then the cell is damaged; if the highest battery system SOC > X1 and the minimum total current ≤ X2, then the cell is normal; if the highest battery system SOC > X1, the minimum total current > X2, and the highest single-cell voltage > X3, then the cell is damaged; otherwise, the cell is normal.

[0055] X1 is the SOC value generated by the decision tree algorithm in the training set for judging cell abnormalities; X1 can be 80%-90%. X2 is the current value generated by the decision tree algorithm in the training set for judging cell abnormalities; X2 can be -30A to -50A. X3 is the highest voltage value generated by the decision tree algorithm in the training set for judging cell abnormalities; X3 can be 3.9V to 4.0V.

[0056] Figure 3 In this context, "sample size" represents the total number of selected cases; "value" represents the number of different categories under the current node's conditions, for example, "value = [18, 35]" represents 18 positive cases and 35 negative cases; "category = 1" represents a negative case, i.e., a single damaged vehicle; "category = 0" represents a positive case, i.e., a normal vehicle; the "Gini coefficient" is used to measure the purity of the node. The less pure the node, the larger the Gini value and the greater the entropy. When the Gini coefficient = 0, the node no longer has branches. In summary, the meaning of each internal node's test output is as follows: Under certain conditions, there exists a Gini coefficient. When the Gini coefficient is not equal to 0, there are a total of 2 categories in the sample size, and the number of each category is fixed. After the node test output, the sum of category 1 in each branch equals the value of category 1 in the internal node, and the sum of category 0 equals the value of category 0 in the internal node.

[0057] 2) Real-time acquisition of battery system characteristic indicators; characteristic indicators include battery system SOC, highest voltage of individual cells, lowest voltage of individual cells, and total current.

[0058] 3) When the battery system experiences an overvoltage alarm, perform fault location.

[0059] Select battery data with an alarm time longer than the set duration, and extract the highest battery system SOC and minimum total current within that period. If the highest battery system SOC is less than or equal to the first set threshold X1, or if the highest battery system SOC is greater than the first set threshold X1 and the minimum total current is greater than the second set threshold X2 and the highest voltage of a single cell is greater than the third set threshold X3, then the single cell is damaged; otherwise, the single cell is not damaged. X1 can be 80%-90%, X2 can be -30A to -50A, and X3 can be 3.9V to 4.0V.

[0060] When the vehicle is charging and is a hybrid vehicle (i.e., a plug-in hybrid), if the battery system SOC is less than the fourth preset threshold X4, the control strategy is deemed unreasonable. When the vehicle is charging and is a hybrid / pure electric vehicle, if the total current before the alarm is negative and the number of records is the fifth preset threshold X5, the charging pile is deemed to be limiting current too slowly; otherwise, it is deemed to be recharging. X4 represents the battery capacity when the plug-in hybrid vehicle is fully charged, and X4 can be 93%-95%. X5 can be 1-10.

[0061] When the vehicle is in motion, if the change in the highest voltage of a single cell exceeds the high-voltage jump threshold and the change in the lowest voltage of a single cell exceeds the low-voltage jump threshold (i.e., high voltage jumps high, low voltage jumps low), it is determined to be a sampling line problem. Verification showed that the jump change in the highest voltage of a single cell is >200mV, and the jump change in the lowest voltage of a single cell is >80mV, indicating a sampling line problem. When the vehicle is in motion, if the total current before the alarm is negative, the number of records is at the fifth set threshold, and the battery system SOC is < the sixth set threshold X6, it is determined to be current feedback; otherwise, it is determined to be circulating current or active balancing. X6 represents the SOC value under high charge conditions, therefore X6 < 100%.

[0062] Example of a method for locating undervoltage faults in a battery system:

[0063] Battery system undervoltage fault location methods, such as Figure 4 As shown, it includes the following steps:

[0064] 1) Establish such Figure 5 The fault tree shown contains all the causes of undervoltage faults; the causes of undervoltage faults include high internal resistance of individual cells and abnormal capacity, etc.

[0065] Training set data is built based on the fault tree; the training set data includes faulty vehicle data and normal vehicle data. Faulty vehicle data includes undervoltage fault cause identifiers and all characteristic index values; normal vehicle data includes normal vehicle identifiers and all characteristic index values.

[0066] The training set data is input into the decision tree classifier for training and learning, thereby establishing a decision tree for each undervoltage fault cause and obtaining the feature indicators and the corresponding set thresholds for each undervoltage fault cause.

[0067] The process of determining each set threshold in the battery system undervoltage fault location method is the same as that in the battery system overvoltage fault location method, and will not be repeated here.

[0068] 2) Real-time acquisition of battery system characteristic indicators; characteristic indicators include battery system SOC, highest cell voltage and its corresponding cell number, lowest cell voltage and its corresponding cell number, total current, SOC corresponding to the highest cell voltage, and SOC corresponding to the lowest cell voltage.

[0069] 3) Locate faults based on characteristic indicators.

[0070] When the battery system experiences a persistent low voltage alarm:

[0071] If the difference ΔSOC between the highest voltage and the lowest voltage of a single cell is greater than the seventh set threshold X7, and the highest voltage of the single cell at the charging end is greater than the high voltage set threshold, and the lowest voltage of the single cell at the discharging end is less than the first low voltage set threshold (i.e., high charge, low discharge), and the cells corresponding to the highest and lowest voltages of a single cell are the same cell, then it is determined that the internal resistance of the single cell is high. Here, the high voltage set threshold is the voltage value corresponding to the highest voltage of the single cell at the charging end, and the first low voltage set threshold is the voltage value corresponding to the lowest voltage of the single cell at the discharging end.

[0072] If the difference ΔSOC between the highest SOC and the lowest SOC of a single cell is greater than the eighth set threshold X8 and shows an upward trend, and the lowest voltage of the single cell at the end of charging is less than the second low-voltage set threshold, and the lowest voltage of the single cell at the end of discharging is less than the third low-voltage set threshold (i.e., low charge, low discharge), and the cells corresponding to the highest and lowest voltages of a single cell are the same cell, then the single cell capacity is judged to be abnormal. Here, the second low-voltage set threshold is the voltage value corresponding to the lowest voltage of the single cell at the end of charging, and the third low-voltage set threshold is the voltage value corresponding to the lowest voltage of the single cell at the end of discharging. X7 and X8 represent the voltage difference limits for consistency differences. In this case, 15% < X7 = X8 < 18%.

[0073] If the lowest voltage of a single cell is less than the ninth set threshold X9 and continues to decrease, and the total current is greater than the tenth set threshold X10, then the discharge relay is judged to be stuck. X9 represents the voltage when the current limit is 0, where 2.0V < X9 < 2.5V. X10 represents the actual current value of the battery when the current limit is 0, and the judgment condition at this time is X10 ≥ 0.

[0074] Of course, there are also some fault locations that can be determined manually, such as:

[0075] If the characteristics of a parked vehicle are present and the minimum voltage of a single cell in the last message reported by the BMS before the parked vehicle is less than the eleventh set threshold X11, then the vehicle is judged to be a high-risk parked vehicle. X11 represents the voltage value of the last data of the battery before the parked vehicle. In this case, 3.0V < X11 < 3.1V, which means that the battery charge was low before the parked vehicle.

[0076] If the alarm duration exceeds the twelfth preset threshold x12, and the alarm is cleared after charging, it is determined that the charging was not timely. X12 represents the duration of the continuous low-voltage alarm, and X12 ≥ 60 seconds.

[0077] When the battery system experiences an intermittent undervoltage alarm:

[0078] If the battery system SOC is less than the thirteenth set threshold X13 and the total current is greater than the fourteenth set threshold X14, and the current reduction alarm is canceled, then it is judged that the current has increased. X13 represents the charge level when the fault alarm occurs, 30% < X13 < 35%; X14 represents the large current that occurs when the charge level is low, 100A < X14 < 150A.

[0079] If the change in the highest voltage of a single cell is greater than the high-voltage switching threshold and the change in the lowest voltage of a single cell is greater than the low-voltage switching threshold, then it is determined to be a sampling line problem.

[0080] This invention analyzes faulty vehicles in a failure database, establishes a comprehensive fault tree, and uses a machine learning approach—the decision tree algorithm—to obtain characteristic indicators of various fault causes and their corresponding set thresholds. Then, based on the real-time collected characteristic indicators, once an alarm is triggered, the fault can be located, ensuring that all faulty vehicles can be promptly relieved of danger and preventing major safety accidents.

Claims

1. A method for locating overvoltage faults in a battery system, characterized in that, Includes the following steps: 1) Real-time acquisition of battery system characteristic indicators; characteristic indicators include battery system SOC, maximum voltage of individual cells, minimum voltage of individual cells, and total current; 2) When an overvoltage alarm occurs in the battery system, perform fault location: Select battery data whose alarm time is longer than the set time, extract the highest battery system SOC and minimum total current within that period, and compare the highest battery system SOC, minimum total current, and highest single cell voltage with the corresponding set thresholds to determine whether a single cell is damaged. When the vehicle is charging and is a hybrid vehicle, the battery system SOC is compared with the corresponding set threshold to determine whether the control strategy is unreasonable. When the vehicle is charging and is a hybrid / pure electric vehicle, the number of records of negative total current before the alarm is compared with the corresponding set threshold to determine whether it is due to slow current limiting of the charging pile or secondary charging. When the vehicle is in motion, the changes in the highest and lowest voltages of individual cells are compared with the corresponding set thresholds to determine whether it is a sampling line problem. When the vehicle is in motion, the number of records of negative total current before the alarm and the battery system SOC are compared with the corresponding set threshold to determine whether it is current feedback, circulating current or active balancing. The process for determining each set threshold is as follows: Establish a fault tree containing all causes of overvoltage failure; Training set data is built based on the fault tree; the training set data includes faulty vehicle data and normal vehicle data. Faulty vehicle data includes overvoltage fault cause identification and all characteristic index values; normal vehicle data includes normal vehicle identification and all characteristic index values. The training set data is input into the decision tree classifier for training and learning, thereby establishing a decision tree for each overvoltage fault cause and obtaining the feature indicators and corresponding threshold values ​​for each overvoltage fault cause.

2. The battery system overvoltage fault location method according to claim 1, characterized in that, If the highest battery system SOC is less than or equal to the first set threshold, or if the highest battery system SOC is greater than the first set threshold and the minimum total current is greater than the second set threshold and the highest voltage of a single cell is greater than the third set threshold, then the single cell is damaged; otherwise, the single cell is not damaged.

3. The battery system overvoltage fault location method according to claim 1, characterized in that, When the vehicle is charging and is a hybrid vehicle, if the battery system SOC is less than the fourth set threshold, the control strategy is deemed unreasonable. When the vehicle is charging and is a hybrid / pure electric vehicle, if the total current is negative and the number of records is the fifth set threshold before the alarm, the charging pile is deemed to be limiting current too slowly; otherwise, it is deemed to be recharging.

4. The battery system overvoltage fault location method according to claim 1, characterized in that, When the vehicle is in motion, if the change in the highest voltage of a single cell is greater than the high voltage jump-off setting threshold and the change in the lowest voltage of a single cell is greater than the low voltage jump-off setting threshold, it is determined to be a sampling line problem; when the vehicle is in motion, if the total current before the alarm is negative, the number of records is the fifth setting threshold, and the battery system SOC is less than the sixth setting threshold, it is determined to be current feedback; otherwise, it is determined to be circulating current or active balancing.

5. The battery system overvoltage fault location method according to claim 1, characterized in that, It also includes a step of pruning the decision tree for each cause of overvoltage fault.

6. A method for locating undervoltage faults in a battery system, characterized in that, Includes the following steps: 1) Real-time acquisition of battery system characteristic indicators; characteristic indicators include battery system SOC, maximum cell voltage and its corresponding cell number, minimum cell voltage and its corresponding cell number, total current, SOC corresponding to the maximum cell voltage, and SOC corresponding to the minimum cell voltage; 2) When the battery system experiences an intermittent undervoltage alarm, compare the changes in the highest and lowest voltage of a single cell with the corresponding set thresholds to determine if it is a sampling line problem; compare the battery system SOC and total current with the corresponding set thresholds to determine if voltage polarization is caused by increased current. When the battery system experiences a continuous low voltage alarm, the system compares the difference between the SOC corresponding to the highest voltage and the SOC corresponding to the lowest voltage of a single cell, the highest voltage of the single cell at the end of charging, and the lowest voltage of the single cell at the end of discharging with the corresponding set thresholds to determine if the cell internal resistance is too high; it also compares the difference between the SOC corresponding to the highest voltage and the SOC corresponding to the lowest voltage of a single cell, the lowest voltage of the single cell at the end of charging, and the lowest voltage of the single cell at the end of discharging with the corresponding set thresholds to determine if the cell capacity is abnormal; and it further compares the lowest voltage of the single cell and the total current with the corresponding set thresholds to determine if the discharge relay is stuck. The process for determining each set threshold is as follows: Build a fault tree that includes all causes of undervoltage faults; Training set data is built based on the fault tree; the training set data includes faulty vehicle data and normal vehicle data. Faulty vehicle data includes undervoltage fault cause identifiers and all characteristic index values; normal vehicle data includes normal vehicle identifiers and all characteristic index values. The training set data is input into the decision tree classifier for training and learning, thereby establishing a decision tree for each undervoltage fault cause and obtaining the feature indicators and the corresponding set thresholds for each undervoltage fault cause.

7. The battery system undervoltage fault location method according to claim 6, characterized in that, If the difference between the SOC corresponding to the highest voltage and the SOC corresponding to the lowest voltage of a single cell is greater than the seventh set threshold, and the highest voltage of the single cell at the charging end is greater than the high voltage set threshold, and the lowest voltage of the single cell at the discharging end is less than the first low voltage set threshold, and the highest voltage and the lowest voltage of the single cell are from the same single cell, then the single cell is judged to have high internal resistance. If the difference between the SOC corresponding to the highest voltage and the SOC corresponding to the lowest voltage of a single cell is greater than the eighth set threshold and shows an upward trend, and the lowest voltage of the single cell at the charging end is less than the second low voltage set threshold, and the lowest voltage of the single cell at the discharging end is less than the third low voltage set threshold, and the highest voltage and the lowest voltage of the single cell are from the same single cell, then the single cell is judged to have abnormal capacity.

8. The battery system undervoltage fault location method according to claim 6, characterized in that, If the lowest voltage of a single cell is less than the ninth set threshold and continues to decrease, and the total current is greater than the tenth set threshold, then the discharge relay is determined to be stuck.

9. The battery system undervoltage fault location method according to claim 6, characterized in that, If the battery system SOC is less than the thirteenth set threshold and the total current is greater than the fourteenth set threshold, and the current reduction alarm is canceled, then the voltage polarization is determined to be caused by increased current. If the change in the highest voltage of a single cell is greater than the high voltage jump setting threshold and the change in the lowest voltage of a single cell is greater than the low voltage jump setting threshold, then the sampling line problem is determined to be the problem.

10. The battery system undervoltage fault location method according to claim 6, characterized in that, It also includes a step of pruning the decision tree for each cause of undervoltage fault.

Images