New energy automobile battery fault diagnosis system

Abstract

The invention provides a new energy automobile battery fault diagnosis system, and relates to the technical field of battery fault diagnosis. The new energy automobile battery fault diagnosis system comprises a data acquisition unit, the data acquisition unit is connected with a dynamic baseline generation unit, the dynamic baseline generation unit is connected with a deviation degree calculation unit, the deviation degree calculation unit is connected with a group calibration unit, and the group calibration unit is connected with an early warning decision unit. And the early warning decision-making unit is connected with a base line self-adaptive unit. According to the method, the false alarm rate is remarkably reduced through dynamic baseline generation and entropy weight method calculation, individual and group faults are intelligently distinguished, invalid early warning is reduced, the full-period detection rate is ensured through an SOH coupled adaptive mechanism, excessive intervention is avoided through a graded response strategy, the maintenance efficiency is improved through a multi-mode report, and a novel battery is adapted through closed-loop optimization; and more accurate, safer and more intelligent battery fault diagnosis is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of battery fault diagnosis, and in particular to a new energy vehicle battery fault diagnosis system. Background Art

[0002] New energy vehicles (NEVs) utilize unconventional fuels (such as electricity and hydrogen) as their power source, incorporating advanced vehicle power control and drive technologies. These vehicles feature advanced technical principles, new technologies, and new structures. These primarily include pure electric vehicles, plug-in hybrid vehicles, and fuel cell vehicles. Battery fault diagnosis is crucial for NEVs because, as their core power source, the battery's performance and safety are directly linked to the overall vehicle's operation. During use, batteries can develop faults due to aging, overcharge, over-discharge, and external short circuits. Failure to detect and address these faults can degrade battery performance and shorten its lifespan, impacting vehicle power and range. In severe cases, these faults can lead to safety hazards such as thermal runaway, fire, and explosion, threatening the lives and property of drivers and passengers. Accurate and timely battery fault diagnosis can also provide a basis for battery repair and replacement, reducing maintenance costs. Therefore, battery fault diagnosis is essential for NEVs.

[0003] Existing new energy vehicle battery fault diagnosis systems generally have defects such as poor scenario adaptability, high false alarm rate, and rigid response mechanism. Traditional methods often use fixed thresholds or simple statistical models to determine anomalies, and cannot effectively distinguish between normal parameter fluctuations of batteries under different charging and discharging modes and ambient temperatures and real faults. For example, a momentary drop in battery voltage under fast charging conditions may be mistaken for a short circuit, while an increase in internal resistance caused by a low temperature environment can easily be identified as aging failure. Such false alarms directly lead to unnecessary system power outages or factory repairs, increasing user costs. At the same time, existing technologies lack a dynamic calibration mechanism, and threshold updates rely on manual experience or fixed-cycle adjustments. It is difficult to adapt to the nonlinear changes in battery performance with the number of cycles and health decay, resulting in a significant increase in the missed detection rate in the later stages of aging, posing a safety hazard. Summary of the Invention

[0004] In view of the shortcomings of the existing technology, the present invention provides a new energy vehicle battery fault diagnosis system, which solves the problem of low detection accuracy in the existing technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a new energy vehicle battery fault diagnosis system, comprising a data acquisition unit, the data acquisition unit is connected to a dynamic baseline generation unit, the dynamic baseline generation unit is connected to a deviation calculation unit, the deviation calculation unit is connected to a group calibration unit, the group calibration unit is connected to an early warning decision unit, and the early warning decision unit is connected to a baseline adaptive unit;

[0006] The data acquisition unit is configured to obtain battery cell voltage, battery pack temperature, charge and discharge current and internal resistance parameters in real time. The dynamic baseline generation unit is used to generate a dynamic threshold range for each battery parameter based on historical data within a preset time window. The deviation calculation unit is configured to calculate the single parameter deviation coefficient based on the degree of deviation between the current parameter value and the corresponding dynamic threshold range and obtain the comprehensive deviation by weighting through the entropy weight method. The group calibration unit is connected to the cloud database and is used to normalize the deviation of the current vehicle based on the deviation data of the same type of vehicle group. The early warning decision unit is configured to trigger a graded early warning instruction based on the calibrated deviation value. The baseline adaptive unit is connected to the battery management system and is used to dynamically adjust the dynamic threshold range according to the battery health attenuation value.

[0007] Preferably, the dynamic baseline generation unit includes a data clustering subunit and a quantile calculation subunit, wherein the data clustering subunit is used to classify the historical data according to the charging mode and the ambient temperature, and the quantile calculation subunit is configured to calculate the 10th percentile Q for the data of each scene category. 10 and the 90th percentile Q 90 , and set Q 10 To Q 90 The interval is the dynamic threshold range.

[0008] Preferably, the quantile calculation subunit performs:

[0009] When the battery health deteriorates, the dynamic threshold range is proportionally expanded to meet the following relationship:

[0010] ΔQ=β·(SOH0-SOH t )

[0011] Where β is the preset attenuation compensation coefficient, and SOH0 is the initial health of the battery.

[0012] Preferably, the single parameter deviation coefficient in the deviation calculation unit is defined as the relative offset when the current parameter value exceeds the dynamic threshold range, and the entropy weight method weight allocation is determined based on the data fluctuation entropy value of each parameter within a preset time window.

[0013] Preferably, the group calibration unit includes a deviation comparison subunit and a calibration decision subunit. The deviation comparison subunit calculates the ratio γ of the current vehicle deviation to the average deviation of the same type of vehicle group. The calibration decision subunit is configured to determine it as an individual abnormality when γ>2.5, and to determine it as the influence of group environmental factors when 1.2≤γ≤2.5.

[0014] Preferably, the early warning decision unit implements a multi-level response strategy:

[0015] Generate a visual diagnostic report at level one warning;

[0016] Limit battery charging power during level 2 warning;

[0017] During the third-level warning, maintenance instructions are sent simultaneously to the vehicle display terminal and service platform.

[0018] Preferably, the system also includes a cloud-based collaborative unit, which receives anonymized deviation data uploaded by each vehicle, updates the group deviation baseline value every 6 hours, and initiates a global review of the dynamic threshold when it is detected that more than 20% of vehicles of the same model trigger the same parameter warning.

[0019] Preferably, the data acquisition unit includes a sliding window filter and a data validity verification module, wherein the sliding window filter is used to eliminate the transient noise of the sensor, and the data validity verification module is configured to 10 To Q 90 range is marked as invalid data.

[0020] Preferably, the baseline adaptive unit automatically expands the temperature dynamic threshold range by 15%-25% when the battery cycle number reaches 500 times.

[0021] The present invention provides a new energy vehicle battery fault diagnosis system. It has the following beneficial effects:

[0022] The present invention provides a new energy vehicle battery fault diagnosis system. This technology systematically addresses the shortcomings of existing solutions through multi-dimensional dynamic modeling, group collaborative decision-making, and a hierarchical response mechanism. Based on scenario-based dynamic baseline generation, the present invention uses a quantile algorithm to construct an adaptive threshold, accurately distinguishing normal fluctuations from true anomalies under different operating conditions. This reduces the false alarm rate compared to traditional fixed threshold methods. The invention introduces an entropy weight method for weighted deviation calculation, dynamically assigning weights based on the parameter's own fluctuation characteristics, avoiding the interference of high-noise parameters on the comprehensive evaluation and improving fault location accuracy. By comparing individual deviations with cloud-based group data in real time and using the calibration factor γ to identify individual faults or batch defects, the system achieves a transition from "single-point false alarms" to "swarm intelligence," reducing invalid warnings. Based on the aging characteristics of batteries, an adaptive threshold update rule coupled with SOH is designed to relax the detection standard synchronously with capacity decay, ensuring a stable fault detection rate throughout the entire life cycle. A hierarchical response mechanism is used to implement progressive risk management, avoiding excessive intervention. Multimodal interactive reports are generated simultaneously, improving the efficiency of maintenance personnel in locating the root cause of faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 is a schematic diagram of the present invention;

[0024] Figure 2 Schematic diagram of the present invention. DETAILED DESCRIPTION

[0025] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0026] like Figure 1-2 As shown, an embodiment of the present invention provides a new energy vehicle battery fault diagnosis system, including a data acquisition unit, the data acquisition unit is connected to a dynamic baseline generation unit, the dynamic baseline generation unit is connected to a deviation calculation unit, the deviation calculation unit is connected to a group calibration unit, the group calibration unit is connected to an early warning decision unit, and the early warning decision unit is connected to a baseline adaptive unit;

[0027] The data acquisition unit is configured to obtain battery cell voltage, battery pack temperature, charge and discharge current and internal resistance parameters in real time. The dynamic baseline generation unit is used to generate the dynamic threshold range of each battery parameter based on historical data within a preset time window. The deviation calculation unit is configured to calculate the single parameter deviation coefficient based on the degree of deviation between the current parameter value and the corresponding dynamic threshold range and obtain the comprehensive deviation by weighting through the entropy weight method. The group calibration unit is connected to the cloud database and is used to normalize the deviation of the current vehicle based on the deviation data of the same model vehicle group. The early warning decision unit is configured to trigger a graded early warning instruction according to the calibrated deviation value. The baseline adaptive unit is connected to the battery management system and is used to dynamically adjust the dynamic threshold range according to the battery health attenuation value.

[0028] The dynamic baseline generation unit includes a data clustering subunit and a quantile calculation subunit. The data clustering subunit is used to classify the historical data according to the charging mode and ambient temperature. The quantile calculation subunit is configured to calculate the 10th percentile Q for the data of each scene category. 10 and the 90th percentile Q 90 , and set Q 10 To Q 90 The interval is the dynamic threshold range.

[0029] The quantile calculation subunit performs:

[0030] When the battery health deteriorates, the dynamic threshold range is proportionally expanded to meet the following relationship:

[0031] ΔQ=β·(SOH0-SOH t ) Where β is the preset attenuation compensation coefficient and SOH0 is the initial health of the battery.

[0032] The single-parameter deviation coefficient in the deviation calculation unit is defined as the relative offset when the current parameter value exceeds the dynamic threshold range. The entropy weighting method assigns weights based on the data fluctuation entropy of each parameter within a preset time window. The group calibration unit includes a deviation comparison subunit and a calibration decision subunit. The deviation comparison subunit calculates the ratio γ of the current vehicle's deviation to the average deviation of the same vehicle group. The calibration decision subunit is configured to determine an individual anomaly when γ > 2.5 and a group-wide environmental influence when 1.2 ≤ γ ≤ 2.5.

[0033] The early warning decision-making unit implements a multi-level response strategy:

[0034] Generate a visual diagnostic report at level one warning;

[0035] Limit battery charging power during level 2 warning;

[0036] During the third-level warning, maintenance instructions are sent simultaneously to the vehicle display terminal and service platform.

[0037] The system also includes a cloud-based collaborative unit that receives anonymized deviation data uploaded by each vehicle and updates the group deviation benchmark every 6 hours. When it detects that more than 20% of vehicles of the same model trigger the same parameter warning, it initiates a dynamic threshold global review. The data acquisition unit includes a sliding window filter and a data validity verification module. The sliding window filter is used to eliminate instantaneous noise from the sensor. The data validity verification module is configured to be 10 To Q 90 When the battery cycle number reaches 500, the baseline adaptive unit automatically expands the temperature dynamic threshold range by 15%-25%.

[0038] Data acquisition unit

[0039] Input the original signals of battery cell voltage, temperature, current and internal resistance sensor, and then filter them through sliding window with a window length of 10 seconds to eliminate instantaneous spike noise. If the sampling value exceeds the historical Q for three consecutive times, 10 To Q 90 Range, determine that the sensor fails, trigger redundant sensor switching (such as enabling the backup temperature probe), and finally output the pre-processed standardized data packet (timestamp, parameter value, validity mark)

[0040] Dynamic baseline generation unit

[0041] Input pre-processed historical data and perform scene classification. According to the charging mode, it is divided into fast charging and slow charging. According to the ambient temperature, it is divided into low temperature, normal temperature and high temperature. Perform quantile calculation and calculate Q for each scene data set. 10 , Q 50, Q 90 Percentile, and use T-Digest algorithm to reduce memory usage, generate dynamic threshold range, the normal interval is Q 10 To Q 90 , the abnormal interval is Q 0.1 ~Q 10 or Q 90 ~Q 99.9 , and then output the dynamic threshold table corresponding to each scene (scene label, parameter type, Q 10 / Q 90 value)

[0042] Deviation calculation unit

[0043] Enter the current parameter value and the dynamic threshold of the corresponding scene to process the single parameter deviation coefficient:

[0044] If the current value>Q 90 :Deviation coefficient = (current value - Q 90 ) / (Q 90 -Q 50 )

[0045] If the current value 10 : Deviation coefficient = (QQ 10 -current value) / (Q 50 -Q 10 )

[0046] Other cases: Deviation coefficient = 0

[0047] The information entropy value of each parameter within the time window is calculated by weight allocation using the entropy weight method (reflecting the degree of parameter fluctuation), where weight = (1-entropy value) / sum of all parameter entropy decays (highly volatile parameters have lower weights). The deviation coefficients of each parameter are weighted and summed by the comprehensive deviation, and the final output is the comprehensive deviation value (range 0 to 5, with higher values ​​indicating greater fault risk).

[0048] Group Calibration Unit

[0049] Input the current vehicle's comprehensive deviation and the cloud-based group data for the same vehicle model. Calculate the calibration factor: γ = current vehicle deviation / average deviation for the same vehicle model group (group data is updated every 5 minutes, excluding faulty vehicles).

[0050] The warning levels are:

[0051] γ>2.5→Individual abnormality (triggering level 3 warning)

[0052] 1.2≤γ≤2.5→Group interference (triggering level 2 warning)

[0053] γ<1.2→Normal fluctuation (only logs are recorded) ​

[0054] Early warning decision-making unit

[0055] Input the calibrated warning level and real-time vehicle status to implement a multi-level response strategy.

[0056] Multi-level response strategy table 1

[0057]

[0058]

[0059] Send control instructions to BMS via CAN bus, output control instruction set and user prompt information.

[0060] Baseline Adaptive Unit

[0061] Input the battery health (SOH) and the number of cycles. The threshold relaxation rule is implemented. For every 5% decrease in SOH, the temperature threshold range is expanded by 3% to 8%. After the number of cycles reaches 500, the upper limit of the internal resistance threshold is increased by 10%. The baseline update cycle is triggered every 24 hours or when the SOH changes by ≥1%. Before the update takes effect, it must pass the group data verification and then output the adjusted dynamic threshold parameter table.

[0062] User interaction unit

[0063] Input the warning level and fault parameter data, generate a natural language report, extract the top 3 deviation parameters, match the preset fault knowledge base (such as "voltage abnormality → single short circuit risk"), and then generate a three-part description, including phenomenon description → possible cause → disposal suggestion. Implement a visual interface, including the superposition of parameter trend graphs to display the current value curve and dynamic threshold range, set color coding, where red is exceeded Q 90 , Yellow Q 90 ~Q 95 , green is the normal interval, and the final output is multimodal interactive content, including text, graphics, and voice broadcast.

[0064] Cloud collaboration unit

[0065] Anonymized deviation data for each vehicle is input, and sensitive information such as the VIN is removed. The baseline value is optimized. The mean and variance of the deviation for the same vehicle group are calculated every six hours. If the group mean exceeds 1.2 times the historical baseline for two consecutive hours, a threshold review is triggered. Once the review is passed, the new threshold coefficient is uniformly distributed to the vehicle terminal, and the optimized baseline parameters and global alarm events are output to the cloud.

[0066] The present invention significantly improves the diagnostic accuracy under different working conditions through scenario-based dynamic baseline generation and quantile algorithm, reduces the false alarm rate compared with the traditional fixed threshold method, adopts the entropy weight method for weighted deviation calculation, dynamically allocates parameter weights, effectively filters noise interference, and improves fault location accuracy; by comparing individual and cloud group data in real time, the calibration factor γ is used to intelligently distinguish individual faults from batch defects, reducing invalid warnings; combined with the SOH-coupled adaptive threshold update mechanism, it ensures a stable detection rate throughout the battery life cycle; the hierarchical response strategy realizes progressive risk disposal to avoid excessive intervention, and multi-modal interactive reporting improves maintenance efficiency; the closed-loop data flow design supports system autonomous optimization and can adapt to new battery technologies without human intervention, achieving more intelligent, accurate and safe battery fault diagnosis and warning as a whole.

[0067] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A new energy vehicle battery fault diagnosis system, including a data acquisition unit, characterized in that: The data acquisition unit is connected to a dynamic baseline generation unit, the dynamic baseline generation unit is connected to a deviation calculation unit, the deviation calculation unit is connected to a group calibration unit, the group calibration unit is connected to an early warning decision unit, and the early warning decision unit is connected to a baseline adaptive unit; The data acquisition unit is configured to obtain battery cell voltage, battery pack temperature, charge and discharge current and internal resistance parameters in real time. The dynamic baseline generation unit is used to generate a dynamic threshold range for each battery parameter based on historical data within a preset time window. The deviation calculation unit is configured to calculate the single parameter deviation coefficient based on the degree of deviation between the current parameter value and the corresponding dynamic threshold range and obtain the comprehensive deviation by weighting through the entropy weight method. The group calibration unit is connected to the cloud database and is used to normalize the deviation of the current vehicle based on the deviation data of the same type of vehicle group. The early warning decision unit is configured to trigger a graded early warning instruction based on the calibrated deviation value. The baseline adaptive unit is connected to the battery management system and is used to dynamically adjust the dynamic threshold range according to the battery health attenuation value.

2. A new energy vehicle battery fault diagnosis system according to claim 1, characterized in that: The dynamic baseline generation unit includes a data clustering subunit and a quantile calculation subunit. The data clustering subunit is used to classify historical data into scenarios according to charging mode and ambient temperature. The quantile calculation subunit is configured to calculate the 10th percentile Q10 and the 90th percentile Q90 for the data of each scenario category, and set the interval from Q10 to Q90 as the dynamic threshold range.

3. A new energy vehicle battery fault diagnosis system according to claim 2, characterized in that: The quantile calculation subunit performs: When the battery health deteriorates, the dynamic threshold range is proportionally expanded to meet the following relationship: ΔQ=β·(SOH0-SOH t ) Where β is the preset attenuation compensation coefficient, and SOH0 is the initial health of the battery.

4. A new energy vehicle battery fault diagnosis system according to claim 1, characterized in that: The single parameter deviation coefficient in the deviation calculation unit is defined as the relative offset when the current parameter value exceeds the dynamic threshold range, and the entropy weight method weight allocation is determined based on the data fluctuation entropy value of each parameter within a preset time window.

5. A new energy vehicle battery fault diagnosis system according to claim 1, characterized in that: The group calibration unit includes a deviation comparison subunit and a calibration decision subunit. The deviation comparison subunit calculates the ratio γ of the current vehicle deviation to the average deviation of the same type of vehicle group. The calibration decision subunit is configured to determine that it is an individual abnormality when γ>2.5, and to determine that it is affected by group environmental factors when 1.2≤γ≤2.

5.

6. A new energy vehicle battery fault diagnosis system according to claim 1, characterized in that: The early warning decision unit implements a multi-level response strategy: Generate a visual diagnostic report at level one warning; Limit battery charging power during level 2 warning; During the third-level warning, maintenance instructions are sent simultaneously to the vehicle display terminal and service platform.

7. A new energy vehicle battery fault diagnosis system according to claim 1, characterized in that: The system also includes a cloud-based collaborative unit that receives anonymized deviation data uploaded by each vehicle, updates the group deviation baseline value every 6 hours, and initiates a dynamic threshold global review when it detects that more than 20% of vehicles of the same model trigger the same parameter warning.

8. The new energy vehicle battery fault diagnosis system according to claim 1, characterized in that: The data acquisition unit includes a sliding window filter and a data validity verification module. The sliding window filter is used to eliminate instantaneous noise of the sensor. The data validity verification module is configured to mark three consecutive sampling points as invalid data when they are outside the range of Q10 to Q90.

9. A new energy vehicle battery fault diagnosis system according to claim 1, characterized in that: The baseline adaptive unit automatically expands the temperature dynamic threshold range by 15%-25% when the battery cycle number reaches 500.

Images