CN-121978445-A - Vehicle-mounted CCS integrated busbar fault diagnosis method and system based on multi-source data fusion

CN121978445ACN 121978445 ACN121978445 ACN 121978445ACN-121978445-A

Abstract

The application discloses a vehicle-mounted CCS integrated busbar fault diagnosis method and system based on multi-source data fusion, comprising the steps of collecting multi-source real-time data and working condition labels; preprocessing data and extracting multidimensional fault sensitive feature vectors, inputting the feature vectors, working condition labels and historical fault labels into a CNN and LSTM fused model, combining working condition weights to determine fault states and confidence degrees, combining environment temperature and working time to adjust the confidence degrees and executing hierarchical early warning. The application improves the fault diagnosis accuracy and early warning reliability, effectively avoids the safety risk, reduces the overhaul cost and completely meets the engineering application requirements of the vehicle-mounted scene.

Inventors

YANG ZHONGQIANG
LI MINGZHONG
LIU XIUZHEN
WANG MING

Assignees

深圳市至臻精密股份有限公司

Dates

Publication Date: 20260505
Application Date: 20260403

Claims (10)

1. The vehicle-mounted CCS integrated busbar fault diagnosis method based on multi-source data fusion is characterized by comprising the following steps of: acquiring multi-source real-time data from a core area of a vehicle-mounted CCS integrated busbar by utilizing a sensor array, and synchronously recording various working condition labels; preprocessing the multi-source real-time data to obtain preprocessed real-time data, and extracting a multi-dimensional fault-sensitive feature vector from the preprocessed real-time data; Inputting the multidimensional fault sensitive feature vector, the various working condition labels and the historical fault labels into a fault diagnosis model fused with CNN and LSTM to obtain a plurality of fault probability distributions of the CCS integrated busbar, determining a final fault probability distribution according to weights corresponding to all working conditions and the plurality of fault probability distributions, and determining a fault state and corresponding confidence level of the CCS integrated busbar based on the final fault probability distribution; And dynamically adjusting the confidence level by combining the ambient temperature and the working time of the CCS integrated busbar, and executing hierarchical early warning by combining the fault state of the CCS integrated busbar.
2. The method for diagnosing a vehicle-mounted CCS integrated busbar fault based on multi-source data fusion according to claim 1, wherein before collecting multi-source real-time data from a core area of a vehicle-mounted CCS integrated busbar by using a sensor array and synchronously recording a plurality of operating mode labels, said method further comprises: a Hall current sensor, a first differential voltage sensor and a first patch type temperature sensor are deployed at a copper-aluminum bar electric connection node in the CCS integrated busbar; A second differential voltage sensor and a second patch type temperature sensor are deployed at the interface of the FPC/PCB signal acquisition assembly in the CCS integrated busbar; a third patch type temperature sensor is deployed at the stress critical position of the plastic structural part in the CCS integrated busbar; and calibrating and clock synchronizing the Hall current sensor, the first differential voltage sensor, the first patch type temperature sensor, the second differential voltage sensor, the second patch type temperature sensor and the third patch type temperature sensor, connecting an output end to a high-speed data acquisition module, and establishing communication with a vehicle-mounted battery management system through a bus interface of the high-speed data acquisition module.
3. The method for diagnosing a failure of an on-board CCS integrated busbar based on multi-source data fusion as set forth in claim 2, wherein said collecting multi-source real-time data from a core area of the on-board CCS integrated busbar by using a sensor array and synchronously recording a plurality of operating mode labels includes: Triggering the Hall current sensor, the first differential voltage sensor, the first patch type temperature sensor, the second differential voltage sensor, the second patch type temperature sensor and the third patch type temperature sensor to acquire data at a preset sampling frequency to obtain busbar working current acquired by the Hall current sensor, cell single voltage acquired by the first differential voltage sensor, busbar surface temperature acquired by the first patch type temperature sensor, signal line voltage acquired by the second differential voltage sensor, interface temperature acquired by the second patch type temperature sensor and structural member surface temperature acquired by the third patch type temperature sensor; synchronous with the acquisition temperature of the sample, collecting a vehicle working condition label from a vehicle-mounted system; and determining the working current of the busbar, the voltage of the battery cell, the surface temperature of the busbar, the voltage of the signal line, the temperature of the interface and the surface temperature of the structural member as multi-source real-time data.
4. The method for diagnosing an integrated busbar fault of an on-board CCS based on multi-source data fusion as recited in claim 1, wherein said preprocessing said multi-source real-time data to obtain preprocessed real-time data, and extracting a multi-dimensional fault-sensitive feature vector from said preprocessed real-time data includes: noise filtering, missing value filling and abnormal value correction are carried out on the multi-source real-time data to obtain preprocessed real-time data; And respectively extracting time domain features, frequency domain features and time sequence features from the preprocessing real-time data, and splicing the time domain features, the frequency domain features and the time sequence features into multidimensional fault-sensitive feature vectors, wherein the time domain features comprise current features, voltage features and temperature features, the frequency domain features comprise frequency spectrum features, main frequency band energy duty ratio and harmonic distortion rate, and the time sequence features comprise current mutation slope, voltage continuous fluctuation duration and pearson correlation coefficients of temperature change trend.
5. The vehicle-mounted CCS integrated busbar fault diagnosis method based on multi-source data fusion is characterized in that a fault diagnosis model fusing CNN and LSTM comprises a fault diagnosis sub-model fusing CNN and LSTM corresponding to different working conditions and a decision layer fusion sub-module, each fault diagnosis sub-model comprises a feature layer fusion module and a CNN-LSTM hybrid diagnosis module which are sequentially connected, the CNN-LSTM hybrid diagnosis module comprises a CNN sub-module, two LSTM sub-modules and a fault classification head which are sequentially connected, the feature layer fusion module is used for splicing input vectors in dimensions to form initial high-dimensional features, then the initial high-dimensional features are fused in a dimension reducing mode to obtain a fusion feature matrix, the CNN-LSTM hybrid diagnosis module is used for remolding the fusion feature matrix into a two-dimensional feature map, then the two-dimensional feature map is subjected to time sequence feature capture through the two LSTM sub-modules, the fault classification head predicts the fault busbar fault probability according to the history fault labels and the features, and the fault classification head predicts the fault probability distribution of the CCS integrated busbar is used for calculating the fault state according to the confidence coefficient of each CCS integrated busbar fault model and the integrated busbar fault state.
6. The method for diagnosing an integrated busbar fault of an on-board CCS based on multi-source data fusion as set forth in claim 5, wherein said decision layer fusion sub-module is configured to assign weights to each of said fault diagnosis sub-models according to different working conditions, including: training and fusing the corresponding fault diagnosis sub-models of the CNN and the LSTM through historical multi-source data under a single working condition, and distributing weights to each fault diagnosis sub-model according to real-time data of different working conditions to obtain the weights corresponding to each fault diagnosis sub-model.
7. The method for diagnosing a failure of an on-board CCS integrated busbar based on multi-source data fusion as set forth in claim 5, wherein said calculating a failure state and an initial confidence of said CCS integrated busbar based on said weights and a failure probability distribution of said CCS integrated busbar predicted by each failure diagnosis sub-model includes: Multiplying each item of the fault probability distribution of the CCS integrated busbar predicted by each fault diagnosis sub-model with corresponding weight, accumulating to obtain final fault probability distribution of the CCS integrated busbar, and selecting the fault state with the highest probability from the final fault probability distribution to determine the fault state of the CCS integrated busbar and the corresponding fault probability as initial confidence.
8. The method for diagnosing a vehicle-mounted CCS integrated busbar fault based on multi-source data fusion as claimed in claim 5, wherein said inputting said multi-dimensional fault sensitive feature vector, said plurality of condition labels and said history fault labels into a fault diagnosis model fusing CNN and LSTM, obtaining a plurality of fault probability distributions of said CCS integrated busbar, determining a final fault probability distribution according to weights corresponding to respective conditions and said plurality of fault probability distributions, and determining a fault state and a corresponding confidence level of said CCS integrated busbar based on said final fault probability distribution includes: Inputting the multidimensional fault sensitive feature vectors, the various working condition labels and the historical fault labels into a fault diagnosis model fused with CNN and LSTM, splicing the input multidimensional fault sensitive feature vectors in dimensions by using the feature layer fusion module in the fault diagnosis sub-model corresponding to each working condition to form initial high-dimensional features, and then carrying out dimension reduction fusion on the initial high-dimensional features to obtain a fusion feature matrix; The CNN-LSTM hybrid diagnosis module in the fault diagnosis sub-model corresponding to each working condition is utilized to remodel the fusion feature matrix into a two-dimensional feature map, and then the two-dimensional feature map is subjected to time sequence feature capture through the two LSTM sub-modules; Predicting the fault probability distribution of the CCS integrated busbar according to the captured characteristics, the historical fault labels and the working condition labels by using a fault classification head; The decision layer fusion sub-module is utilized to distribute weight to each fault diagnosis sub-model according to different working conditions, and based on the weight and the fault probability distribution of the CCS integrated busbar predicted by each fault diagnosis sub-model, the fault state and the initial confidence of the CCS integrated busbar are calculated; And verifying the fault diagnosis model by using a verification set to calculate classification accuracy, and multiplying the classification accuracy by the initial confidence to obtain the confidence corresponding to the fault state of the CCS integrated busbar.
9. The method for diagnosing a failure of an on-board CCS integrated busbar based on multi-source data fusion as set forth in claim 1, wherein said dynamically adjusting said confidence level in combination with an ambient temperature and a working time of said CCS integrated busbar, and performing a hierarchical early warning in combination with a failure state of said CCS integrated busbar includes: Judging whether the ambient temperature and/or the working time length of the CCS integrated busbar are/is greater than a preset threshold value, if so, increasing the corresponding step length of the confidence level corresponding to the fault state of the CCS integrated busbar on the original basis so as to dynamically adjust the confidence level; and executing hierarchical early warning according to the confidence level after dynamic adjustment and combining the fault state of the CCS integrated busbar.
10. An on-vehicle CCS integrated busbar fault diagnosis system based on multisource data fusion is characterized by comprising: the acquisition module is configured to acquire multi-source real-time data from a core area of the vehicle-mounted CCS integrated busbar by utilizing the sensor array and synchronously record various working condition labels; the processing module is configured to preprocess the multi-source real-time data to obtain preprocessed real-time data, and extract a multi-dimensional fault-sensitive feature vector from the preprocessed real-time data; The prediction module is configured to input the multidimensional fault-sensitive feature vector, the various working condition labels and the historical fault labels into a fault diagnosis model fused with CNN and LSTM, obtain a plurality of fault probability distributions of the CCS integrated busbar, determine final fault probability distribution according to weights corresponding to various working conditions and the plurality of fault probability distributions, and determine fault states and corresponding confidence degrees of the CCS integrated busbar based on the final fault probability distribution; And the grading module is configured to dynamically adjust the confidence level in combination with the ambient temperature and the working time of the CCS integrated busbar and execute grading early warning in combination with the fault state of the CCS integrated busbar.

Description

Vehicle-mounted CCS integrated busbar fault diagnosis method and system based on multi-source data fusion Technical Field The application belongs to the technical field of fault diagnosis, and particularly relates to a vehicle-mounted CCS integrated busbar fault diagnosis method and system based on multi-source data fusion. Background With the iteration of the new energy automobile to the high endurance, fast charging and long service life direction, the safety redundancy design and the dynamic monitoring capability of the power battery system become the core competitiveness of the industry. The vehicle-mounted CCS integrated busbar is used as a nerve center for connecting a battery cell array, transmitting power current and collecting status signals, the running stability of the vehicle-mounted CCS integrated busbar directly determines the power output and a safety base line of the whole vehicle, the vehicle-mounted CCS integrated busbar not only carries large current transmission from the battery cell to an electric control system, but also realizes accurate collection of voltage and temperature of single battery cells through an FPC/PCB, and the vehicle-mounted CCS integrated busbar is a key carrier for realizing overcurrent protection and thermal runaway early warning of a BMS (battery management system). Unlike the traditional discrete busbar, the integrated busbar integrates the copper-aluminum conductor, the insulating package and the signal interface, has extremely complex working environment, is required to bear current impact and temperature fluctuation in charge-discharge circulation, is required to resist strong electromagnetic interference generated by a vehicle-mounted motor and a controller, and is required to adapt to mechanical stress caused by vehicle body vibration. Latent faults (such as contact resistance increase caused by poor contact, conductor fusing caused by overcurrent and insulation layer aging and damage) often have the characteristics of recessive development and burst explosion, and directly threaten the safety of drivers and passengers. Therefore, the implementation of fault diagnosis of 'real-time monitoring, early warning and accurate positioning' on the CCS busbar has become a rigid requirement for the upgrading of the safety technology of the new energy automobile. At present, the fault diagnosis of the vehicle-mounted CCS busbar mainly depends on a basic monitoring function and a post-maintenance means carried by the BMS, and the specific scheme comprises a single parameter threshold monitoring method and a generalized intelligent diagnosis model, and the two schemes have defects, and the specific scheme is as follows: The single parameter threshold monitoring method is the most commonly used diagnosis mode of the current BMS, and realizes fault judgment by setting fixed thresholds of current, voltage and temperature (such as current >600A triggering overcurrent alarm), and has the core defects that (1) the single physical quantity threshold is only relied on, multiple parameter coupling characteristics cannot be associated, such as local temperature rise, voltage drop increase and current fluctuation aggravation caused by 'poor contact', the local temperature rise, the voltage drop increase and the current fluctuation aggravation are caused at the same time, the temperature is only monitored and is easy to be misjudged due to difference of heat dissipation conditions, the voltage is only monitored and is easy to be interfered by fluctuation of consistency of a battery core, (2) the fixed threshold cannot be matched with different working conditions, (3) the hidden faults such as early contact failure, insulation layer micro breakage and the like cannot be effectively identified, and the faults are the sources of follow-up safety accidents. The universal intelligent diagnosis model adopts a single CNN model and an LSTM model to diagnose busbar data, but has no obvious bottleneck in practical application, namely (1) model training does not distinguish working condition characteristics such as starting, accelerating and charging, and fault characteristic difference is obvious under different working conditions, so that the accuracy of cross-working condition diagnosis is suddenly reduced, (2) an original electric signal is directly input into the model, special processing is not carried out on vehicle electromagnetic interference, high-frequency noise generated by motor operation can mask fault sensitive signals, so that the model misjudges interference fluctuation as fault characteristics, and (3) only simple time domain characteristics (such as current average value) are extracted, and characteristics of unfused frequency domains (such as current harmonic distortion) and time sequences (such as voltage mutation slope) cannot be constructed, so that early weak faults are difficult to identify. Therefore, the current related technology is difficult to meet the comprehensi