CN-121978539-A - Method and system for identifying internal faults of battery pack

CN121978539ACN 121978539 ACN121978539 ACN 121978539ACN-121978539-A

Abstract

The invention discloses a method and a system for identifying internal faults of a battery pack, belonging to the technical field of battery safety detection, wherein the method comprises the steps of obtaining a standard impedance signal, a signal residual value and a standard charge-discharge quality signal of the battery pack to be detected; the method comprises the steps of extracting an internal detection area temperature difference and a three-dimensional hot spot morphology map based on an internal detection area of a battery pack to be detected, carrying out feature extraction, multi-scale feature fusion and feature mapping conversion on the battery pack to be detected based on a preset improved YOLOv model to obtain visual feature data, and judging a standard impedance signal, a signal residual value, a standard charge and discharge quality signal, an internal detection area temperature difference, the three-dimensional hot spot morphology map and the visual feature data based on a preset fault diagnosis model to obtain an internal fault result of the battery pack. The method for identifying the internal faults of the battery pack can be used for comprehensively identifying the faults of the battery pack.

Inventors

JIN LI
LEI ERTAO

Assignees

广东电网有限责任公司电力科学研究院

Dates

Publication Date: 20260505
Application Date: 20260311

Claims (10)

1. A method for identifying an internal fault of a battery pack, comprising: Acquiring a standard impedance signal of a battery pack to be detected, a signal residual value of the battery pack to be detected, a standard charge and discharge quality signal of the battery pack to be detected and an initial battery pack internal visual image of the battery pack to be detected; Extracting an internal detection area temperature difference and a three-dimensional hot spot morphology map based on the internal detection area of the battery pack to be detected and the spatial distribution position of preset detection points; Performing feature extraction and multi-scale feature fusion on an internal visual image of an initial battery pack based on a preset improvement YOLOv model, acquiring a first visual feature map, a second visual feature map and a third visual feature map, performing feature mapping conversion on the first visual feature map, the second visual feature map and the third visual feature map, and acquiring visual feature data; And carrying out multidimensional identification on the standard impedance signal, the signal residual value, the standard charge-discharge quality signal, the internal detection area temperature difference, the three-dimensional hot spot morphology graph and the visual characteristic data based on a preset fault diagnosis model to obtain a battery pack internal fault result.
2. The method for identifying a fault inside a battery pack according to claim 1, wherein the steps of obtaining a standard impedance signal of the battery pack to be detected, a signal residual value of the battery pack to be detected, a standard charge-discharge quality signal of the battery pack to be detected, and an initial visual image of the battery pack inside of the battery pack to be detected, comprise: acquiring an original impedance signal of a battery pack to be detected, and performing low-pass filtering on the original impedance signal to acquire a purified impedance signal; acquiring a fitting impedance signal and a signal residual value based on a preset exponential fitting model and a purified impedance signal; Performing phase correction on the fitted impedance signal to obtain a standard impedance signal; and acquiring a standard charge and discharge quality signal of the battery pack to be detected and an initial battery pack internal visual image of the battery pack to be detected.
3. The method for identifying internal faults of a battery pack according to claim 2, wherein the step of acquiring a standard charge-discharge quality signal of the battery pack to be detected and an initial battery pack internal visual image of the battery pack to be detected comprises the steps of: Acquiring an initial charge and discharge quality signal of a battery pack to be detected and an initial battery pack internal visual image of the battery pack to be detected, and performing DC removal processing on the initial charge and discharge quality signal to acquire a purified charge and discharge quality signal; Performing continuous wavelet transformation on the purified charge-discharge quality signal to extract high-frequency characteristic data; and carrying out signal reconstruction on the high-frequency characteristic data to obtain a standard charge-discharge quality signal.
4. The method for identifying an internal fault of a battery pack according to claim 1, wherein the extracting an internal detection area temperature difference and a three-dimensional hot spot morphology map based on the internal detection area of the battery pack to be detected and the spatial distribution position of preset detection points comprises: acquiring a plurality of detection point temperature values based on an internal detection area of a battery pack to be detected, and calculating an internal detection area temperature difference based on the plurality of detection point temperature values; And extracting a three-dimensional hot spot morphology graph based on the spatial distribution position of the preset detection points and the temperature difference of the internal detection areas.
5. The method for identifying the internal faults of the battery pack according to claim 1, wherein the preset improvement YOLOv model comprises a main network unit, a neck network unit and a head network unit, the method for extracting features and fusing multi-scale features of the internal visual image of the initial battery pack based on the preset improvement YOLOv model, obtaining a first visual feature map, a second visual feature map and a third visual feature map, performing feature mapping conversion on the first visual feature map, the second visual feature map and the third visual feature map, and obtaining visual feature data comprises the following steps: correcting and filtering the internal visual image of the initial battery pack to obtain the internal visual image of the standard battery pack; Performing feature extraction on the internal visual image of the standard battery pack based on the backbone network unit to obtain an initial bottom layer feature map, an initial middle layer feature map and an initial shallow layer feature map; performing multi-scale feature fusion on the initial bottom layer feature map, the initial middle layer feature map and the initial shallow layer feature map based on the neck network unit to obtain a first visual feature map, a second visual feature map and a third visual feature map; and performing feature mapping conversion on the first visual feature map, the second visual feature map and the third visual feature map based on the head network unit to acquire visual feature data.
6. The method for identifying an internal fault of a battery pack according to claim 5, wherein the neck network unit comprises a first neck branch subunit, a second neck branch subunit and a third neck branch subunit, the multi-scale feature fusion is performed on an initial bottom layer feature map, an initial middle layer feature map and an initial shallow layer feature map based on the neck network unit to obtain a first visual feature map, a second visual feature map and a third visual feature map, and the method comprises the following steps: performing enhancement fusion on the initial bottom layer feature map, the initial middle layer feature map and the initial shallow layer feature map based on the first neck branch subunit to obtain a first enhancement feature map and a second enhancement feature map; convolving and fusing the first enhancement feature map, the second enhancement feature map and the initial bottom layer feature map based on the second neck branch subunit to obtain a third enhancement feature map and a fourth enhancement feature map; Performing convolution expansion and spatial feature reconstruction on the second enhancement feature map, the third enhancement feature map and the fourth enhancement feature map in sequence based on the third neck branch subunit to obtain a first initial visual feature map, a second initial visual feature map and a third initial visual feature map; And carrying out channel compression and feature reformation on the first initial visual feature map, the second initial visual feature map and the third initial visual feature map based on the third neck branch subunit to obtain the first visual feature map, the second visual feature map and the third visual feature map.
7. The method for identifying an internal fault of a battery pack according to claim 6, wherein the performing enhancement fusion on the initial bottom layer feature map, the initial middle layer feature map and the initial shallow layer feature map based on the first neck branch subunit to obtain a first enhancement feature map and a second enhancement feature map comprises: dynamically upsampling the initial bottom layer feature map to obtain a reconstructed bottom layer feature map; Splicing and fusing the reconstructed bottom layer feature map and the initial middle layer feature map to obtain a first fused feature map; performing feature enhancement on the first fusion feature map to obtain a first enhancement feature map; Dynamically upsampling the first enhancement feature map to obtain a reconstructed enhancement feature map; splicing and fusing the reconstructed enhanced feature map and the initial shallow feature map to obtain a second fused feature map; And carrying out feature enhancement on the second fusion feature map to obtain a second enhancement feature map.
8. The method for identifying an internal fault of a battery pack according to claim 6, wherein the convolution fusion is performed on the first enhancement feature map, the second enhancement feature map and the initial bottom layer feature map based on the second neck branch subunit, so as to obtain a third enhancement feature map and a fourth enhancement feature map, which comprises: performing convolution downsampling on the second enhancement feature map to obtain a first convolution feature map; splicing and fusing the first convolution feature map and the first enhancement feature map to obtain a third fusion feature map; performing feature enhancement on the third fusion feature map to obtain a third enhancement feature map; performing convolution downsampling on the third enhancement feature map to obtain a second convolution feature map; splicing and fusing the second convolution feature map and the initial bottom layer feature map to obtain a fourth fusion feature map; and carrying out feature enhancement on the fourth fusion feature map to obtain a fourth enhancement feature map.
9. The utility model provides a battery package internal fault identification system which characterized in that includes electric vision data acquisition module, thermal characteristic data extraction module, vision characteristic data acquisition module and multidimensional fault identification module, specifically does: The electric vision data acquisition module is used for acquiring a standard impedance signal of the battery pack to be detected, a signal residual value of the battery pack to be detected, a standard charge and discharge quality signal of the battery pack to be detected and an initial battery pack internal vision image of the battery pack to be detected; The thermal characteristic data extraction module is used for extracting temperature difference and a three-dimensional hot spot morphology graph of an internal detection area based on the internal detection area of the battery pack to be detected and the spatial distribution position of a preset detection point; The visual feature data acquisition module is used for carrying out feature extraction and multi-scale feature fusion on an internal visual image of an initial battery pack based on a preset improvement YOLOv model, acquiring a first visual feature map, a second visual feature map and a third visual feature map, carrying out feature mapping conversion on the first visual feature map, the second visual feature map and the third visual feature map, and acquiring visual feature data; The multi-dimensional fault identification module is used for carrying out multi-dimensional identification on the standard impedance signal, the signal residual value, the standard charge and discharge quality signal, the internal detection area temperature difference, the three-dimensional hot spot morphology graph and the visual characteristic data based on a preset fault diagnosis model, and obtaining the internal fault result of the battery pack.
10. The system of claim 9, wherein the electrical visual data acquisition module is configured to acquire a standard impedance signal of a battery pack to be detected, a signal residual value of the battery pack to be detected, a standard charge-discharge quality signal of the battery pack to be detected, and an initial battery pack internal visual image of the battery pack to be detected, and comprises: acquiring an original impedance signal of a battery pack to be detected, and performing low-pass filtering on the original impedance signal to acquire a purified impedance signal; acquiring a fitting impedance signal and a signal residual value based on a preset exponential fitting model and a purified impedance signal; Performing phase correction on the fitted impedance signal to obtain a standard impedance signal; and acquiring a standard charge and discharge quality signal of the battery pack to be detected and an initial battery pack internal visual image of the battery pack to be detected.

Description

Method and system for identifying internal faults of battery pack Technical Field The invention relates to the technical field of battery safety detection, in particular to a method and a system for identifying internal faults of a battery pack. Background Along with the acceleration of global energy transformation, the method is used as an electrochemical energy storage technology which is a key technology for supporting renewable energy consumption and power grid peak shaving, and the application scale of the method in the fields of smart grids, distributed energy systems, electric automobiles and the like is continuously expanded. The lithium ion battery pack becomes the main stream choice of electrochemical energy storage by virtue of the advantages of high energy density, long cycle life and the like, but the internal structure is complex, and in the actual operation process, the problems of aging of a battery core, local overheating, abnormal charge and discharge and the like are extremely easy to cause thermal runaway, so that safety accidents such as fire and explosion are caused, not only can serious property loss be brought, but also personnel safety is seriously threatened. Therefore, the realization of accurate detection and early fault early warning of the internal state of the battery pack is a core requirement for restricting the safe and stable development of the electrochemical energy storage industry. Under the background of the prior art, the current battery pack detection technology still has a plurality of problems, and is difficult to meet the actual safety detection requirement, firstly, most detection technologies rely on single parameter detection equipment, and the internal complex fault characteristics are difficult to be covered comprehensively, firstly, an independent impedance detection device adopts an off-line measurement mode, so that the dynamic impedance change in the charging and discharging processes of the battery pack can not be captured in real time, and the early-stage battery cell degradation problem is difficult to be identified; secondly, the problem that the high-frequency component capturing precision is insufficient generally exists in the charge and discharge quality monitoring equipment, and potential risks caused by current and voltage waveform distortion cannot be effectively identified. The single parameter detection means can only reflect the local state of the battery pack, and cannot form complete fault diagnosis logic, so that early fault missed judgment and high false judgment rate can be caused. Secondly, most of the existing multi-parameter detection schemes are simple superposition of various independent devices, and lack of systematic integration and data cooperation capability, namely, devices such as impedance detection, charge and discharge quality monitoring, temperature detection and the like independently operate, data are stored in different systems, and synchronous analysis and association judgment cannot be achieved. Meanwhile, part of detection devices are complex to operate, professional personnel are required to debug respectively, the probe structure is not matched with a battery pack access hole, the detection is difficult to penetrate into the battery pack, and the detection device is not suitable for a high-efficiency inspection scene of a large-scale energy storage power station. More importantly, the risk accounting system of the traditional detection scheme has logic contradiction, operation and maintenance personnel lack of reliable judgment basis when facing complex faults, early warning level misjudgment is easy to occur, and early intervention time of thermal runaway is missed. Disclosure of Invention The invention aims to provide a method and a system for identifying faults in a battery pack, which are used for solving the technical problems, avoiding the problems of missed judgment, misjudgment and early warning lag of faults in the battery pack caused by incomplete detection of the inside of the battery pack, and realizing comprehensive fault identification of the inside of the battery pack. In order to solve the above technical problems, the present invention provides a method for identifying an internal fault of a battery pack, including: Acquiring a standard impedance signal of a battery pack to be detected, a signal residual value of the battery pack to be detected, a standard charge and discharge quality signal of the battery pack to be detected and an initial battery pack internal visual image of the battery pack to be detected; Extracting an internal detection area temperature difference and a three-dimensional hot spot morphology map based on the internal detection area of the battery pack to be detected and the spatial distribution position of preset detection points; Performing feature extraction and multi-scale feature fusion on an internal visual image of an initial battery pack based on a preset improvement Y