Research On Failure Diagnosis And Detection Technology Of Lithium-ion Batteries Based On Photoelectric Characteristics

Lithium-ion batteries contain reactive and inflammable materials,which may fail during production,transportation and operation,thus causing safety accidents.Therefore,the safety of lithium-ion batteries has become the focus of attention.Lithium ion battery thermal runaway induction factors can be generally divided into three categories: mechanical abuse,electrical abuse and thermal abuse,these three factors will cause internal short circuit,resulting in battery temperature rise,aerosol generation,fire and explosion.The development of lithium-ion battery failure diagnosis and detection technology is very important for the early warning of battery failure.In order to overcome the problems of aerosol and smoke detection in the current lithium-ion battery failure diagnosis,a new failure detection method is proposed in this paper.Based on the Mie scattering theory and dual-wavelength photoelectric detection technology,this method combines temperature,voltage,current,and characteristic gas parameters to achieve online monitoring of lithium-ion battery failures.The main research contents of this article are as follows:1.Design of a lithium-ion battery failure detection system based on a dualwavelength photoelectric detector.This system is designed using an STM32 embedded system to transplant the Free RTOS operating system,complete relevant application layer software design,and use Lab VIEW to implement the design of a remote monitoring interface for the host computer.The embedded hardware circuit design is completed,which mainly includes control module and signal acquisition module.The photoelectric detector uses blue light and infrared LEDs with wavelengths of 470 nm and 850 nm,respectively,and different combinations of wavelengths can distinguish different types of aerosols and smoke particles.The data output and function configuration are carried out through the serial peripheral interface(SPI)port.2.Implementing photoelectric signal preprocessing using the Hilbert-Huang transform algorithm.The noise generated by the photoelectric detector belongs to a "nonlinear non-stationary" signal.First,the photoelectric signal is decomposed into the modulation signals and frequency spectra of each IMF component through EMD.Then,the Hilbert transform and Hilbert spectral analysis are performed to discover the noise signal generated during photoelectric detection,and finally remove the noise signal to reconstruct the photoelectric signal.3.Conducting experimental research on lithium-ion battery failure detection under overcharge,over-discharge,and puncture abuse conditions.In the three groups of abuse experiments,it was found that the photoelectric detector could quickly detect the gas and aerosols generated by battery failure and rupture,and had a faster response speed compared to electrochemical gas sensors.The experimental results showed that by combining parameters such as temperature,voltage,current,and characteristic gas,the accuracy of lithium-ion battery failure diagnosis could be significantly improved,ultimately achieving online detection of lithium-ion battery failure.4.Studying lithium-ion battery failure diagnosis based on the entropy weight method and grey correlation analysis.Firstly,important indicators that can determine the diagnosis of lithium-ion battery failure,such as photoelectric signal,temperature,voltage,and feature gas,are selected.Then,the weight values of the above important indicators are obtained by using the entropy weight method.Subsequently,the gray correlation analysis method is used to obtain the correlation values of each important indicator,and then a performance evaluation matrix of lithium-ion battery operation is created.Finally,by calculating the correlation degree of each lithium-ion battery operating condition,accurate diagnosis of lithium-ion battery failure can be achieved.