Research On Safety Warning And Fault Location Of Lithium Battery Energy Storage Cabin Based On Acoustic Signal

Large-scale electric energy storage facilities can promote the development of smart grids,solve the problem of new energy volatility,and have a bright future.Lithium battery energy storage has the advantages of high energy density and low cost,so it is the most widely used energy storage technology.However,due to the high activity of lithium battery and the lack of fault warning ability,the frequent fire and explosion accidents of lithium battery energy storage power stations have seriously restricted its application and promotion.Based on the experimental results of thermal runaway of battery unit and module,the effectiveness and sensitivity of acoustic signal warning are verified.On this basis,the method of early warning of thermal runaway based on the identification of acoustic signal of safety vent in the energy storage tank and the fault battery positioning method based on sound source location are proposed,so as to provide a kind of alert hand with both sensitivity and economy for lithium battery energy storage power station.The main research works are as follows:(1)The validity and sensitivity verification of acoustic signal warning.Build a thermal runaway experiment platform,overcharge single cells and battery modules to cause thermal runaway.The experiments show that cutting off the battery’s external power supply in time after the safety vent is opened can effectively curb the development of thermal runaway.The sound signal of the safety vent can be used as an effective warning signal for thermal runaway,and the sensitivity is more sensitive than that of visible,infrared and gas detection methods.(2)Effective recognition of the safety venting acoustic signal.A threshold denoising scheme based on wavelet theory was designed for the noise environment inside the energy storage cabin,and the parameter settings were determined.The Mel-frequency cepstral coefficients(MFCC)is applied to the feature extraction of acoustic signals inside the energy storage cabin,and the 40-dimensional MFCC feature coefficients of the acoustic signal are extracted.Finally,a data set based on the 40-dimensional MFCC feature coefficients is constructed.The KNN,PNN,and SVM algorithms are respectively applied to the recognition of safety venting acoustic signals,and 10-fold cross-validation is used to find the better parameters.And the results show that SVM has the best accuracy for the safety venting acoustic signal,and the recognition rate of the trained SVM model for unknown data is 97.4%.(3)Acoustic source location of the safety venting acoustic signal.An energy storage cabin was built to simulate the internal acoustic environment.Time difference of arrival(TDOA)method is selected for source location,and a quaternary microphone synchronous sound signal acquisition system was designed.The quaternary microphones were arranged in a three-dimensional manner and the safety valve sound signal was released at a specific position.For four-microphone signals with severe reverberations in real environments,an effective signal selection method based on threshold is designed,and the microphone time difference is successfully extracted by using the generalized cross-correlation algorithm.The spherical interpolation method based on TDOA is used to solve the extracted cross-correlation time difference.The results show that the average error of sound source position is in cm level according to the designed system and positioning method,which can reach the battery module level positioning requirements.