Research On Connection Fault Diagnosis Method Of On-board Lithium-ion Power Battery Pack

Vigorous development and promotion of new energy vehicles is one of the important ways to solve the energy crisis and environmental pollution problems in China.Thanks to the strong support and encouragement of the national policies,new energy vehicles and their power batteries are developing rapidly and are attracting widespread attention and research.Lithiumion batteries are increasingly used as the main energy storage component and power source for new energy vehicles due to their high energy density,long life,and low self-discharge.However,battery is a very complex electrochemical system,with strong nonlinear characteristics.Mechanical stress and aging of materials are likely to lead to battery connection faults,manifested by abnormal increase in contact resistance or even loosening of the battery unit,which will lead to rapid decline of battery capacity and continuous accumulation of local heat.If the fault is not diagnosed and removed timely,serious battery safety problems such as thermal runaway and explosion may occur.Aiming at the connection fault diagnosis of battery packs,this paper analyzed the connection fault characteristics of parallel battery packs and series battery packs respectively.The data-based battery pack connection fault diagnosis methods are proposed without increasing the hardware cost,and the connection fault experiments of different degrees are designed to verify the validity and reliability of the diagnosis methods.The main research contents of this paper include:Firstly,the internal structural composition and working principle of lithium-ion battery are analyzed.The inconsistency of lithium-ion battery packs,formation methods and connection fault characteristics are studied.The effect of battery formation mode and inconsistency on connection fault diagnosis is considered to provide reliable theoretical support for the research of connection fault diagnosis.Aiming at the lack of cell current information in parallel battery packs,a connection fault diagnosis method based on the current distribution estimation of the parallel battery pack was proposed.The long short-term memory(LSTM)neural network is used to estimate the internal current distribution of the battery pack that is hard to be measured directly,through the terminal voltage,total current and state of charge that can be measured or estimated by battery management system.The estimated current distribution can be used to diagnose different connection faults inside the parallel battery pack without extra cost and complex battery models.Based on the estimated current distribution,two residuals are generated to diagnose the connection fault and its degree with better robustness.Multiple sets of experiments are designed for different types of connection faults to validate this method.Experimental results show that the proposed method can achieve an accurate and effective diagnosis of connection faults.Aiming at the problem of diagnosis and quantification of connection fault for series battery packs,a connection fault diagnosis method for series-connected battery packs based on the combination of similarity calculation of charging curve and conversion of charging quantityvoltage(QV)curve is given.Based on the edit distance on real sequence(EDR),the similarity of the charging curve of the battery cell was calculated,so as to determine the location of the faulty cell and the preliminary quantification of the fault degree.Then,by calculating the QV curve conversion parameters,the corresponding contact resistance value can be calculated and distinguished from the internal resistance.Finally,a set of experiments were designed on seriesconnected batteries to simulate the connection faults with different severity to validate the proposed method.The experimental results show that the method can effectively locate the connection fault and diagnose its severity in the series-connected batteries without a complex battery model.Aiming at the influence of the battery inconsistency on the connection fault diagnosis in series battery packs,a fault diagnosis method based on principal component analysis(PCA)was studied to diagnose and isolate battery connection faults and inconsistency fault.By establishing PCA statistical model of series battery pack,the squared prediction error(SPE)generated by measurement data in PCA model was calculated as the fault index,and the connection fault and inconsistency fault of battery pack could be distinguished by using the numerical variation rule and shape of SPE curve.Then,by calculating the contribution of each cell to the SPE,the degree of it deviates from the normal level can be determined respectively,thus to locate the faulty cell.Finally,the connection fault and inconsistency fault experimental data of the series battery pack with eight cells were used to verify.Experiment results show that this method can realize accurate fault diagnosis and fault location.