Study On Short Circuit Diagnosis Of Lithium Ion Power Battery

Lithium-ion power battery system is a key component in battery electric vehicle.To ensure battery safe working,and extand battery service life,it is necessary to equip a battery management system(BMS)to monitor and control battery working.However,power battery is a complicatedly chemical energy storage system.Battery internal state cannot be measured directely with sensors.Therefore,accurate battery model and state estimation algorithm are required to derive battery internal state.Micro-short circuit(MSC)is a latent problem.If it cannot be monitored and managed in time,it will affect battery power perfomence and even induce thermal runaway.Quantitative MSC diagnosis is available to give a reference for fault management.The paper proposes an on-line MSC quantitative diagnosis method for a SeriesConnected Lithium-Ion Battery Packs.An Electrochemistry-based equivalent circuit model(E-ECM)is built based on battery mechanism model analysis.And a Decoupling parameters extraction(DPE)method is investigated to derive more reliable model parameters.The E-ECM has a higher voltage simulation accuracy compared with traditional ECMs.Based on Mean-difference model(MDM),the paper achieves a high state of charge inconsistency estimation for a small series connected battery pack.Then,an on-line MSC quantitative diagnosis method is proposed on the basis of SOC difference varying characters.Finally,a mutual information(MI)model is employed to identify the fault of small capacity and MSC.Firstly,an accurately on-line E-ECM is built.It is helpful to get entire SOC range estimation with high accuracy.Traditional ECMs(e.g.,second-order RC model)cannot describe battery non-linear characters especially at low SOC area.Combining the advantages of battery mechanism model and ECMs,the paper adds an extra part to represent particle solid phase diffusion on ECMs.A DPE method is also introduced which can derive a reliable model parameters reference value.Experiment results show that the proposed E-ECM has high simulation accuracy both under dynamic and constant current discharge working condition.Specifically,it has a remarkable improvement at low SOC area.In addition,the E-ECM can also predict battery rate capability.Next,SOC inconsistency estimation of lithium-ion battery pack based on MDM and extended Kalman filter(EKF)is studied in this paper.Second-order RC model is employed as cell mean model(CMM)to simulate battery pack mean performance.A hypothetical Rint model is employed as cell difference model(CDM)to simulate the difference between cell and battery mean state.The parameters of MDM are identified by particle swarm optimization(PSO)method.EKF algorithm is applied for battery pack mean state estimation and cells SOC inconsistency estimation.Experiment is carried out on a twelve series connected small battery pack.The results show that adopting the proposed SOC inconsistency estimation method can derive a high accuracy ?SOC estimation with low computational effort.Then,battery pack quantitative MSC diagnosis method is investigated.Different short resistances are adopted to simulate varies degree short level.The MSC diagnosis method is validated at difference battery working situation.Through a long time scale,the proposed method for SOC inconsistency estimation can follow the reference value well.Recursive least squares(RLS)algorithm is further applied for on-line MSC quantitative diagnosis.The experiment results show that the fault diagnosis method can calculate short resistance of MSC cells with low computational cost.Finally,the paper proposes a fault identification method based on MI.The essential characters and differences of small capacity cell and MSC cell are analized.A decision tree is established to identify fault types.The identification of CDM parameters is also introduced in this part.The discriminant quantity of MI is adopted here to identify the two similar performance fault based on the trend of?SOC estimation.Experiment results show that the fault identification method is able to separate these two similar performance fault cells.