Research On State Estimation And Fault Diagnosis Of Li-ion Battery In Energy Storage System

As the power source of pure electric vehicles and the core component of intelligent microgrid,energy storage system plays an important role in energy storage,scheduling,conversion and supply.Lithium-ion battery is widely used in energy storage system due to its high energy density,high output voltage,long cycle life and no memory effect.With the extensive use of lithium-ion batteries,the electrical behavior characteristics,temperature sensitivity and aging characteristics of the system become extremely complicated.At the same time,due to the continuous improvement of specific energy of lithium-ion batteries and more complex application scenarios,the safety problems caused by battery failures are becoming more and more prominent.These problems bring great challenges to real-time monitoring and safety control of lithium-ion battery energy storage system.Therefore,research on lithium-ion battery state estimation and fault diagnosis methods,to ensure the efficiency and reliability of lithium-ion battery operation,and improve the security of the energy storage system has an important theoretical support and application value,but also for the reliable operation of the intelligent micro grid and pure electric vehicles promoting strong power.In this paper,the state estimation and fault diagnosis of commercial lithium iron phosphate batteries were the main research objects,and the following research work is carried out:In order to solve the problem of on-line identification of electrical behavior model parameters and the low robustness of State of Charge(SOC)estimation method for lithium-ion battery under the condition of variable temperature,this paper firstly based on Thevenin equivalent circuit,an Alternative Generalized Least Squares with Forgetting Factor(FF-AGLS)parameter identification method is proposed.The model parameter identification accuracy in low temperature and low SOC region is improved.Then,a SOC estimation method based on Singular Value Decomposition(SVD)H∞ robust volume Kalman Filter(CKF)is proposed.Solve the problem of SOC estimation accuracy decline in wide temperature region and working conditions;Finally,experiments show that the proposed method can accurately estimate the SOC of lithium-ion battery under the condition of variable temperature.Incremental Capacity Analysis(ICA)is firstly introduced in this paper to solve the problem of modeling the aging behavior of lithium-ion batteries and accurately estimating the State of health(SOH)under the condition of small samples.The relationship between IC curve and battery aging mode is analyzed.Then,the relationship between Health Indicator(HI)and battery Health status is quantitatively analyzed by using grey correlation analysis method.At the same time,an information dimension reduction method based on Weighted Principal Component Analysis(WPCA)is proposed to solve the interference of redundant information on battery modeling accuracy.Finally,in order to improve the generalization performance of SOH estimation under the condition of small sample data,this paper proposes a Structural Weighted Twin Support Vector Regression(SW-TSVR)algorithm.The validity of the proposed method is verified with several battery aging data samples.Aiming at the SOC estimation problem of the whole life cycle of lithium-ion batteries,a battery state space model with temperature change and aging factors is established.Based on this model and Interacting Multiple Model(IMM),a joint estimation scheme for SOC and SOH of lithium-ion batteries is proposed.On the basis of the battery data of different aging degrees and temperatures,the IMM-based SOC and SOH joint estimation algorithm is verified by experiments,which proves that the proposed method can achieve accurate SOC estimation and real-time battery capacity tracking.Aiming at the problem of Fault diagnosis method for lithium-ion battery,an IMM fault diagnosis method(LN-IMM)based on Low Inertia Noise Reduction(LN)is proposed.Based on the analysis of typical battery electrical fault classification and mechanism,the transfer probability correction function based on model probability n-order difference is adopted based on IMM to suppress the external disturbance caused by complex working conditions,reduce the fault false positive rate and improve the fault diagnosis accuracy.The model jump threshold method is used to reduce the inertia of the fault model during the switchover,which can guarantee the fast separation and control of the fault.Finally,the validity of the proposed fault diagnosis method is verified under different working conditions,and it is proved that the proposed method can achieve accurate detection and rapid separation of battery faults.Based on the above work,the state estimation accuracy of lithium-ion battery used in energy storage system is effectively improved and the battery failure is effectively controlled.The research results can be used in engineering practice,and also provide strong support for the further application of energy storage system in the industry.