Research On The State Estimation For Liquid Metal Batteries

The application of advanced energy storage technology can facilitate the integration of renewable energy into the large-scale grid and improve the efficiency of power systems.As an emerging battery energy storage technology,liquid metal battery has shown considerable potential in the field of electric energy storage applications owing to the advantages of large capacity,low cost and long life.For the safe and efficient utilization of liquid metal batteries,a matching battery management system(BMS)is of great significance.For this new battery technology,it is necessary to study the working characteristics and establish an accurate battery model to estimate the state of charge(SOC)and state of health(SOH)online,and further provide the basis for the operation of the BMS.This paper has carried out a series of work around the state estimation of liquid metal batteries.The specific research contents and results can be summarized as follows:1.A combined model and Thevenin model are established for liquid metal batteries.The parameters are identified offline by linear least squares estimation and the simulation model was further built in Matlab/Simulink.Through the simulation analysis of the battery test,it is judged that the Thevenin model is more in line with the actual dynamics of the liquid metal battery,and thus is more suitable for the work of state estimation.2.The state space is established based on the Thevenin model of liquid metal batteries.The state vector including the battery state and the model parameters is estimated online by using the adaptive extended Kalman filter(AEKF)and the adaptive unscented Kalman filter(AUKF)algorithm.Online estimation of parameters can ensure the fidelity of the model,which in turn enhances the accuracy of SOC estimation.In addition,the adaptive estimation algorithm can update the system noise information in real time,thus improving the stability of the estimation.Experimental results using actual battery test data show that AUKF has the highest estimation accuracy.3.Considering that adding model parameters directly to the state vector will lead to highorder matrices in the calculations,the parameters in Thevenin model are estimated online by recursive least squares method and combined with the AUKF algorithm to a states and parameters co-estimator.This method is able to estimate the required state variables more efficiently.Subsequent battery dynamic tests have further verified its superior performance.4.The BMS prototype for liquid metal batteries was designed and applied to the threecell liquid metal battery pack for functional test,from which the feasibility of the prototype and the effectiveness of AEKF algorithm in the BMS were verified.