Study On States Estimation And Equalization Management Of Electric Vehicle Battery

During the development of the electric vehicle(EV), energy storage elementshave always been the bottleneck which hinder the development of EVs. Powerbatteries as the energy storage elements of EVs, we need to know their SOC in thedriving process real-timely. The battery is a nonlinear dynamical system, and thebattery model parameters also will be changed affected by temperature, aging andother factors in the using process, so it is very difficult to estimate the battery SOConline. We in this paper use adaptive Kalman filter to estimate the battery states,which not only can identify the model parameters of the battery real-timely, but alsocan improve the estimation accuracy. The EV batteries are used in groups, in order toimprove their use efficiency and prolong their lives, equalization management ofbattery pack is necessary. The balance control strategy based on expert system is usedin this paper, which can reduce system’s equalization energy loss and improvesystem’s balancing speed. Through the study, we have achieved the followingachievements:1. The method of statisticing battery outside signals of current and terminalvoltage is used to estimate battery SOH. Battery external terminal voltage and currentsignals can be obtained conveniently, and we in this paper do statistical analysis forthem from three angles, and find that battery outside signals have statistical law on thebattery SOH. The above conclusions provide a new way to estimated battery SOH.2. Multiple model adaptive method is used to estimate battery SOC. When usingKalman filter to estimate SOC, the estimation precision is affected greatly by theaccuracy of the battery model. The battery model parameters will be changed with thebattery SOH and temperature, so the traditional Kalman filter will increase theestimation error. According to the problem, we in this paper use the multi-modeladaptive method and the estimation precision is improved. The multiple modelsKalman filter is selecting several different SOH battery within the SOH range to buildmodels, and designing Kalman filters based on each battery model, and parallelestimating battery SOC using each filter, and calculating the weight of each model,and the sum of each model weighted SOC estimation is the final estimation.3. Adaptive unscented Kalman filter is used to estimate battery SOC and ohmicresistance. Using unscented Kalman filter doesn’t need to linearize the system, which not only can reduce the computation but also can improve the estimation precision.We in this paper use unscented Kalman filter to estimate battery SOC, and useextended Kalman filter to identiy battery ohmic resistance. The two simultaneousfilters are formed into iteration algorithm, and it can update battery model parametersin real time and improve the accuracy of the model, thereby improving the estimateprecision of the battery SOC. Since battery ohmic resistance can characterize thebattery SOH, we can further estimate the battery SOH.4. Battery pack non-energy loss voltage balance control strategy based on expertsystem is used in this paper. The aim of battery voltage balance control is to maintainthe batteries be the same terminal voltage during work process. The principle ofbalance control is to reduce energy consumption and improve the balancing speed inthe equalization process. We in this paper take a switched capacitor equalizing circuitfor example, analysing the relationships among equalizing circuit capacity, switchingfrequency and battery working current, and provide a theoretical analysis method fordesigning equalizing circuit.