SOC Estimation Of Lithium Battery Based On Dual-filter Structure Of APSO-MUKF

In order to alleviate the increasingly serious environmental problems,countries around the world are actively promoting the development of the new energy automobile industry.In the new energy vehicle industry,power battery management technology is the core and key to measuring battery life and ensuring vehicle safety.Therefore,accurate and efficient battery SOC estimation is of great significance.In this study,lithium-ion batteries were selected as the research object.Aiming at the problems of inaccurate SOC estimation and uncertain noise after long-term use of batteries,this study proposed an adaptive particle swarm optimization untracked Kalman double filter structure algorithm under different time scales based on the second-order RC equivalent circuit model and unscented Kalman Filter.The battery state and battery parameters（including capacity,internal resistance,etc.）are respectively estimated by micro and macro estimators,which are input to each other,so as to achieve a more accurate estimation effect.Moreover,the accuracy and robustness of battery SOC estimation are analyzed and verified through battery dynamic condition experiments.The main work of this study is as follows:The structure and working principle of lithium-ion battery were briefly introduced at first,then the research object was determined and the battery test platform was built,and the battery basic characteristic test and battery cycle test were designed.Among the battery basic characteristic tests,mainly completed the maximum available capacity test,battery capacity influencing factors test（under different temperatures and different discharge rates）,open circuit voltage characteristic test,hybrid impulse characteristic test and dynamic condition test.According to the experimental results,the characteristics of the battery were analyzed,which laid a foundation for the subsequent battery modeling and SOC estimation.Secondly,the common battery models were briefly described,and their advantages and disadvantages were analyzed.By weighing the accuracy and complexity of the battery model,the second-order RC equivalent circuit model was selected.After determining the battery model,the MATLAB toolbox and the recursive least square method were used to identify the battery model parameters offline and parameter identification.Finally,the accuracy of parameter identification was verified by the experimental data of battery dynamic conditions.In the off-line parameter identification,the error between the battery test voltage and the battery simulation voltage is less than 40mV.In the online parameter identification,the error between the battery test voltage and the battery simulation voltage is less than 20mV,which proves the accuracy of the online parameter identification results.Then,the principle of untracked Kalman filter is introduced.In the process of estimating the battery SOC,particle swarm optimization algorithm is introduced to solve the problem that the battery untracked Kalman filter is difficult to obtain the error covariance of the optimal solution.At the same time,the particle swarm inertia weight is selected incorrectly,which leads to missing the optimal solution or falling into the local optimal solution.In this study,a self-use dynamic adjustment of inertia factor is proposed to form APSO-UKF algorithm,which realizes the estimation of battery SOC.The accuracy of the algorithm is verified by DST and UDDS dynamic experiments.Experimental results show that APSO-UKF method has higher estimation accuracy than UKF method,and the algorithm does not increase the calculation pressure,and does not affect the online estimation.Finally,in view of the sources of error in battery SOC estimation after battery aging,this study proposes a joint estimation of power battery SOC and capacity parameters at different time scales,and a macro filter is used to estimate battery parameters with slow time-varying characteristics（capacity C,internal resistance R₀,etc.）,A microscopic filter estimates the battery SOC with faster time-varying characteristics,and the two alternate and input each other,achieving better estimation performance.Through the verification of battery capacity and SOC estimation under different time scales and different cycle times,the analysis shows that compared with a single time scale,the estimation method of different time scales can obtain a more accurate estimation effect.Through the verification and analysis of dynamic operating conditions under different cycles,the estimation error of battery SOC can be kept within 4%,which can improve the estimation accuracy of battery SOC after battery aging and has good robustness.