Sliding Mode Observation Of Lithium Battery SOC Based On Order Adaptive AR Equivalent Circuit Mode

With the increasingly serious problems of energy shortage and environmental pollution,development of renewable energy has become a mainstream measure to promote sustainable development,vehicle power battery is regarded as a vital part of the global energy strategy.As the core of the energy storage system,the battery management system（BMS）determines the scientific and reasonable use efficiency of the battery and the advantage and disadvantage of vehicle performance.As the core technology of BMS,the state of charge（SOC）of vehicle power battery is a key factor to restrict the development of electric vehicles,accurate estimation of SOC is a prerequisite for stable operation of power system,the guarantee for extend the service life of the drive equipment,promote the improvement of vehicle power battery efficiency of energy conversion and efficacy of control system.Therefore,research SOC of vehicle power battery has practical significance for the theoretical exploration and engineering practice in improve driving mileage and safety performance.This paper conducts in-depth research around the estimation problem for lithium-ion power battery of SOC,the specific work content and related conclusions are as follows:（1）Describe the present research status,select lithium-ion battery as the research object after compare the performance of common battery types,explains the significance of studying the SOC,discuss the key factors affecting the estimation of SOC,systematically research and analyze the existing methods of estimating SOC,lay the foundation for the build model of lithium battery.（2）Aiming at the traditional method for estimating the remaining SOC of a type of vehicle power battery based on the equivalent circuit mode,the estimation accuracy is highly dependent on the model accuracy,and the model accuracy is directly proportional to the model complexity,which is difficult to apply to the embedded control unit problem,Propose a new equivalent circuit model with relatively low complexity and capable of adaptively determining the optimal model order--Gray box model of equivalent circuit of vehicle power battery based on order adaptive autoregressive（AR）model.The demand for fast computing of real vehicle battery management systems was considered,use the double iteration Brockwell-Dahlhaus-Trindade（BDT）algorithm to realize the online parameters identification of the order iteration of the adaptive AR equivalent circuit model,the Kullback information criterion()to determine the optimal order of the model.（3）We will compare the performance of computational complexity,model quality,and computational time for adaptive AR equivalent circuit model and traditional n RC equivalent circuit model in the noise-free and gaussian white noise environment based on AIC.The new equivalent circuit model uses an autoregressive model to replace the Resistor-Capacitance circuit（RC）in the traditional equivalent circuit model to achieve the continuity of the number of parameters to be identified.In this way,compared to traditional equivalent circuit model,the new equivalent circuit model can have better model quality,which means the model is correctly specified and not being overfitted and also have the optimal balance between model computational complexity and accuracy of the model.（4）Based on the above adaptive variable structure AR equivalent circuit model,this paper further presents the process of design the sliding mode observer（SMO）for the SOC of the lithium-ion battery and the process of proving the observability and stability.The A123 battery in the open data set of the University of Maryland CALCE battery experiment was used to compare the model accuracy and the estimated SOC accuracy of the new equivalent circuit model and the Thevenin model under DST conditions in this paper.The A123 battery in the open data set of the University of Maryland CALCE battery laboratory was used to compare the model accuracy and the estimated SOC accuracy of the new equivalent circuit model and the Thevenin model under DST conditions in this paper.The verification shows that the SOC estimation method based on adaptive AR-sliding mode observer gray box model proposed in this paper has low model complexity,high accuracy,strong robustness and fast convergence performance.