Improved splice‐electrochemical circuit polarization modeling and optimized dynamic functional multi‐innovation least square parameter identification for lithium‐ion batteries. (9th May 2021)

• Record Type: Journal Article
• Authors: Shi, Haotian; Wang, Shunli; Fernandez, Carlos; Yu, Chunmei; Fan, Yongcun; Cao, Wen
• Is Part Of: International journal of energy research. Volume 45:Number 10(2021)
• Page Start: 15323
• Page End: 15337
• Publication Date: 2021-05-09
• Subjects: dynamic function optimization -- lithium‐ion batteries -- multi‐innovation least squares -- parameter identification -- splice‐electrochemical circuit polarization model; Power resources -- Periodicals; Power (Mechanics) -- Periodicals; Power resources -- Research -- Periodicals; 621.042
• DOI: 10.1002/er.6807
• Languages: English
• ISSNs: 0363-907X

Abstract

Summary: The internal nonlinearity of the lithium‐ion battery makes its mathematical modeling a big challenge. In this paper, a novel lithium‐ion battery splice‐electrochemical circuit polarization (S‐ECP) model is proposed, which integrates the strengths of various lithium‐ion battery models and refines the ohm and polarization characteristics of the electrochemical Nernst model and the differences in charge‐discharge internal resistance. Moreover, by applying the one‐sided limit to the discrete system, a multi‐innovation least squares algorithm optimized based on the dynamic function (F‐MILS) introduced to constrain the original innovation weight is put forward, which tackles the problem of large algorithm errors caused by huge changes in the system input. In order to evaluate the regulating ability of weight constraint factors, the relevant parameter values in the dynamic function are discussed as independent variables. Furthermore, parameters of the S‐ECP model are identified online by HPPC experimental data combined with the multi‐innovation least squares (MILS) algorithm ameliorated by the dynamic function, and the convergence speed of parameters in the identification process is analyzed. Finally, an urban dynamometer driving schedule experiment is carried out on the lithium‐ion battery under more complex working conditions. It is revealed that the accuracy of F‐MILS is 0.5% higher than that of unoptimized MILS, further confirming the accuracy of the S‐ECP model and the superiority of the F‐MILS algorithm. Highlights: A novel lithium‐ion battery splice‐electrochemical circuit polarization (S‐ECP) model is proposed, which integrates the strengths of various lithium‐ion battery models and refines the ohm and polarization characteristics of the electrochemical Nernst model and the difference in charge‐discharge internal resistance. By introducing a dynamic function to constrain the original innovation weight and taking the influence of noise on identification accuracy into account, an optimized multi‐innovation least squares algorithm based on the dynamic function (F‐MILS) is put forward. In order to evaluate the regulating ability of weight constraint factors, the relevant parameter values in dynamic functions are discussed as independent variables. A large number of experiments, including hybrid pulse power characterization and urban dynamometer driving schedule condition experiment, are designed to verify the accuracy of S‐EP model and the superiority of F‐MILS algorithm.