An extended single particle model-based parameter identification scheme for lithium-ion cells

Abstract

The accurate modeling and parameter identification of lithium-ion battery are of great significance in real-time control and high-performance operation for advanced battery management system (BMS) in electrified vehicles (EVs). However, it is difficult to obtain the information about the interior state inside battery, because it cannot be directly measured by some electric devices. In order to accurately identify the key state parameters of lithium-ion cell applied to electric ground vehicles, an extended single particle model of lithium-ion cell with electrolyte dynamics behaviors is first built up based on the porous electrode theory and concentration theory in this article. Compared with the conventional single particle cell model, the parameter description of the solid electrolyte interface film is incorporated into this model, and the coupled effects of temperature-dependent and electrolyte-dependent electrochemical parameters on the cell discharge are also taken into consideration. Based on this extended single particle cell model, a simplified parameter sensitivity analysis method and a comprehensive parameter identification scheme for lithium-ion cell are proposed herein, in which a sensitivity analysis of the capacity to a subset of electrochemical parameters that are hypothesized to evolve throughout the battery's life, is conducted to determine the highly sensitive parameters to be identified under some particular operation scenarios, and further to solve the parameter optimization problem using the genetic algorithm. Based on this method, the test data under the working condition of 1 C discharge rate at 23℃ are employed to evaluate the identified parameters of lithium-ion battery cell with a peak value of voltage error less than 3.8%. Afterwards, the effectiveness and feasibility of the proposed parameter identification scheme are validated by the comparative study of the simulated output voltage and the experimental output voltage under the same input current profile. Specifically, the 0.05 C discharge and HPPC (hybrid pulse power characterization) current profile are used to verify the evaluated parameters under the 1 C discharge condition, and the maximum relative errors of voltage with 0.05 C galvanostatic discharge profile at 23 and 45℃ are 3.4% and 2.6% by using our proposed SPMe_SEI model, and 5.7% and 4.0% by using the traditional SPMe model, respectively. Moreover, the maximum relative errors of voltage with HPPC discharge profile at 23 and 45℃ are 1.9% and 1.5% by using our proposed SPMe_SEI model, and 2.1% and 1.8% by using the traditional SPMe model, respectively. It is concluded that the proposed parameter identification scheme for a lithium-ion cell model can provide a solid theory foundation for facilitating the estimation of state-of-health in BMS application.