An Novel Battery State Estimation Method with High Robustness Based on Improved DEKF

Jun Wang, Yu Fang, Rui Xiong

Abstract


The accuracy of common battery state estimation method is usually influenced by the temperature input. The capacities at different temperatures were found to be similar when the starting temperature and the final temperature only have subtle differences by carrying out a series of capacity experiments. Based on the finding, the original DEKF algorithm is improved by using the curved surface of OCV-SOC-CAP instead of the one of OCV-SOC-Temperature to estimate battery state without temperature input. Moreover, piecewise debugging parameters of Q and R is used to estimate battery state when the SOC is at high level and low level respectively. Large amounts of data at different temperatures and at different lifecycles are used to verify the accuracy and find the scope of the method. Results show that the error of estimated SOC is lower than 3% when the temperature is between 45℃ and -10℃ and the SOH is between 80% and 100%. Finally, the HIL experiments also verify the effectiveness of the method.

Keywords


battery state estimation, dual extended kalman filter, piecewise debugging parameters, without temperature input, hardware in loop


DOI
10.12783/dteees/iceee2018/27812

Refbacks

  • There are currently no refbacks.