An Energy Compensation Method With Dropout Technique Based on Data-Driven Approach for SOE Estimation of Lithium-Ion Batteries

Accurate state of energy (SOE) estimation of lithium-ion batteries (LIBs) is critical for providing feedback on the remaining range of electric vehicles, which is one of the most concerning metrics for drivers. The power integration (PI) method is a simple approach for SOE estimation and serves as the foundation for other modeling techniques. The performance of the PI method, however, is limited by its cumulative error. To tackle this issue, this article proposes an SOE estimation method with energy compensation, and the method employs a data-driven approach to reduce the cumulative error of the PI method. The causes of the cumulative error are first analyzed, and a variable maximum energy metric is presented to improve the accuracy of the cumulative discharge energy calculation. The parameters characterizing the energy errors between the maximum energy and the reference energy are extracted via data-driven methods. To use data-driven parameters for reducing cumulative error, an energy-compensated SOE estimation method is proposed by designing the scaling mechanism that considers multiple discharge factors. In addition, a dropout technique is introduced into the compensation method to prevent overcorrection under unknown working conditions. The experiments validate the effectiveness of the proposed method under complex working conditions, with satisfactory estimation results.