An Improved Pelican Optimization Algorithm - Kernel Extreme Learning Machine for High Accurate State of Charge Estimation of Lithium-Ion Batteries in Energy Storage Systems

Abstract

The accurate estimation of the state of charge (SOC) of lithium-ion batteries is of great significance in real-time monitoring and safety control of batteries. To solve the problems of difficult real-time estimation and low estimation accuracy of lithium-ion batteries, this paper takes terpolymer lithium-ion batteries as the research object and optimizes the kernel extreme learning machine based on the Pelican optimization algorithm (POA) improved by fusion Tent chaotic and weight factor and Cauchy variation and sparrow search algorithm (TWCS-POA-KELM) to estimate the SOC of lithium-ion batteries. First, aiming at the randomness of the initial population of the basic Pelican search algorithm, Tent chaotic mapping is used to increase the diversity of particles. Secondly, to enhance local optimization capability, the inertia weight factor was improved by using a nonlinear inertia weight factor ω to adjust the pelican's position and update the correlation with the current pelican's position information. At the same time, to address the issues of pelican diversity and the weak optimization capability of the algorithm, this paper introduces a Cauchy mutation strategy and a sparrow warning mechanism in the second stage, enhancing the optimization performance of the pelican algorithm in this phase. Finally, the TWCS-POA-KELM model was constructed on the kernel extreme learning machine (The model is abbreviated as the IPOA-KELM model.). The experimental results demonstrate that the IPOA-KELM model accurately estimates the SOC of lithium-ion batteries, with MAE, RMSE, and MAPE under BBDST conditions at 0.143%, 0.172%, and 1.344%, respectively, indicating strong tracking ability and robustness.

Keywords: Lithium-ion battery, State of charge, Improved pelican optimization algorithm, Kernel extreme learning machine, IPOA-KELM model