<p>The state of charge (SOC) of lithium-ion batteries is a critical parameter for ensuring safe and stable battery operation. Therefore, accurate estimation of lithium-ion battery SOC is required. The forgetting factor recursive least squares (FFRLS) algorithm and extended Kalman filter (EKF) algorithm are widely applied in SOC estimation. The selection of initial parameters matrix to be identified in the FFRLS algorithm and initial noise covariance matrices in the EKF algorithm directly affects the accuracy of SOC estimation. However, determining optimal initial matrices is particularly challenging. To more accurately estimate the SOC of lithium-ion batteries, the pattern search algorithm is implemented using the patternsearch function in MATLAB. This approach optimizes arbitrarily selected initial parameters matrix to be identified and initial noise covariance matrices. After optimization, the optimized initial matrices are used to perform online parameter identification and SOC estimation respectively. The simulation results demonstrate that after optimizing the initial matrices, the average SOC estimation accuracy under different temperature environments improved by 79.82% for Dynamic Stress Test (DST) and by 80.20% for Federal Urban Driving Schedule (FUDS) that simulates city driving environments. This optimization provides assurance for the accuracy and stability of SOC estimation using the FFRLS algorithm, its improved variants, and the EKF algorithm.</p>