<p>Accurate state of charge (SOC) estimation is a crucial function of the battery management system to optimize the longevity, safety, and performance of lithium-ion batteries. However, the conventional extended Kalman filter (EKF) method suffers from significant SOC estimation errors due to model inaccuracies under dynamic working conditions and temperature variations. To address some of these challenges, this work proposes an adaptive dynamic correction factor-extended Kalman filter (ADCF-EKF) based on the second-order resistor–capacitor electrical equivalent circuit model to accurately estimate the SOC of lithium-ion batteries across a range of temperatures and working conditions. By introducing an adaptive correction factor, the EKF dynamically switches its reliance between the model and actual measurements using a sigmoid function based on the innovation sequence to intelligently update the priori error covariance matrix. Experimental data from hybrid pulse power characterization tests, dynamic stress tests, and the Beijing bus dynamic stress tests are used to verify the proposed ADCF-EKF method at different temperatures. The comprehensive results show that the ADCF-EKF method achieves an overall optimal mean absolute error of 0.77%, a mean square error of 0.0096%, and a root mean square error of 0.98%, representing error reductions of 18.95%, 40.12%, and 22.83%, respectively, compared to the conventional EKF method. These improvements are consistent under complex working conditions and various temperatures, confirming the method’s robustness for real-world battery management applications.</p>