Enhancing Battery SOC Estimation Method Through ANN-Based Incorporation of SOH Degradation

A massive number of electronic devices that use batteries as the main power source demand to maintain the battery’s safe operation and lengthen its lifespan. To respond to these demands, a state of charge (SOC) estimation has been extensively developed. Nevertheless, the existing SOC estimation methods often overlook the degradation of the battery state of health (SOH), which can lead to inaccurate estimations. This is because crucial parameters in SOC estimation, such as capacity, vary with battery SOH. An artificial neural network is used for modeling actual SOC trajectories and to predict the actual SOC. To improve accuracy, the model adjusts the SOC parameter by updating the initial SOC value using the updated SOH. This addresses the inadequacy of conventional SOC estimation models trained on fresh cells in accurately estimating SOC in aged batteries. Results indicate that the proposed model consistently achieves lower errors, with the largest improvement observed in the estimation of aged batteries. This reduction in errors by up to 45% in mean absolute percentage error highlights the effectiveness of the proposed method. The proposed model leverages measurable battery data for SOC estimation, reducing computational complexity. By utilizing computationally less complex measurable battery data, the model can be implemented on edge devices, alleviating the computational burden and complexity associated with traditional estimation methods. This further enhances the overall performance and reliability of battery-powered electronic devices, meeting the demands for efficient battery management in practical scenarios.