Model Parameter Identification of Supercapacitors Using Metaheuristic Gradient Based Optimization

Abstract

Energy storage plays an essential role in both conventional and renewable energy systems, serving as a backup power source and maintaining grid stability between load-demand cycles. The effective control of energy transfer between the storage systems and the power source is of the utmost importance. Supercapacitors are notable within the realm of storage alternatives due to their suitability for high-power density applications. These technologies find applications in several domains, such as in the use of regenerative braking systems in electric vehicles and the utilization of burst mode power sources. This study investigates the parameterization of the Zubieta model, which is an electrical circuit model employed for supercapacitors. This is carried out through the utilization of a hybrid metaheuristic gradient-based optimization (MGBO) methodology. The Zubieta model is composed of three RC branches and an additional self-discharge branch, which necessitates the identification of seven parameters. The research compares the modified MGBO (M-MGBO) approach with particle swarm optimization (PSO) and two PSO variations. One approach combines Particle Swarm Optimization (PSO) and (M-MGBO), while the other incorporates a Local Escaping Operator (LCEO) to enhance the creation of positions and prevent convergence to local minima. The evaluation of performance encompassed the assessment of convergence rate, accuracy, and convergence time. The study's findings indicate that the hybrid PSO-MGBO and PSO-LCEO versions outperformed the conventional PSO approach, showing an average enhancement percentage of 51% and 94%, respectively. Additionally, both variants demonstrated a comparable level of effectiveness to the M-MGBO technique. These variations offer an effective approach for estimating the parameters of the Zubieta model, which has implications for the design and implementation of energy storage systems utilizing supercapacitors. This study highlights the potential of hybrid optimization strategies in improving the precision and effectiveness of supercapacitor model parameterization.