Comparisons of Fitness Functions in Identifying an Electromagnetic Energy Harvester

Abstract

Objective: System identification is a part of necessary preparations for vibration analysis and dynamic control of a complex system which usually integrates with mechanical and electric components, or the so-called mechatronic system. Methology: The fitness functions adopting energy balance modeling with full-state errors are employed in the real-coded genetic algorithm (RGA) to search for the optimal parameters of an electromagnetic energy harvester. Results: During the optimization and refactoring of the RGA, it is found that the fitness function cannot use only one part of the system responses to avoid losing accuracy in system identification. Hence, fitness functions adopting energy balance modeling with full-state errors are suggested to identify system’s parameters, and the best fitness function is found from numerical simulations. Thus, the proposed methodology in this study can also be utilized to analyze any mechatronic system.