Parallel Differential Evolution Based on Distributed Cloud Computing Resources for Power Electronic Circuit Optimization

Abstract

Power electronic circuit (PEC) design and optimization is a significant problem in engineering area. Due to its complexity, evolutionary computation algorithms such as differential evolution (DE), genetic algorithms, and particle swarm optimization have been used successfully to obtain optimal components for PEC. However, since the fitness evaluation of PEC is often very expensive, these methods are computationally demanding and cannot easily be used for real time control or large scale problem. Therefore, finding a simple and powerful method to reduce the computational time is an important work. In this paper, a distributed parallel DE (PDE) is proposed to implement on a set of distributed cloud computing resources in order to accelerate the computation. The experimental results indicate that more computational resources for parallel implementation can indeed help to reduce the computational time efficiently. Therefore, the PDE paradigm significantly speeds up the computation for expensive fitness evaluation, making it more suitable for complex optimization problems in big data environments.