Hybrid Intelligent Dynamic Optimization of Switched Systems

In this article, a hybrid intelligent dynamic optimization method integrating of particle swarm optimization (PSO) and gradient-based optimization (GBO) is proposed for optimal control of switched systems. The method both improves the solution accuracy of PSO and avoids falling into local minima generated by the deterministic optimization. First, the PSO algorithm with ring topology is used to explore the whole search space to detect the global optimum area. Second, the GBO algorithm is deployed in the detected global optimum area to achieve faster convergence rate and higher precision solution than those of pure PSO. Finally, the simulation results show that the algorithm outperforms both PSO and GBO in terms of solution accuracy and computational cost.

I. Introduction

Various fields of application, such as fermentation process [1], electrical circuit systems [2], air traffic control [3], computer process [4], and robotic systems [5], are concrete examples of switched systems. Such systems involve a series of discrete or continuous subsystems represented by differential equations and switching laws that determine, which subsystem is active at each moment of time horizon [6], [7], [8], [9], [10], [11].