关键词: 粒子群优化, 差分进化, 混合优化算法, 软测量

Abstract: Particle Swarm Optimization（PSO） algorithm and Differential Evolution（DE） algorithm are analyzed.A novel Differential Evolution-Particle Swarm Optimization based hybrid optimization algorithm（DEPSO） is proposed by taking advantage of both PSO and DE.The procedure of optimization is divided into two phases and the particles are divided into two sub-swarms.One sub-swarm searches via PSO and the other searches via DE at the same time.Evolution speed factor of the swarm is introduced in this algorithm and two sub-swarms exchange information in each iteration to avoid local optimum.Experiments on four complex functions with high dimension show that the improved algorithm outperforms traditional PSO and DE in robustness，converging speed and precision，global searching ability.The improved algorithm is applied to construct a soft sensor model for real-time measuring the ethylene yield.Application results show that this model has high prediction precision as well as good generalization ability.

Key words: particle swarm optimization, differential evolution, hybrid optimization algorithm, soft sensing