Multi-velocity particle swarm optimization and its application in soft sensor
多速微粒群优化算法及其在软测量中的应用

Multi-velocity particle swarm optimization algorithm (MVPSO) is an improved PSO algorithm, which can overcome the shortcomings of easily getting in the local extremum. Besides, this improved algorithm is characterized by clear definition, simple operation and easy realization. Both MVPSO and PSO were used to resolve several typical test function optimization problems. The results indicate that MVPSO has higher efficiency, better performance and more advantages in many aspects than PSO. MVPSO was applied to train artificial neural network to construct a practical soft sensor of product concentration of penicillin fermentation process. The experimental results indicate that this method is feasible and effective in soft sensor application.