三群粒子群优化算法及其在丙烯腈收率软测量中的应用

提出了一种三群粒子群优化算法（THSPSO, three sub-swarms particle swarm optimization）．该算法将整个粒子群分为三群，第一群粒子朝全局历史最优方向飞行，第二群粒子朝着相反方向飞行，第三群粒子在全局历史最优位置周围随机飞行．分别将该算法和基本粒子群优化算法（PSO, particle swarm optimization）用于一些常用测试函数的优化问题；结果表明,与PSO相比，THSPSO具有更好的优化性能．然后，用THSPSO训练神经网络，并将其用于丙烯腈收率软测量建模，结果显示了三群粒子群优化算法在丙烯腈软测量建模中的可行性与有效性．

Three Sub-swarms Particle Swarm Optimization Algorithm and Its Application to Soft-sensing of Acrylonitrile Yield

A three sub-swarms particle swarm optimization algorithm(THSPSO) is proposed.This algorithm divides the particles into three sub-swarms.The first sub-swarm flies toward the global historical best particle. The se-(cond sub-swarm flies) in the opposite direction.The last sub-swarm flies randomly around the global historical best particle.Both THSPSO and particle swarm optimization algorithm(PSO) are used to resolve the optimization problems of several widely used test functions,and the result shows that THSPSO has better optimization performance than PSO.Then,THSPSO is employed to train artificial neural network and applied to soft-sensing of acrylonitrile yield.The results indicate that THSPSO is feasible and effective in soft-sensing of acrylonitrile yield.