摘要
在软测量建模中,最常见的非机理建模方式就是利用神经网络进行建模,而近年来兴起的粒子群算法目前已应用于神经网络的训练。在对粒子群算法提出改进方案后,提出了基于改进的粒子群算法的前馈神经网络训练方案。然后再将神经网络应用到焦化装置分流塔柴油95%点软仪表模型参数估计中,得到了满意的结果,可以满足工业过程中的实际需要。
Nowadays, Artificial Neural Networks （ANNs） was widely used for soft sensor modeling. The Particle Swarm Optimization （PSO）, a new algorithm, gained its popularity in ANN training. A BP neural network training approach based on improved PSO algorithm was represented. And the BP neural network was applied in parameters estimation of soft sensor models of diesel 95 % point in coke sets. The result showed that the proposed technique satisfied the requirement of industrial process.
出处
《广州化工》
CAS
2009年第2期40-42,共3页
GuangZhou Chemical Industry
关键词
粒子群算法
神经网络
软仪表
particle swarm optimization
neural networks
soft sensor
