摘要
过程神经网络是传统神经网络的拓展,它的输入是与时间有关的函数和过程。过程神经元增加了一个对于时间的聚合算子,使网络同时具有时空二维信息处理能力。从而更好地模拟了生物神经元的信息处理机制。污水处理过程是一个与时间有关的过程,将过程神经网络引入到污水处理软测量中是一种新颖的软测量预测算法。在对基于正交基展开的过程神经网络算法的污水处理软测量仿真研究的基础上,通过分析原网络收敛速度慢等问题,借鉴传统BP的改进算法,在原算法的基础上提出两方面的改进,仿真取得较好的训练结果。
Process Neural Network is an extention of traditional neural network, in which the inputs are function or procedures associated with ＂time＂. Pocess neuron has increased a for ＂time＂get together operator, makes network at the same time have 2 space - time dimensional information processing ability, so to imitate the information processor system of biological neuron better. Sewage disposal is related with ＂time＂,so we introduce the process neural network into sewage disposal . It is a new algorithm for soft measurement. Algorithm research based on function orthogonal basis expansion for Process neural network in soft measurement, through analyzing the low - speed reasons of the network, using the improvement of traditional BP algorithm as referece, put two improvements on the foundation of original algorithm are put forward and better trainning result is obtained.
出处
《计算机仿真》
CSCD
2006年第9期108-111,共4页
Computer Simulation
基金
北京市优秀人才项目
北京市教育委员会科技发展项目(KM200310011040)
关键词
过程神经网络
污水处理
软测量
算法研究
Process neural network
Swage disposal
Soft measurement
Algorithm research
