摘要
为了提高模型效率,更好地反映实际系统的动态特性,根据氧乐果合成过程特点确定了PSO-回归BP网络结构.采用惯性权重动态调整的粒子群算法进行初始寻优,并基于改进的BP算法对优化的网络权阈值进一步精确优化,建立了氧乐果合成过程的PSO-回归BP网络模型.仿真结果表明,所建模型误差小、收敛速度快、网络泛化能力强,能更好地反映实际对象特点.
In order to improve the model efficiency and show dynamic characteristic of the system, the modeling method of PSO-recurrent BP network for omethoate synthesis process was studied. Firstly, the structure of PSO-recurrent BP network was determined according to the features of the object. Secondly, PSO algorithm was used to optimize the weight and threshold of BP neural network. Finally, the improved BP algorithm was used to train the pre-optimized weight and threshold for getting further accurate parameters of the model. The simulation results showed that this model not only had small error, fast convergence speed and strong ability of network generalization, but also show characteristics of the actual object well.
出处
《郑州大学学报（理学版）》
CAS
北大核心
2011年第3期113-117,共5页
Journal of Zhengzhou University:Natural Science Edition
基金
国家自然科学基金资助项目
编号60774059
关键词
粒子群算法
回归BP网络
氧乐果合成
温度对象
PSO algorithm
recurrent BP neural network
omethoate synthesis
temperature object
