基于鱼群算法和Hopfield网络的PID参数寻优

Optimization methodology of PID parameters based on artificial fish-swarm algorithm and Hopfield neural network

提出1种融合了人工鱼群算法与Hopfield神经网络的PID参数优化算法.该算法前期利用鱼群算法快速随机的群体性全局搜索能力生成问题较优的可行解域,后期利用Hopfield神经网络硬件易实现简单快速的优点得到最优解,有效弥补了Hopfield网络对初始值过于依赖容易陷入局部极值的缺陷.将该算法用于某发动机PID控制中的参数寻优,结果表明新混合算法的整定效果好于Hopfield神经网络,且该算法简单易实现.

This paper proposes a new algorithm combining Artificial Fish-Swarm Algorithm （AFSA） with Hopfield Neural Network （HNN）. This new algorithm utilizes the fast and stochastic global searching capacity of AFSA to find the relatively excellent feasible solution region in the previous period and then finds the optimal solution by using HNN＇s advantages of being simple and fast, so as to make up for the deficiency of HNN being prone to fall into local extremum due to its overdependence on initial value. The proposed algorithm is applied to optimize PID parameters of a particular engine. The results show that the tuning availability of the new algorithm is better than the HNN and it is a simple and feasible but effective optimization methodology of PID parameters.