基于搜寻者优化算法的PID神经网络解耦控制  

PID neural network decoupling control based on seeker optimization algorithm

作　　者：刘俊[1] 

机构地区：[1]商洛学院电子信息与电气工程学院,陕西商洛726000

出　　处：《工业仪表与自动化装置》2015年第5期97-100,共4页Industrial Instrumentation & Automation

摘　　要：传统的PID神经网络,由于初始权值随机选择,权值学习采用BP算法,所以容易陷入局部极值,进而导致该方法无法得到高精度的控制结果。该文提出采用搜寻者优化算法优化PID神经网络初始权值,再把最优初始权值带入PID神经网络,实现解耦控制。对一个耦合系统进行仿真实验,结果表明,与目前控制效果较好的粒子群算法优化PID神经网络相比,该算法收敛速度更快、稳态误差更小,同时也具有良好的自适应和抗干扰能力,能够实现快速、高精度、稳定的解耦控制。Traditional PID neural network , in which initial weights are randomly selected and weight learning method uses BP algorithm , tends to fall into local extremum , so the method cannot get precise control.Seeker optimization algorithm（SOA） is adopted to optimize the initial weights of PID neural net-work, and the decoupling control is realized by putting the optimal initial weights into PID neural net -work.A coupling system is simulated , and the result shows that , compared with PID neural network which is based on particle swarm algorithm （ PSO） , PID neural network which is based on SOA （ SOA-PIDNN） shares better control effect and smaller steady -state error;moreover, SOA-PIDNN is endowed with adaptive and anti -interference ability , which facilitate rapid , accurate and stable decoupling con-trol.

关 键 词：搜寻者优化算法 解耦控制 神经网络 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]