基于改进粒子群算法的解耦控制研究与仿真

2021-06-11 10:58:34陈晓全
计算机时代 2021年5期

陈晓全

摘  要: 针对标准粒子群算法优化多变量系统的解耦控制时存在陷入非全局最优值的问题,设计出一种基于自适应变异的粒子群算法优化多变量系统解耦控制器的方法。该方法以标准粒子群算法为基础,在种群进化过程中引入变异操作,对不同进化程度的粒子以不同的概率进行更新,其中没有达到个体最优的粒子以随机的大概率进行位置与速度的初始化,对已经早熟的粒子以一定的小概率更新进化路径,以此来提高种群搜索全局最小值的能力。种群寻到最优值的状态就是控制器效果最好的状态,利用新的网络模型即可解决上述问题。经过对比仿真,验证了该方法的可行性。

关键词: 多变量系统; 解耦控制; 粒子群算法; 自适应变异; 全局最优值

中图分类号:TP391.9          文献标识码:A     文章编号:1006-8228(2021)05-68-04

Research and simulation of the decoupling control based on the

improved particle swarm optimization

Chen Xiaoquan

(Jingjiang College, Jiangsu University, Zhenjiang, Jiangsu 212000, China)

Abstract: Aiming at the problem that the standard particle swarm optimization, which optimizes the decoupling control of the multivariable control system, has a problem of falling into non-global optimal value, a method of optimizing the decoupling controller based on the adaptive mutation particle swarm optimization is proposed. Based on the standard particle swarm algorithm, this method introduces mutation operations during the evolution of the population to update particles with different degrees of evolution with different probabilities. If some particles have not reached individual optimality, their position and velocity will be initialized with a random high probability, while the premature particles' evolution path will be updated with the certain low probability in order to improve the population's ability to search for the global minimum. The state where the population finds the optimal value is the state where the controller has the best effect, and the above problem could be solved by using the new network model. Finally, the feasibility of the method is verified by the contrast simulation.

Key words: multivariable system; decoupling control; particle swarm optimization; adaptive mutation; global optimal value

0 引言

在現代控制系统中,随着控制量的不断增多,系统的耦合性与复杂性越来越强,应用在解耦控制领域的方法也是层出不穷,从最初的对角矩阵解耦到现在的神经网络,不同方法所具有的优缺点也很明显。在工业应用中,PID神经元网络一直是主流应用,但是,面对更加复杂的系统,PID网络的性能已经不能满足工业需求,因此,需要对传统网络模型优化,才能改进控制效果。文献[1]给出的方法为遗传算法,利用遗传算法中种群迭代进化的思想寻找出最优值,并以此来PID神经元网络。文献[2]中给出的方法为PSO,利用各粒子之间的合作与竞争来寻得最优解,并以此来优化PID网络,结果表明该方法在一定范围内达到了优化效果。……

登录APP查看全文