基于改进动态域遗传算法的神经元控制器参数优化  

Parameter Optimization of Neuron Controller Based on Improved Dynamic Domain Genetic Algorithm

作　　者：鲜阳 谭飞 XIAN Yang;TAN Fei(School of Automation and Information Engineering,Sichuan University of Science Engineering,Yibin 644000,China)

机构地区：[1]四川轻化工大学自动化与信息工程学院,宜宾644000

出　　处：《组合机床与自动化加工技术》2022年第12期36-39,43,共5页Modular Machine Tool & Automatic Manufacturing Technique

基　　金：国家自然科学基金(61902268);四川省科技计划(2019YFSY0045);大学生创新创业训练计划项目(S202010622090,cx2021192,cx2019259)。

摘　　要：针对传统遗传算法的选择、交叉与变异操作容易造成算法全局收敛概率低,求解精度不够等问题,提出了一种基于精英策略的动态域十进制整数编码遗传算法。算法以十进制整数编码遗传算法为基础,通过设计包括基因补全变异、重生变异等变异操作以及精英交叉操作,来提高算法的收敛全局性;设计搜索变量空间同步与独立动态调整策略,提升算法的收敛精度。仿真函数测试表明算法性能优异,同时该算法用于优化挖掘机伺服系统神经元PID控制参数,取得优良的控制品质,说明算法具有很好的工程适应性。This paper puts forward a dynamic domain integer-coded genetic algorithm based on elitist strategy by analyzing the traditional genetic algorithm's limitation in selection,crossover and mutation operation that lead to algorithm's low global convergence probability and low accuracy.Based on the decimal integer-coded genetic algorithm with the elitist crossover strategy,the modified algorithm improves the algorithm′s global performance of convergence by designing mutation operators that include gene completion mutation,rebirth mutation.The design of search domain for synchronous and independent dynamic adjustment improves the algorithm′s convergence accuracy.The overall test of simulation function indicates that the algorithm has sound global performance of optimization,at the same time,the algorithm is used for optimizing neuron PID control system parameters,and good control quality is achieved.This proves that the algorithm has good adaptability to the project.

关 键 词：遗传算法 基因补全 变异 动态域 神经元PID 

分 类 号：TH165[机械工程—机械制造及自动化] TG659[金属学及工艺—金属切削加工及机床]