一种改进的遗传算法及其在系统辨识中的应用  

Improved Genetic Algorithm and Its Application in System Identification

作　　者：边润强[1] 陈增强[2] 袁著祉 

机构地区：[1]天津理工学院自动化系,300191 [2]南开大学计算机与系统科学系

出　　处：《控制与决策》2000年第5期623-625,640,共4页Control and Decision

基　　金：国家 8 6 3CIMS应用基础研究项目!(86 3- 5 11- 945 - 0 10 );天津市自然科学基金项目!(9736 0 0 311)

摘　　要：提出一种基于遗传算法的大惯性系统辨识方法。该方法将动态补偿和遗传算法相结合 ,实现了系统的参数辨识。为避免遗传早熟而进入局部最优 ,对交换和变异操作进行了改进 ,提出一种防止近亲繁殖的交换策略 ,在不明显增加基因操作计算量的前提下 ,有效地避免了基因缺失。仿真结果表明 ,该方法对过阻尼系统的辨识是有效的。A genetic algorithm based approach for the identification of system with big inertia is presented. The dynamic compensation and the genetic algorithm are organically combined in this approach to realize the identification of the system parameters. In order to avoid the genetic algorithms from the premature and local optimum, the crossover and the mutation operators are improved and an incest preventing crossover strategy is proposed, which can obviate genetic loss without significantly increasing the calculation of the genetic operator. The simulation results show that the approach is effective to the identification of over damping systems.

关 键 词：遗传算法 系统补偿 系统辩识 参数优化 

分 类 号：O231.3[理学—运筹学与控制论] O242.23[理学—数学]