PARAMETER IDENTIFICATION FOR LUGRE FRICTION MODEL USING GENETIC ALGORITHMS

摘要：

Parameter identification for mechanical servo systems with nonlinear friction term is very difficult, and linear identification techniques are not adoptable because that the parameters can not be linear parameterized as well as the local minimum problem. Based on genetic algorithms, this paper presented a two-step offline method for the parameter identification of mechanical servo embedded with LuGre friction model. In the first step, four static parameters were estimated through the Stribeck curve, and in the second step,two dynamic parameters were obtained by the typical limit cycle output of the system. Genetic algorithms with different control parameters and objective functions were used in both steps to minimize the identification errors. At last, the simulation are developed for a typical nonlinear mechanical servo systems, and the results have shown that the convergence of identified friction parameters are robust and not affected by the coupling property between the dynamic parameters and static parameters.

关键词： Friction Genetic algorithms Servo system Parameter identification

作者: DE-PENG LIU

作者单位: School of science, Hangzhou Dianzi University, Hang Zhou Zhejiang, 310018, China

会议类型: 国际会议

会议名称: 2006 International Conference on Machine Learning and Cybernetics(IEEE第五届机器学习与控制论坛)

会议地点: 大连

会议语种:英文

页码: 3419-3422

在线出版日期: 2006-08-13（万方平台首次上网日期，不代表论文的发表时间）