基于遗传算法的伺服系统摩擦参数辨识研究

LuGre摩擦模型能够精确描述摩擦环节的动态特性 ,但由于其高度非线性 ,使得参数辨识非常困难。针对LuGre摩擦模型 ,提出一种新型的基于遗传算法的模型参数两步辨识方法。首先通过Stribeck曲线 ,辨识出摩擦模型中的静态参数 ,然后由系统的稳态极限环振荡响应 ,辨识出摩擦模型的动态参数。在每一步辨识中 ,均采用遗传算法作为优化工具 ,从而避免了采用线性辩识方法时的局部极小问题。对提出的方法进行了数字仿真 ,并通过设计摩擦补偿环节 ,对辨识参数进行验证 ,结果表明了该方法的有效性。

Research on the Parameter Identification of Friction Model for Servo Systems Based on Genetic Algorithms

The LuGre friction model can accurately describe the dynamic characteristic, but because of its high nonlinearity, it is very difficult to identify the parameters of the model. A two step offline method for the parameter identification of LuGre friction model based on genetic algorithms is presented in this paper. In the first step, four static parameters are estimated via Stribeck curve, and in the second step, two dynamic parameters are obtained by the limit cycle output of the system. In each step, genetic algorithms are used to minimize the identification error, thus the local convergence problem existing in the linear identification method can be avoided. Computer simulation results are given, and a friction compensation term is designed to verify the identified parameters.The results show the effectiveness of the proposed scheme.