一种基于遗传算法的高木-关野型模糊模型及其应用

A Takagi-Sugeno Fuzzy Model Based On Genetic Algorithms and Its Application

文章提出了一种基于遗传算法建立高木-关野型(T-S)模糊模型的编码方法.该方法将模糊模型的输入变量、规则的选择及结构、对应于任意规则中每个输入变量的隶属度函数的中心位置及宽度同时编码进染色体中,并在遗传算法中进化.为了验证此方法的有效性,文中通过此方法对磁流变(MR)阻尼器的逆动力学模型进行了训练,并通过仿真对训练得到的模糊模型的精度与训练数据进行了比较.结果表明,通过此方法得到的进化T-S模糊模型可以实现高非线性模型的参数辨识,且具有较少的输入变量、规则以及较高的精度.

A new encoding scheme for modeling a Takagi-Sugeno（T-S） fuzzy model based on genetic algorithms is proposed in this paper. In this scheme, the chromosome includes the number and structure of input variables and rules, the centers and widths of the antecedent membership function to every input variable in every rule, and all these structures and parameters are evolved simultaneously. In order to validate this method, the inverse dynamics of Magneto-rheological（MR） damper is trained, and the accuracy of the resulting fuzzy model and the training data is compared through simulation. The result shows that the parameter identification in high nonlinear model can be achieved by the resulting fuzzy model with less input variables, rules and higher accuracy.