几个非线性逼近器模糊模型的性能比较

Performance Comparison of Several Fuzzy Models Using Nonlinear Approach

模糊控制器可逼近非线性函数,故以非线性目标函数建立模糊模型逼近其 MSE 值,并建立改进模型。再分别用 ANFIS 和遗传算法整定参数,得出改进模型的 MSE 值。在 x1 和 x2 方向上分别取 100 点形成矩阵,重新计算MSE 的值,用列表的形式对各模型的 MSE 与训练样本集的 MSE 比较,能得到综合性能最优解。

According to the theory of any nonlinear function approached by fuzzy controller, fuzzy model was founded with selected nonlinear target function to approach it, and the MSE value was gained, and then the improved model was founded. The parameters were adjusted respectively with ANFIS and Genetic algorithm, educed the MSE value of improved model. 100 point were selected separately to form a matrix in the directions of x1 and x2 of target function, the MSE value was recalculated. MSE of all models and MSE of training sample-set was compared with list form, and the optimum result in all-around capability was gained.