基于模糊竞争学习的模糊模型一体化辨识

An integrated identification of fuzzy model based on fuzzy competitive learning

提出了一种利用MGS(modified Gram-Schmidt)算法建立非线性系统模型的建模方法,并给出了基于MGS算法的模型结构和参数辨识的一体化方法,即利用MGS正交变换对通过模糊竞争学习的聚类结果进行变换,确定对模型贡献大的规则,删除对模型贡献小的规则,同时对模型中的参数进行估计,实现模糊模型结构和参数的优化.仿真结果表明,提出的方法能够对非线性系统进行模糊建模.

The modeling method is proposed to build the model of nonlinear system by the modified Gram-Schmidt method. An integrated algorithm is used to confirm the structure and the parameters of the model by means of the modified Gram-Schmidt algorithm. The fuzzy competitive learning is transformed to confirm the fuzzy rules by means of orthogonal transform. The modified Gram-Schmidt orthogonal transform is used to acquire the important rules and remove the less important rules. The parameters of fuzzy model are estimated via the proposed method. The structure identification and the parameter identification of fuzzy model are synchronously identified in the proposed algorithm. The structure and parameters of fuzzy model are optimized. With the illustration of the simulating result, the fuzzy model of non-linear system can be built by the proposed algorithm.