基于BP神经网络的模糊参数辨识

Fuzzy parameters identification based on BP neural network

提出一种与TSK模糊模型相似的模糊模型—M-2模型,证明了M-2模型与一个4层前向神经网络是等价的,在此基础上提出基于BP神经网络的模糊模型参数辨别算法,即通过BP神经网络对样本数据的学习,直接从样本数据获取模型参数,建立M-2模糊模型,通过仿真实例验证了该算法的有效性。

A fuzzy model called M-2 model which is similar to TSK fuzzy model is proposed.The mathematical equivalence between M-2 model and a 4 layer feedforward neural network is verified.Then a new algorithm to identify parameters of fuzzy model based on BP neural network is put forward.All parameters required for establishing M-2 fuzzy model are extracted directly from a BP neural network trained by sample data.A simulation example is given to demonstrate the effectiveness of the algorithm.