基于改进型模糊聚类的模糊系统建模方法

结合减法聚类和模糊C均值聚类，提出了一种改进型聚类算法，加快了收敛速度．利用改进后的算法对模糊系统输入或输出的样本集聚类，对聚类结果采用Trust-Region法拟合高斯型和S型函数，以实现模糊系统输入、输出空间的划分和隶属度函数参数的确定．结合MATLAB的模糊和曲线拟合工具箱，详述了如何在标准算法上进行改进和模糊系统建模．通过对IRIS标准数据聚类实验以及在解决机械加工误差复映问题上的应用，验证了改进后算法和建模方法的有效性．

Modeling method of fuzzy inference system based on improved fuzzy clustering arithmetic

An improved clustering method is presented by combining subtractive clustering and fuzzy C-means clustering(FCM) to speed up the convergence.Swatch collection of fuzzy inference system input or output is clustered using improved clustering method.Division of fuzzy inference system input and output space and determination of member function parameters are realized by fitting gauss type function and S type function with trust-region method.The improvement of the arithmetic and modeling of the fuzzy inference system on the basis of standard arithmetic are discussed with fuzzy toolbox and curve fitting toolbox of MATLAB.Improved clustering method and model method are validated by clustering experiments with IRIS standard data and application in solving error reflection in machining.