Fuzzy identification based on improved clustering arithmetic and its application

The traditional modeling methods are hard to identify the nonlinear system like in-well environment simulation system which is multivariable, stochastic, strong coupling and large time delay. Thus, it is difficult to express complex system and implement the whole optimal control accurately. This paper proposes a kind of fuzzy identification method based on improved clustering algorithm in connection with the traditional fuzzy C-means clustering algorithm´s defects which are sensitive to the initial value and unable to definite the optimum rule numbers. The method determines initial clustering centers by the subtractive clustering and the validity function, then finds the final clustering centers by the global fuzzy C-means clustering algorithm. Subsequently the suitable area radius by the principle of nearest neighbor is formulated. The system T-S model by weighted recursive least-square method is built finally. In this paper, the temperature model of in-well environment simulation system is proposed to illustrate the method accurate and effective.