Controller Design Using ANFIS-Based Estimation Method for Unmodeled Dynamics

---------------------------------------------------------------------***--------------------------------------------------------------------Abstract— A Nonlinear system is represented as a combination of linear part and Unmodeled Dynamics. The unmodeled dynamics is estimated by using ANFIS. At first the unmodeled dynamics is divided into two parts by using differential expansion of the control input at the last time instant, then the two parts are estimated by using ANFIS. This method improves the precision of Unmodeled Dynamics. Second to deal with the input and output of the unmodeled dynamics one to one mapping and regularization techniques are used with an assumption that growth rate of unmodeled dynamics does not exceed its input vector from which the data vector lies inside the compact set. The datum of the system is fully used to obtain the parameters (centers and widths) in membership functions that are considered as network connection weights in ANFIS offline training. These parameters are tuned online to improve estimation convergence rate. Finally the proposed estimation method is applied to nonlinear switching control design. By the simulation results it is proved that the nonlinear switching control that adopts the proposed estimation method provides stability and convergence of the system that exhibits a desired dynamic performance for the closed loop system.