A Robust Controller Design Method Based on Parameter Variation Rate of RBF-ARX Model

Abstract

As an extension of the exponential autoregressive model and radial basis function (RBF) network, the RBF-ARX model has been widely used in nonlinear system modeling and control. Considering conservativeness of the previous method, which only uses the upper and lower limits of the RBF-ARX model parameters to construct a system's polytopic state space model, in this paper, the model's parameter variation rate information is also utilized to compress variation range of the coefficient matrices in the system's state space model. And then, a robust predictive control (RPC) strategy for output tracking without using system's steady state information is designed. The method of constructing the system's polytopic state space model takes advantage of the fact that the RBF-ARX model itself is a special quasi-LPV model, and there is no need to assume the time varying parameters and/or the variation rate of the parameters in the system model are known or measurable. The effectiveness of the proposed control strategy is verified on a continuous stirred tank reactor (CSTR) process.