Robust Predictive Control Algorithm Based on Parameter Variation Rate Information of Functional-Coefficient ARX Model

Considering the conservativeness caused by the polytopic linear parameter varying (LPV) model, which is constructed using only the upper and lower bound information of the state-dependent auto-regressive model with eXogenous input and radial basis function network type coefficients (RBF-ARX model). In this paper, a robust predictive control (RPC) algorithm based on the parameter variation rate information of the RBF-ARX model is proposed. By using the information, the size of the convex polytopic sets used to wrap the system's polytopic LPV model is compressed greatly. Thus, it improves greatly the control performance and reduces the conservativeness of the subsequent RPC algorithm. The conversion from the RBF-ARX model to the polytopic LPV state space model just uses the RBF-ARX model itself, and the derived LPV model is a special quasi-LPV autoregressive model. So, it is not necessary to assume that the time varying parameters and/or the bounds of the parameter variation rate in the polytopic LPV model must be known or measured. An example of a widely used continuous stirred-tank reactor process control is studied to illustrate the effectiveness of the proposed approach in terms of using parameter variation rate information of the RBF-ARX model to improve step response control performance and anti-jamming performance.