Time-Varying Fault Tolerant Control of Nonlinear MIMO Systems

In this study, a recently proposed discretization based observer is designed for the estimation of unknown time-varying fault variable. Using the estimated fault, a model predictive control based fault tolerant controller is proposed and applied for the real-time control of nonlinear multi-input multi-output (MIMO) systems. The main contributions are given as follows. i) An unknown parameter of the nonlinear grey model is estimated to compensate the uncertainty of the nonlinear system. ii) Second, an unknown process fault function of the nonlinear system is estimated and then the model inaccuracy is compensated using the estimated fault function. Fault and unknown parameter estimations are used to improve control performance of model predictive fault tolerant controller. In experiments, time-varying parameter and fault variables are accurately estimated and hence precise tracking performances are provided in real-time.