Disturbance Suppressing Model Predictive Control Using State Space Approach

Abstract

This paper proposes a method to suppress the effects of disturbance using an extended controller of state feedback and state estimating observer. Model predictive control(MPC) is widely applied in industry and safety is the most important issue. One measure for safety control is a fault tolerant control which controls plants safely even when a fault occurs. In many cases, effects of fault are represented by effects of unknown disturbances and to suppress such effects of unknown disturbances is an effective method of fault tolerant control. Hence, this paper considers a control to suppress the effects of disturbance for model predictive control. Also frequently fault occurs at measurement of outputs and feedback loop is cut and this causes the output being unstable and plant being dangerous. If the poles of controller are designed to be stable, then the unstable output is avoided and a method to specify poles of the controller being stable is also obtained for the fault tolerant GPC. Hence this paper gives a control scheme (1) to suppress the effects of disturbances and (2) to prevent output being unstable when feedback is cut. To ensure these controls, this paper proposes two control schemes, (1) to derive a mathematical expression of the effect of disturbance, (2) to design stable poles of the controller. To design the controller poles, this paper uses an extended generalized controller by state space coprime factorization approach. The extended controller brings additional poles in the closed-loop. Hence, to design fault tolerant control, (3) as the third control schemes, this paper also derives mathematical expressions of the cancelled poles of the closed-loop systems and the poles of the controller. Once, these expressions are obtained, then these poles are easily designed by selecting design parameters in the controller.