Feedforward Control with Disturbance Prediction for Linear Discrete-Time Systems

A typical process control system, in most operations, works to regulate the output against the frequent and variable disturbances, while there is rarely a set-point change. Disturbance rejection solely based on the feedback control tends to be poor when the disturbance propagates to the output quickly. With measurable disturbances, the feedforward control has a great potential for better performance. However, the inversion-based feedforward controller is not realizable if it is noncausal. In this paper, a novel feedforward control scheme is proposed with disturbance prediction capability and yields extraordinary improvement under accurate prediction, but its performance deteriorates with the prediction error. To overcome this, an offsetting mechanism is designed and augmented to the former control for the error compensation. Theoretical analysis and design procedure of the proposed scheme are given. Its effectiveness is demonstrated against the existing feedforward control schemes through simulation on typical examples and a case study on a timing belt actuator.