On the filtered Smith predictor with feedforward compensation

• We propose a general structure for dead-time and measurable disturbance compensation.
• Simple tuning rules for typical industrial models are proposed.
• Discrete implementation of the controller is discussed to ensure internal stability.
• Performance benefits are shown in simulation including three industrial processes.

This paper discusses about a general implementation structure of the filtered Smith predictor for single-input single-output processes with a measurable disturbance. Dead-time free models are used to feed-forward the measurable disturbance into the feedback controller. It is demonstrated that the proposed controller can be applied to control open-loop unstable processes with different time delays associated with control and disturbance signals. Different simulated industrial processes are used to illustrate the proposed strategy properties and advantages over other time-delay compensators.