Providing Effective Control of Processes with Variable Delay Response

摘要

Many important processes are characterized by a response that is dominated by transport delay associated with plant throughput. When such a process is operated over a wide throughput range, then the changes in response delay will impact control performance. Model Predictive Control (MPC) may be used to address the process dynamics, interactions and constraints associated with multi-variable processes. However, a MPC controller is generated from a process model identified through process testing. Thus, the control performance degrades if the process response changes from that assumed in the process model. A common approach for addressing changes in process dead-time is to identify multiple process models and generate different MPC controller for each situation or range of operation. The control system must then switch between models and controllers for each process condition. The maintenance of multiple models and associated control often complicates MPC commissioning and maintenance. In this paper we present a new approach in MPC design that explicitly addresses the control of variable dead-time processes, by adapting the MPC execution rate to the process condition. Several processes with variable time delay will be presented that illustrate the potential of this approach for industry applications.