The design and evaluation of a PLC-based model predictive controller for application in industrial food processes.

摘要

Model Predictive Control (MPC) is a viable control strategy for industrial processes that display relatively large variations in the process variable, have complex process variable interactions, or display a large amount of process deadtime. The objective of using MPC in manufacturing is to reduce overall process variability, the result being an increase in process accuracy, precision and efficiency. This study focused on the implementation of model predictive control techniques on an industrial sugar cooking process. The goal was to implement a successful MPC solution directly on a programmable logic controller (PLC) rather than on a personal computer (PC). Although there are many commercially available MPC controllers for implementation on a stand-alone PC, to date there are no control packages for realizing model-based control techniques directly on the ubiquitous PLC.; This study implemented and evaluated three PC-based, commercial MPC technologies for the sugar cooking process, and a new model state feedback (MSF) MPC implementation directly on Rockwell Automation's Allen-Bradley ControlLogix® PLC. A standard proportional-integral-derivative (PID) control implementation was used as a baseline for comparing the MPC strategies. There were three main areas on which the overall comparative analysis focused. These comparison areas were the dynamic response of each strategy at startup, including both temperature rise time and overshoot, and the steady-state disturbance rejection capabilities of each strategy.; The test results showed that the MPC strategies controlled the sugar cooking process better than the traditional PID control method in regards to temperature rise time, temperature overshoot, and disturbance rejection based on feed rate disturbances. It was seen that the differences between the various MPC strategies was not significant relative to temperature overshoot and disturbance rejection. The PLC-based MPC strategy was comparable, but not superior, to the PC-based commercial MPC applications. However, this strategy has several benefits such as requiring no external hardware, software, and communications protocols, which may result in a less expensive implementation than the commercial MPC strategies. The PLC-based strategy is also easier and cheaper to maintain because it is developed on the existing, well-known control platform with existing tools.