Ethernet-Based Intelligent Switching of Controllers for Performance Improvement in an Industrial Plant
This paper presents an Ethernet-based intelligent system for remote switching of the controllers of an industrial plant with the objective of on-line improvement of the performance of the plant. The plant considered in the present paper is an industrial fish-processing machine, which operates using one of several adaptive controllers. The scheme utilizes a remote supervisor, which incorporates knowledge-based decision making to continuously monitor the performance of the plant. The performance metrics deduced from the observations are then used to infer the best adaptive controller for the plant under the existing conditions. A knowledge-based system that incorporates both human expertise and analytical knowledge regarding the plant and the controllers is developed. The proposed intelligent switching is implemented in real-time for controlling the hydraulic-actuated cutter of the fish-processing machine. A client/server supervisory control architecture for remote networked-based controller switching is developed. Switching has to be done in such a manner that the transition from one controller to another takes place in a smooth manner. Proper design of the intelligent switching system is key to achieving this objective.