Title :
Development and application of an integrated neural system for a HDCL
Author :
Lu, Yong-Zai ; Markward, S.W.
Author_Institution :
Dept. of Inf. Technol., Bethlehem Steel Corp., Burns Harbor, PA, USA
Abstract :
This study presents the development and application of an integrated neural system in coating weight prediction and adaptive control for a modem hot dip coating line (HDCL). The neural system consists of two multilayer feedforward networks (with backpropagation learning) and a neural adaptive controller (with adaptive learning). They perform coating weight prediction, feedforward control and adaptive feedback control respectively. The production environment analysis and the design of the neural system architecture, associative memory, generalization and stability are addressed in detail. The real-time industrial application results are also given to show the benefits through applying an integrated neural control system in a modem hot dip coating line
Keywords :
adaptive control; backpropagation; coating techniques; coatings; feedback; feedforward neural nets; neurocontrollers; process control; weight control; adaptive control; adaptive learning; associative memory; backpropagation learning; coating weight prediction; feedback; galvanizing line; generalization; hot dip coating line; integrated neural system; multilayer feedforward networks; neural system architecture; production environment analysis; stability; Adaptive control; Backpropagation; Control systems; Dip coating; Feedback control; Modems; Nonhomogeneous media; Production systems; Programmable control; Weight control;
Conference_Titel :
American Control Conference, Proceedings of the 1995
Conference_Location :
Seattle, WA
Print_ISBN :
0-7803-2445-5
DOI :
10.1109/ACC.1995.531382
