Acquisition of tuning rules for hot strip looper system based on fuzzy classifier system

In these days, electro mechanical systems are widely automatized in industry. However, the intervention of human is essential to enhance the performance of electro mechanical systems. Therefore, the methods and systems that have a technical substitution of expert's skilled technique are needed for teaching and assisting the unexperienced workers. PID controller of a hot strip looper control system is treated in this study. In hot strip looper control system, control parameters are to be optimized according to the rolled material, because control dynamics are greatly influenced by rolling conditions. We aimed to develop a support technology that autonomously acquires tuning rules like a skillful expert's decision-makings from accumulated operating data. A fuzzy classifier system is used for decision-making and learning hypotheses. A fuzzy classifier system could generate plural rules and pick up available rules from them. IF-THEN descriptions based on fuzzy theory were used for the representation of decision-making rules.