Genetic Fuzzy Tension Controller for Tandem Rolling F. Janabi-Sharifi, Senior Member, IEEE J. Liu Department of Mechanical, Aerospace, and Industrial Engineering, Ryerson University Department of Electrical and Computer Engineering, University of Waterloo Toronto, Ontario, Canada, M5B 2K3 Waterloo, Ontario, Canada N2L 3G1 Email: fsharifi@acs.ryerson.ca Email: j5liu@engmail.uwaterloo.ca Abstract-- A fuzzy logic control (FLC) is designed to maintain constant tension for tandem rolling mills. Genetic algorithm (GA) is used as self-organization technique for FLC to locate the optimal parameters. Simulation results verify the superior performance of genetic fuzzy controller. Keywords-- tension control, fuzzy logic, genetic fuzzy system, rolling mill. I. INTRODUCTION Tandem rolling mills usually consist of a number of mill stands arranged in alignment. Various cross-sectional long metal workpieces are reduced step-by-step under high pressure as they proceed through mill stands sequentially [1]. To meet the dimension requirement and regulate mass flow, automatic gage controllers (AGCs) and automatic speed reguators (ASRs) are employed respectively (Fig. 1). Fig. 1 A rolling mill stand A specific problem associated with tandem rolling mills is the presence of tension, a longitudinal force inside the workpiece resulting from the unequal mass flow of two adjacent mill stands. To optimize the performance of AGC and ASR, it is desirable to keep tension constant via addtitional control action. However, resonance effect, i.e., activities of AGC and ASR will incur the variation of tension and in turn tension adjustment will worsen gage and speed control, will perplex the situation especially in harsh noisy environments. II. FUZZY TENSION CONTROLLER Conventional tension control schemes need exact mathematical models and complete knowledge of real- time operation [5]. However, it is difficult to identify rolling processes from the measurement of rolling data because of the complicated characteristics of rolling stands, resonance effect, noisy environment and strong demand for instrumentation. Multivariable controllers, based on advanced control theory, have been proposed. However, they are difficult to be implemented and configured [2]. The situation particularly becomes very difficult for bar and structural mills that are very noisy and polluted and it is either very expensive or impossible to use tension sensors. In practice, human experts (operators) can manipulate the rolling stands and manage the inter-stand tension in almost satisfactory manner by manual intervention. This fact motivated us to formalize expert knowledge as fuzzy IF-THEN rules [3] to develop a controller based on fuzzy logic to circumvent incomplete plant knowledge and inexact operation information. This controller will emulate the way that human brain processes ambiguous information with intuition and experience. Greatly inspired by these merits and the competence of human operators, an attempt is made to apply fuzzy logic to tension control for complex nonlinear rolling process. To eliminate the tension variation, a fuzzy logic controller (FLC) is superimposed on the ASR to adjust the reference speed of ASR. In this scheme, armature current of roll stand driving motor is used as a rough direct indicator of tension between controlled and downstream stands. Therefore terms tension and current