An Integrated Near-Optimum Design of Cold Rolling Mills

摘要

An Integrated Near-Optimum Design Method is introduced which makes a systematical use of the existing techniques which curbs many complexities involved in industrial systems. The method consists of a 'Subsystem' decoupling stage which represents the originally high order model by a set of weakly coupled subsystems of much lower order. The second stage of the design is a 'Time-Scale' decoupling in which the variables are divided into three 'slow', 'basic' and 'fast' categories according to the length of their response times. The slow and fast variables are eliminated and the near-optimum control is obtained by computations involving basic variables only. These two stages of decoupling are readily interchangeable depending on whether the model is linear or nonlinear. The method has been applied to an Allegheny Ludlum Steel's high speed three-stand mill which can represent a 45-variable control system. Both linear and nonlinear dynamic and state space models are presented for the mill. In a digital computer simulation of the nonlinear model, two of the subsystems are studied in detail and an open loop near-optimum control as a power series of a decoupling parameter is obtained. The application of the method to the linear model resulted in an approximate solution of a 17th-order Riccati equation hence a state regulator feed-back gain matrix as a function of a plant parameter. (Author)