应用多变量系统辨识建立浮选回路在线模型
来源期刊:中南大学学报(自然科学版)1992年第2期
论文作者:尹蒂 李松仁
文章页码:149 - 155
关键词:系统辨识; 浮选; 模型; 程序包
Key words:system identification; flotation; model; program package
摘 要:浮选回路的在线模型是实现浮选回路计算机高级控制的基础.本文将多变量系统辨识理论和方法应用于浮选回路在线模型的建立,以带外生可观测干扰变量的多变量受控自回归滑动平均模型作为浮选回路在线时间序列模型的基本形式.在IBM-PC/XT微型机上建立了通用的多变量递推辨识程序包MVRIMPB,用以估计模型的阶和参数.最后用仿真方法考查了程序包的实用性.
Abstract: The advanced computer control of flotation circuits is based on the implementation of on-line flotation model. The theory and the method of the multivariable system identification are used to develop on-line model of flotation circuits. The multi-variable, controlled, auto-regressive and moving average model with an extraneous measurable input term is selected as the form of online empirical time-series models of flotation circuits. The computational algorithm of the Recusion Extended Least Squares (RELS) is described in detail. A multi-variable system identification program package, MVRIMPB, developed in IBM-PC/XT is used to estimate all sub-orders and parameters of the model. Finally, the availability and the reliability are illustrated by computer dynamic simulation.