基于区间数度量的运动模式建模与控制

Moving pattern measured by interval number for modeling and control

针对一类复杂的生产过程,本文作者在前期研究成果中,提出了基于运动模式的建模方法与控制方法的思想.针对该方法中模式类别变量的度量问题,本文提出采用区间数来度量模式类别变量,进而提出了一种基于区间数度量的运动模式建模与控制方法.首先,采用K均值聚类算法对收集的足够长时间内的工况数据进行聚类,得到C个模式类别,进而构成模式刻度"空间".为了描述模式的运动,本文提出了带输入的区间自回归模型(IARX).在此基础上,采用IARX模型建立模式类别变量的控制模型并给出了相应的控制算法.最后,以烧结生产过程为例验证了本文所提建模与控制方法的有效性.

On the basis of our previous work on moving pattern-based modeling and control method for a class of complex production processes, we propose using the interval number as the measure of the pattern class variable, and develop a new modeling and control method of the moving pattern measured by the interval number. In this approach, using K-means clustering algorithm, the long-time collection of patterns in operating conditions is clustered into C pattern classes to build a scaled space of patterns. To describe the motion of patterns, we introduce the interval autoregressive model with exogenous input （IARX）. With the IARX model, we develop the control model and the control algorithm for the pattern class variable. Practical experimental results are presented for demonstrating the validity and feasibility of the proposed approach of modeling and control.