基于子空间方法的精馏塔系统辨识建模研究

System Identification Modeling of Distillation Column Based on Subspace Method

子空间辨识方法作为一种有效的针对多输入-多输出系统(MIMO)的辨识建模方法近年来受到广泛的重视。目前主要采用的子空间辨识算法只能适用于白噪声环境,而实际的工业现场数据很多是受到较大有色噪声干扰的。针对问题采用了一种新的子空间辨识算法,利用马尔可夫参数用于处理随机性部分,同时引入辅助变量用以去除噪声的干扰,能够适用于存在较大有色噪声干扰情况下的辨识建模,并可得到对象的无偏模型,建模的精度优于通常所采用的子空间辨识算法。通过对精馏塔仿真模型的辨识结果证明了该方法的可行性和有效性,以及在实际工业过程对象建模中良好的应用前景。

Subspace identification method as an effective identification modeling way for multi input and output system has drawn much attention recently. The mainly used subspace identification algorithms are effective only in white noise,while the most industrial field data are impacted by colorednoise. To solve this problem, a new subspace identification algorithm is used in this paper, which uses Markov Parameters for stochastic part and adds instrumental variables to remove the effects of the unmeasured noise sources. This subspace identification algorithm is suited for the situation disturbed by colored - noise, and could obtain an unbiased model. The accuracy of model is better than that of the normally used subspace identification algorithm. The simulation results demonstrate the effectiveness and feasibility of the method.