具有随机噪声污染的连续过程辨识的一种频域方法  

A Frequency-domain Approach to the Identification of Continuous Processes Corrupted with Random Noisies

作　　者：陶永林[1] 倪祖彭[1] 章文斌[1] 

机构地区：[1]上海第二工业大学电子电气工程系

出　　处：《上海第二工业大学学报》1990年第1期57-62,共6页Journal of Shanghai Polytechnic University

摘　　要：已知过程输入、输出的自谱密度以及它们之间的互谱密度在若干个频率上的数据,可以辨识过程模型和噪声模型的阶次、参数(包括过程的滞后时间).同时,这里证明了:当辨识涉及的频率数目∠≥2max(n,m)时,辨识问题有唯一解,其中n和m分别为过程模型噪声模型的阶次.If the data of the spectral densities of both the process (under consideration) input and output, and the input-output cross spectral density at certain frequency points are given, it is possible to identify the orders and the parameters (including the dead time of the process) of both the process model and the noise model.Meanwhile it is proved that if the number of frequency points involved in the identification, 1≥2max(n, m), then a unique solution to the problem of the identification is found, when n and m represent the orders of the models for both the process and the random moise respectively.

关 键 词：连续过程辨识 随机噪声 频域方法 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]