摘要随着控制对象复杂性的提高，控制理论及其应用也日益广泛，但其实际应用不能脱离被控对象的数学模型。过程模型建立主要有两种类型的方法，一类是基于机理模型的方法，一类是基于数据驱动的方法。系统辨识方法属于数据驱动建立模型方法类，对于一些难以建立机理模型的复杂系统，系统辨识法充分利用系统的输入、输出数据，基于系统的数据，按照一个准则在一组模型类中选择一个与数据拟合得最好的模型，其中输入输出数据是辨识的基础，准则是辨识的优化目标，模型类是选择模型的范围。33516
本文主要以MATLAB为编程仿真工具，对系统辨识的基本原理及系统辨识的相关算法进行了大量仿真实验，包括一般最小二乘法、递推最小二乘法、广义最小二乘法、增广最小二乘法、辅助变量法等。在此基础上，利用系统辨识相关算法辨识实际系统模型，利用编程语言实现系统辨识的各种方法。并对数据处理和建模准确度问题进行了深入讨论。
关键词：系统辨识，数学模型，最小二乘法，数据处理
毕业论文设计说明书外文摘要
Title Research on system identification method under deficient data situation
Abstract
With the improving of the complexity of the object control, control theory is applied more and more widely, but its actual application can not be porced from mathematical model of controlled device. Process modeling have two mainly types of methods, one is a method based on the mechanism of the model, another is based on data driven method.The system identification method is data driven model is established, for some is difficult to establish the mechanism model of complex system, input and output data of the system identification method to make full use of the system, based on the data of the system, according to a criterion in a set of model class selection a number according to fit the best model, the input and output data is identified, the criterion is the optimization target identification, model is model selection.
This paper mainly using MATLAB simulation tools for programming, of the identification system and the basic principle of system identification algorithms were a large number of simulation experiments, including ordinary least square method, recursive least square method, generalized least squares, augmented least square method, instrumental variable method. On this basis, the system identification using correlation algorithm to identify the actual system model, the use of programming language, the method of system identification. And the accuracy of data processing and modeling are discussed.
Keywords:System identification, mathematical model, least squares method, data processing.
目 次
1 绪论 1
1.1 本课题的研究背景 1
1.2 系统辨识的研究状况 1
1.2.1 系统辨识的基本概念及发展 1
1.2.2 系统辨识的主要步骤 2
1.2.3 系统辨识的主要方法 4
1.3 本论文的工作 5
2 常用辨识方法的仿真 6
2.1 建立数学模型的基本方法 6
2.2 几种常见的数学模型的数学表示 6
2.2.1 脉冲响应函数 6
2.2.2 线性差分方程 7
2.2.3 状态空间模型 7
2.3 基于最小二乘法的辨识方法 7
2.3.1 最小二乘法 7
2.3.2 递推最小二乘法 9 MATLAB不完整数据下的系统辨识方法研究:http://www.751com.cn/zidonghua/lunwen_30665.html