过去的30年来，自动化界对过程监测与故障诊断进行了大量研究，产生了很多理论和方法。但是，该领域整体上没有形成实用的技术，工业应用很少；特别是与模型预测控制技术的广泛应用相比。导致这一困境的最大问题是动态建模成本太高。为了解决动态建模问题，推动过程监测与故障诊断理论和方法在大型工业过程的应用，我们提出“面向过程监测与故障诊断的系统辨识方法研究” 作为研究课题。因为在过程监测与故障诊断建模中，过程的输入大都含有测量噪声，该项目的研究重点是输入含测量噪声的errors-in-variables (EIV) 系统辨识。本项目将研究多变量闭环EIV系统的辨识，包含离散时间系统和连续时间系统。在研究中，将全面解决系统辨识中的四个基本问题，即实验设计、参数估计、阶次确定和模型验证，产生实用的EIV系统辨识理论和方法。

In the last three decades, process monitoring and fault diagnosis have been studied intensively in the control community and many theories and methods developed. However, this research field has not created applicable technologies and there are very few industrial applications. In the same 30 year period, model predictive control (MPC) technology has been widely applied in process industries and has generated enormous economical and social benefits. The major obstacle that prevented the industrial applications of process monitoring and fault diagnosis is the high cost of modeling. In this project, in order to solve the modeling issue in process monitoring and fault diagnosis, we propose the research project of “system identification for process monitoring and fault diagnosis” that aims to promote industrial applications of process monitoring and fault diagnosis. In process monitoring and fault diagnosis modeling, process inputs are corrupted by measurement noises. Therefore, this project will focus on errors-in-variables (EIV) system identification. We will study multivariable closed-loop EIV system identification in both discrete-time and continuous-time. We will provide systematic solutions to the four basic identification problems: test design, parameter estimation, order selection and model validation. The work will result in applicable EIV system identification methods.