大规模强耦合非线性多变量系统辨识方法研究

系统辨识是研究建立(动态)系统数学模型的理论与方法。目前，非线性系统辨识已成为领域研究热点，但主要集中于单变量系统，即单输入单输出非线性系统。本项目针对化工过程中广泛存在的一类有不可测变量、变量多（维数高）、结构复杂的大规模强耦合非线性多变量系统，将线性系统辨识方法拓展到这类复杂非线性系统，推进复杂非线性多变量系统辨识的研究进程。主要内容包括（1）利用辅助模型技术，研究这类存在不可测变量复杂系统的辨识方法；（2）基于辨识模型分解的递阶辨识原理，系统地研究和提出这类复杂非线性系统的辨识方法，解决大规模非线性系统辨识方法计算量大的问题；（3）应用滤波辨识技术，研究和提出能抑制噪声干扰的复杂非线性多变量系统辨识新方法；（4）研究提出方法的性能，提高非线性系统辨识方法的有效性，以解决一类大规模强耦合非线性工业过程的模型化问题。本项目属于应用基础研究，研究成果在我国化工等流程企业中有重要的应用前景。

System identification is a discipline for establishing mathematical models of (dynamic) systems. At present, nonlinear system identification has become a hot research field, but mainly focuses on single variable systems, namely the single-input single-output nonlinear systems. In chemical process industries, there widely exists a class of nonlinear systems with the characteristics of multivariate, complex structures, strong coupling between variables and existing unknown disturbances and variables. This project is aiming at presenting effective methods and extending linear system identification methods to such complex nonlinear systems, so as to promote the research process of complex nonlinear multivariable system identification. The main work includes (1) study identification methods for complex systems with unmeasurable variables, based on the auxiliary model technology; (2) systematically study and present identification methods for such complex nonlinear systems based on the hierarchical identification principle so as to solve calculation problems for large-scale nonlinear system identification methods; (3) apply the filter identification technology, study and put forward identification methods for such complex nonlinear multivariable system; (4) study performances of the proposed methods so as to improve the effectiveness of nonlinear system identification methods and solve modeling problems for a class of large-scale, strong coupling nonlinear industrial processes. This project belongs to applied basic researches and has very important values in theory and wide application prospects in chemical process industries in our country.