系统辨识算法复杂性、收敛性、计算效率研究

项目名称： 系统辨识算法复杂性、收敛性、计算效率研究

项目编号： No.61472195

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 自动化技术、计算机技术

项目作者： 丁锋

作者单位： 青岛科技大学

项目金额： 80万元

中文摘要： 在许多工程系统中，经常会遇到大型计算问题、优化问题，使得求解问题算法的复杂性、计算量，计算精度等成为突出问题。本项目研究大规模非线性系统、多变量系统辨识算法的复杂性、收敛性、计算效率等重要科学问题。主要研究内容为（1）利用信息滤波技术和多新息辨识理论，研究能提高辨识精度的大规模非线性系统辨识理论与方法及其复杂性；（2）利用递阶辨识原理，结合优化技术，根据系统的输入输出数据，研究和提出维数高、变量数目多的多变量系统递阶辨识方法，减小算法的计算量和复杂性；（3）利用鞅收敛理论，研究和建立非线性系统和多变量系统辨识方法的收敛理论；（4）采用并行计算方法与递阶计算方法，分析非线性系统和多变量系统辨识算法的计算量，探讨辨识算法的在线实现方式，提高计算效率，以解决一类大规模非线性系统和多变量系统辨识算法计算效率和模型化问题。本项目属于应用基础研究，研究成果在我国炼油、化工等流程企业中有广泛的应用前景。

中文关键词： 算法设计；算法分析；计算复杂性；并行计算

英文摘要： In many engineering systems, one often encounters large-scale computational problems and optimization problems, so that the complexity, computation and calculation accuracy of algorithms for solving problems become a prominent issue. The project studies some important scientific issues about the complexity, convergence, computational efficiency of identification algorithms for large-scale nonlinear systems and multivariable systems. The main contents include (1) to use the information filtering technology and the multi-innovation identification theory, study the identification methods and their complexity for large-scale nonlinear systems, which can improve the identification accuracy; (2) to use the hierarchical identification principle and the optimization techniques, based on the input-output data of systems, study and propose the hierarchical identification methods for multivariable systems with high dimensionalities and more variables so as to reduce computational complexity; (3) to use the martingale convergence theory, study and establish the convergence theory of nonlinear system and multivariable system identification methods; (4) to use the parallel computing and the hierarchical computation, analyze the amount of work involved in nonlinear system and multivariable system identification algorithms, explore the implementation of the online identification algorithms, improve the computational efficiency so as to solve the modeling problems of a class of large-scale nonlinear systems and multivariable systems. This project belongs to the basic theory and applications, the achieved results have important applications in China's oil refining and chemical industry.

英文关键词： Algorithm design;Algorithm analysis;Computional complexity;Parallel computation