项目名称: 基于时间序列的非线性系统优化迭代学习控制策略和应用边值问题研究
项目编号: No.61263008
项目类型: 地区科学基金项目
立项/批准年度: 2013
项目学科: 自动化技术、计算机技术
项目作者: 郝晓弘
作者单位: 兰州理工大学
项目金额: 43万元
中文摘要: 本课题采用优化理论和时间序列研究建立快速收敛迭代学习控制算法,以便提高学习收敛速度和控制跟踪精度。内容包括:对克隆选择优化迭代学习算法采用实数编码和限制变异扩展半径方法,提高克隆选择算法的搜索性能;研究压缩感知的数据预处理方法,从较少的数据中提取尽量多的信息,减少数据的运算量,提高学习效率;基于L2空间,采用时间序列的理论建立基于时间序列的迭代学习算法,通过构建数据时间序列,构建对应的迭代算法模型;针对被控系统模型不确知、时变、非线性等问题,展开应用边值的研究。同时对构建的新算法进行收敛性和有效性论证,使得新算法在学习效率、收敛速度和跟踪精度上大大提高。通过本课题的研究为优化迭代学习控制的快速算法奠定理论基础,为大时滞、非线性不确定工业过程跟踪问题的控制提供科学手段。
中文关键词: 优化迭代学习控制;随机过程;时间序列;L2空间;边值问题
英文摘要: The subject of the use of optimization theory and time series study the establishment of rapid convergence of iterative learning control algorithm, in order to improve the learning convergence speed and tracking accuracy.The Clone Selection and optimal iterative learning algorithm using real number coding and limit variation propagation radius method, improved clonal selection algorithm to search performance; study of compressed sensing data preprocessing methods, from less data to extract as much information as possible, to reduce the data quantity of operation, improve the efficiency of learning; based on L2 space, using time series theory was established based on the time series of the iterative learning algorithm, by constructing a data time series, constructs the corresponding algorithm model; for the controlled system model is not known with certainty, time-varying, nonlinear problem, launches the application boundary value research. At the same time to the construction of the new algorithm convergence and validity of proof, show that the new algorithm in the convergence speed and learning efficiency, tracking precision. The research will lay the theoretical foundationfor the iterative learning control algorithm , and provide a scientific basis for large time delay, uncertain nonlinear industrial process c
英文关键词: Optimal iterative learning control;Stochastic process;Time series;L2 space;Boundary value problem