项目名称: 模型参数辨识中的试验优化设计问题研究
项目编号: No.61203116
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 自动化学科
项目作者: 刘雨
作者单位: 哈尔滨工业大学
项目金额: 22万元
中文摘要: 最优试验设计在科学实验中的数学建模、控制工程中的系统辨识、工业生产过程中的质量控制等方面,都具有非常广阔的应用前景。但是,传统的设计方法仅适用于回归模型和线性系统模型,而对于航空、航天等领域中不断出现的非线性动态系统辨识试验设计问题,则缺少必要的理论支撑和有效的解决办法。申请人在国内首次开展了相关的理论和试验研究工作,率先将全局最优化方法应用于非线性动态系统参数辨识的最优试验设计。研究内容包括:(1)针对非线性动态系统参数估计中的最优输入信号设计问题,给出相应的试验计划随机化设计方法,并研究非随机化方法的逼近问题;(2)将此类试验设计问题转化为有约束最优化问题,在此基础上,研究随机和确定性的全局最优化方法,关键是给出一类有约束非凸函数极小化问题的全局最优化条件;(3)以惯导平台系统自标定试验中旋转轨迹设计问题为应用背景,通过仿真和试验研究来检验所提出设计方法的工程实用性。
中文关键词: 试验设计;全局优化;非线性动态系统;参数估计;惯导系统自标定试验
英文摘要: Optimal experiment design has broad application prospects in many aspects such as mathematical modeling in scientific experiments, system identification for control engineering and quality control of industrial production. But traditional design methods can only apply to regression model and linear system model, while for the emerging experiment design problems of nonlinear dynamic system identification in the field of aerospace, necessary theoretical support and effective approaches are very absent. The applicant carries out relevant theoretical and experimental research for the first time in China and takes the lead in the application of global optimization method to optimal experiment design for parameter identification of nonlinear dynamic systems. The research contents include:(1)for the optimal input design problems in parameter estimation of nonlinear dynamic systems, the corresponding randomized design method of test plan is presented and the approximation problem of nonrandomized method is studied as well;(2)on the basis of transforming the experiment design problem into a constrained optimization problem, the stochastic and deterministic global optimization methods are studied with more emphasis on the global optimality conditions for a kind of constrained nonconvex minimization problem;(3)with the des
英文关键词: experimental design;global optimization;nonlinear dynamic system;parameter estimation;self-calibration test of INS