基于非平稳时间序列的连续变速颤振试验与在线分析方法研究

项目名称： 基于非平稳时间序列的连续变速颤振试验与在线分析方法研究

项目编号： No.11302175

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 数理科学和化学

项目作者： 郑华

作者单位： 西北工业大学

项目金额： 23万元

中文摘要： 连续变速是近年来提出的一种全新颤振试验理念，具有更加贴近于飞行器的实际工作状态且极大幅度提高试验效率等优势，但同时这类颤振试验的风险也随之成倍增加。为了确保试验安全，急需一套可靠的面向该类试验的在线分析理论。然而这类试验的观测均为非平稳信号，相应的数学理论尚不健全，传统的颤振信号处理方法一般均在某些假设下进行近似处理，从而直接影响到了分析结论的质量，因此迫切需要发展并解决与之相适应的信号在线监测和分析技术。 本课题首先通过对非平稳信号的定性研究，给出以非平稳度为核心的定量分析方法；经过对现有时变参数建模方法的系统研究，探讨基于状态空间模型的粒子滤波算法的优势，并从数学上进行算法结构优化，以显著提高其实时性；依据连续变速颤振试验机理和观测信号特征，研究递推时频谱的构造以及颤振特征量提取方法。借助数值仿真和气弹模型设计及其风洞试验，对所研究的理论方法和实现技术的性能进行全面验证和评估。

中文关键词： 时变参数建模；非平稳度；粒子滤波；连续变速颤振试验；颤振边界预测

英文摘要： Flutter test with progression variable speed is a new technique proposed in recent years and extensive attended at home and abroad, which can be much closer to the real flight and significantly improve the test efficiency, but also along with multiple risks. Additionally, due to the nonstationarity of data from such test and the short of corresponding mathematical theorie, the traditional analysis conclusions are directly affected by approximate solutions under certain assumptions, and few flutter signal processing methods could apply. Consequently, there is a critical need to research and develop the new techniques to monitor and analyze signal from the flutter test with progression variable speed on-line. Based on qualitative studies of non-stationary signals, the degree of non-stationary is defined firstly. Through the research of the existing time-varying parameter modeling methods, the advantages of particle filter based on state space model is explored, and the structure of particle filter is optimized mathematically to improve its on-line performance. On the basis the mechanism and signal characteristics of flutter test with progression variable speed, methods of recursive time-frequency analysis implementation and the flutter characteristic quantity extraction are researched in depth. With the numerical

英文关键词： time-varying parameter modeling；non-stationary degree；particle filter；Flutter test with progression variable speed；Flutter Boundary Prediction