摘要
研究AR和Prony法的线性时变振动系统的模态参数识别。提出可以处理非平稳信号的时变Prony法。采用递推最小二乘算法估计模型参数,并讨论基函数的选择,其中以时间多项式基、离散长球面基以及离散余弦基下所识别的效果较好。为验证方法的有效性,进行了数字仿真和附加质量随时间连续变化的悬臂梁的模态分析试验,试验结果和理论结果比较,表明了提出的线性时变系统的模态参数识别方法有效、可行。
In this paper, modal parameter identification of a time-variant linear vibration system is studied using autoregressive modeling and Prony analysis. Based on the classical Prony analysis, an improved method to process non-stationary signals is presented. The recursive least squares method is adopted to estimate model parameters. Choice of the base functions is discussed. It shows that as the base functions, time polynomial functions, discrete long spherical functions and discrete cosine functions are more suitable than the others for modal parameter identification. To validate the proposed method, numerical simulation and a vibration mode test of a cantilever beam are completed, where the mass varies continuously with time. Comparison between identification results and theoretical results indicates that the proposed method is effective and feasible to identify modal parameters of time-variant linear system.
出处
《噪声与振动控制》
CSCD
北大核心
2009年第1期5-8,20,共5页
Noise and Vibration Control
基金
国家自然科学基金项目(50678173)
上海铁路局资助项目(2006076)
关键词
振动与波
模态参数识别
时变Prony法
基函数
线性时变系统
vibration and wave
modal parameter identification, time varying Prony method, base function, time-variant linear system
