一种时变参数模型法在连续变速颤振试验信号处理中的应用研究

A New Method of Time-Varying Parameter Modelling Suitable for Data Processing of FTPVS (Flutter Test with Progression Variable Speed)

针对连续变速颤振试验信号的非平稳特点及在线分析需求,提出了一种基于基函数的时变参数模型分析方法。该方法采用递推最小二乘算法对数据建立时变的参数模型,并将时变参数用基函数的加权和来表示,从而将1个线性非平稳问题转化为线性时不变问题。数值仿真及低速风洞气弹模型的试验结果证明了文中方法的计算精度和速度可以满足连续变速颤振试验的要求。

Purpose. At the beginning of the full paper, we discuss in detail the requirements of real-time data processing of FTPVS, which is short for what we call ＂flutter test with progression variable speed＂. Then we utilize the ideas of time-varying parameter modelling to develop a new method that can meet the requirement of FTPVS on processing precision. In the full paper, we explain in detail our new methods here we give only a briefing. A time-varying parameter model based on basis functions is constructed and reeursive least squares （RLS） algorithm is adopted to estimate the parameters. We represent time-varying parameters with summations of weighted basis functions, thus the linear nonstationary problem is transformed into a linear time-invariant ones RLS algorithm is able to lessen the amount of computation and storage. Finally, we give an application example involving aero-elastic tests in a low-speed wind tunnel. In this example, we show how to process the data obtained from a certain test condition with an initial wind velocity of 7.1 m/s; as this speed increased gradually, we found experimentally the critical speed to be 22 m/s. And the predicted speed by using our data processing method is 22.6 m/s. This shows that our method is able to meet the requirement of FTPVS on high data processing precision.