链状非线性振动系统识别的无偏一致算法

An Un-Biased and Consistent Approach of Identification for Chainlike Nonlinear Vibration Systems

本文提出链状非线性振动系统非参数识别的新算法。利用多项式级数逼近一般无记忆非线性恢复力,以回归方法求展开系数,对系统非线性引入的相关残盖,用时间序列分析方法处理,迭代修改初估值,最后得到具有无偏性和一致性的识别结果。

This paper presents a new approach of nonparametric identification for nonlinear dynamic systems. The Chebyshev polynomials series is used to express general memoryless nonlinear restoring forces and the regression technique is utilized to aquire the expansion coefficients. Because system nonlinearity causes corelated residue, the special attention has been put to use the time series analysis method to deal with the residue so can get a modification of pre-estimation by iteration. We also show that the obtained ultimate estimation has un-biased and consistent properties.