关键词: 工作模态参数识别, 线性慢时变结构, 滑动窗, 等变自适应源分离, 变步长

Abstract: To identify the transient and time-varying modal parameters of Slow Linear Time-Varying (SLTV) structures with weakly damped online only from non-stationary vibration response signals,a Moving Window Variable step-size Equivariant Adaptive Separation via Independence (MWVEASI) based operational modal identification method was proposed.Through the fixed length window technique in MWVEASI,non-stationary vibration response signals could approximately be seen as stationary random time series in each window and SLTV structures could approximately be seen as multiple linear time-invariant structures.In each window,the variable step-size EASI algorithm was used to estimate the modal shapes and natural frequencies as the transient modal parameters at the middle moment.The whole time period time-varying modal parameters of the SLTV structure were obtained through moving the window.The identification results on simulation data of the three-degree-freedom spring oscillator structure with slow time-varying mass showed that two different MWVEASI methods could well identify the modal shapes and natural frequencies of SLTV structures.Compared with moving window fixed-step EASI and variable-step EASI,MWVEASI had better tracking time-varying characteristics,higher recognition accuracy and faster convergence.

Key words: operational modal parameter identification, slow linear time-varying structure, moving window, equivariant adaptive separation via independence, variable-step