改进的NExT-ERA时域模态识别法的误差分析

Error analysis of time-domain modal identification methods of improved NExT-ERA

为进行结构损伤识别,用时域模态识别的NExT-ERA法,从脉动激励和测量噪声下有限时长和时间间隔的结构反应中提取高精度模态参数。改进的相关函数估计公式可消除传统相关估计引起的模态参数误差;利用稳定图对不同系统的阶进行识别,以减小系统定阶误差并选择高精度模态;根据Hankel阵奇异向量、奇值和模态参数的对应关系,分析时域模态识别误差的规律和影响因素。用某三自由度框架和主跨1 650 m的西堠门悬索桥模型验证有效。

Accurate modal parameters are needed for effective structural damage detection. The NExT-ERA for modal identification in time domain was used to identify high-accuracy modes from noisy structural ambient dynamic responses of limited recording length and time intervals. The correlation estimation in the NExT-ERA was improved to eliminate errors due to traditional correlation estimaties of short-recording-length structural responses The mean values and standard deviations of the modal parameters identified for different system orders in the stabilization diagram were used to reduce the order-determining errors and to select the modes. The modal parameter errors were analyzed for rules and influencing factors based on the relations between the singular vectors and values of the Hankel matrix and the modal parameters. A model of a 3 degree-of-freedom frame and the Xihoumen suspension bridge of 1 650 m main span was used as an example.