基于联邦扩展卡尔曼滤波的结构损伤识别方法

Structural damage identification based on the federal extended kalman filter

当结构损伤与传感器故障同时存在时,两者的相互影响会导致损伤识别结果的劣化;为此,提出了一种基于联邦扩展卡尔曼滤波的结构损伤识别算法。利用分散化滤波计算量小、滤波精度高的优点,联邦扩展卡尔曼滤波方法能根据正常的传感器信号准确识别结构的损伤位置与程度,具有良好的鲁棒性;同时利用联邦滤波容错性好的特点,能实现对故障信号的自动检测和剔除,并将剩余的正常子系统进行组合,以继续提供准确的损伤识别结果。梁式结构的数值算例及实验分析验证了该算法的有效性及对故障传感器信号的检测隔离能力。

The coexistence of structure damages and sensor faults will deteriorate identified results evidently,so an algorithm for the identification of structure damages based on the Federated Extended Kalman Filter method（ FEKF） was proposed by using free vibration signals. The presented method can identify the location and extent of damages accurately,and shows good robustness when the sensors work normally. Combined with the residual chi-square test,the FEKF also can eliminate the effects of fault sensors by the automatic detection and removal of the fault sensor signals. Numerical simulations and experiments show that the FEKF can ensure the accuracy and stability of the damage identification results and detect fault signals effectively.