Structural Damage Identification Based On Federal Extended Kalman Filter

The damage identification of bridge structure is crucial in bridge health monitoring system. Currently, the structural damage identification method based on the time-domain vibration signal is developing fastly. The damage identification method based on the time-domain method contains extended Kalman filter and least square method etc. The extended Kalman filtering method(EKF) is a kind of real-time recursive algorithm to identify the physical parameters of the nonlinear system structure. The algorithm can accurately identify the location and degree of the variation of structure parameters. However, the traditional extended Kalman filtering method has two limitations:(1) the state vector in EKF contains structural physical parameters. The relatively large dimensions of state vector will result in a low computational efficiency for a complex structure;(2) EKF belongs to a kind of centralized filter structure with a bad fault tolerance. If the fault exists in a vibration signal, the result of the structural damage identification will be affected significantly.The federal extended Kalman filtering method(FEKF) is selected to erase the limitations of EKF. FEKF has a flexible structure with a small amount of calculation and a good fault tolerance. It can accurately identify the fault signal to avoid a bad result in structural damage identification. The state vector can be built by modal coordinates. The order modal coordinates are intercepted to reduce the dimension of the state vector. Thus, the stability of damage identification results will be improved. The following is the main research contents.(1) The extended Kalman filtering method(FEKF) that is widely used in the field of navigation is s to identify the damage of bridge structures. The damage identification recursive formula is derived based on the free vibration response of the structure. A numerical analysis of free vibration of a simply supported beam is designed for verifying the valid of FEKF. The results show a high precision and stability.(2) Currently, few scholars apply the kind of recursive method, such as FEKF, into the damage identification of the bridge structure under a moving load. However, in practical engineering, it is difficult to measure the free vibration signal because bridge structures are in service stage. FEKF can be used to identify the damage of the bridge structure under a moving load and which will be satisfied to the requirements of online damage identification of bridges in service stage. The damage identification recursive formula is derived based on the forced vibration response of the bridge structure under a moving load. The results of numerical analysis show that FEKF can accurately identify the location and degree of damage.(3) Currently, there is seldom consideration of existing the fault in vibration signals of sensors to the work of the damage identification of structures. The vibration signal from fault sensors will lead to the erroneous results of damage identification of structures. The residual chi-square test is proposed for the fault detection of the vibration signals in this dissertation. A numerical analysis with different vibration signal fault types is executed. The results show that the residual chi-square test can accurately identify the fault signals and distinguish the type of fault. Then the fault signal can be rejected from all the signals. Thus the results of damage identification will be accuracy and stability. Through the application of the fault detection function, FEKF can still guarantee the good accuracy of the identification results even if a fault signal exists.