Research On Vibration-based Structural Damage Identification Method

Large-scale engineering structures occupy a pivotal position in the national economy, and their health status is closely related to people's life and property. Research on the structural damage identification methods, which can help avoid disaster and reduce maintenance costs, has very important theoretical and practical significance. Vibration-based structural damage identification method has been a research hotspot in recent years. In this paper, a series of studies are carried out based on the previous achievements and the development direction of this field. The main contents are as follows:(1) A comprehensive overview of the research on damage identification is carried out, in which the vibration-based structural damage identification method (dynamic method) is emphatically discussed. A clear and reasonable classification for dynamic method is conducted, and the research status of all kinds of methods is introduced. The current issues and future development direction are analyzed and summarized, which provide guidance for this research.(2) The damage identification method based on curvature mode is systematically studied. The curvature calculation formula for two special cases, one is uneven measuring point and the other is the end point, is proposed. The deficiencies of commonly used normalization method for curvature mode are analyzed, and the optimal normalization technique is proposed, which can make the injury peak much more significant. Coordinate transformation technology is used to solve the calculation difficulty of curvature mode in complex structure so as to lay a good foundation for the popularization and application of the curvature mode method. The interference peak problem of the curvature mode in continuous beam structure is discovered, and then the wavelet transform technology is used to eliminate the interference peak, which extends the application scope of the curvature mode index. The quantitative damage identification theory of curvature mode is proposed, which proves the deficiencies of the curvature mode difference index. Then a new index called the superimposed curvature mode change rate is developed, together with the corresponding damage factor estimation methods. The new index can not only locate the damage but also reflect damage degree, which means a significant advantage.(3) The damage identification method based on the modal flexibility matrix is studied. The analysis points out that the traditional modal flexibility matrix will lost its effects when the mode shape is not mass normalized. And the proportional flexibility matrix, which is difficult to calculate, is suitable for this condition. The improved calculation method and formula for proportional flexibility matrix are proposed, which can greatly reduce the computational difficulty and achieve satisfying damage identification results.(4) The damage identification method based on modal strain energy is studied. The zero point problem of modal strain energy in continuous beam is discovered, which can cause the failure of the commonly used modal strain energy change rate index. Two improved indexes called superimposed modal strain energy difference and superimposed modal strain energy change rate are proposed, which can effectively overcome the above problems.(5) The noise problem with damage identification process is studied. Firstly the disadvantages of current research are analyzed and pointed out. Then a unified, coordinated, intuitive noise simulation and measurement method as well as two noise immunity evaluation methods for damage identification index, called the recognition success rate evaluation method and the noise amplification factor evaluation method, are proposed. The effectiveness of the two evaluation method is proved by numerical examples. The noise immunity of curvature mode, modal flexibility matrix and the modal strain energy index is studied, and the results show that the curvature mode index has the best noise immunity, but there is still a large room for improvement. This conclusion is validated using the test data of1-40bridge provided by Los Alamos National Laboratory. The noise amplification theory of curvature calculation is proposed, thanks to this the one-sided point of view which says "more measuring point, more accuracy in curvature", is corrected. Two difference calculation methods with the noise suppression effect are proposed, which are called as the interval differential and smooth differential technique. These techniques improve the noise immunity of the of curvature mode index.(6) The damage identification method based on the autocorrelation function square root is studied. The effects of three autocorrelation function damage identification index are analyzed, and results show that their effect is not satisfactory. The concept called autocorrelation function square root is proposed, together with the corresponding damage identification theory. And then two damage identification indexes called autocorrelation function square root difference and autocorrelation function square root change rate are put forward. Numerical example proves that both of the new indexes have excellent damage localization ability and anti-noise performance, and the latter can even reflect the damage degree, which has a significant advantage over other indexes..(7) The new idea of damage identification based on movable sensors is proposed. A damage identification technology based on movable sensors is developed by the help of autocorrelation function square root index. Numerical example shows that only a few movable sensors are required for this technology, but can achieve the same effect of a large number of fixed sensors. So this technology can significantly reduce the dependence of the sensor's quantity and has a broad application prospect.(8) The damage identification method based on Unscented Kalman filter (UKF) is studied. Firstly the UKF method in case of known excitation is proposed. Then the case of unknown excitation is researched. An extrapolation UKF algorithm is put forward. Combining with the least squares technique, a new UKF method which can predict excitation, is developed. The effectiveness of UKF method is studied using the experimental data of a three-story architectural model provided by Los Alamos National Laboratory. The results show that UKF method has very excellent performance in damage localization, quantification and noise immunity.