Damage Detection Based On Dynamic Characteristics Of Structures Under Control And Information Fusion Techniques

Methods of structural damage detection by using the data of dynamic characteristics of structures under control, information fusion techniques and two-step damage diagnosis are proposed in the thesis. Multiple damage location assurance criterion (MDLAC) is described and frequency change square ratio correlation (FCSRC) is proposed in the study. To obtain more accurate identification results, information fusion techniques are presented and the use of Dempster-shafer theory applied to fusion of the diagnosis results obtained from different damage detection methods is described in the thesis. Damage detection methods based on dynamic characteristics of structures under control are discussed. How to design structures under control is also discussed in detail. MDLAC and FCSRC are used to detect the damage of structures under control and structures without control separately. Numerical results indicate that the methods using dynamic characteristics of structures under control is better. The effects of the number and location of control forces on the damage detection are discussed. The two-step damage detection strategy is suitable for complex structures. In the thesis, a two-step damage detection method with the data of dynamic characteristics of structures under control is proposed. The structure under detection is divided into several sub-domains and the probabilistic neural network is employed to identify the damaged sub-domain. Then MDLAC and FCSRC are used as the damage indicators to locate the specific damaged element and the information fusion technique is further employed to fuse the diagnosis results from MDLAC and FCSRC. Numerical results show that the damaged sub-domain can be well identified using the probabilistic neural networks and the specific damaged element can be identified more accurately using the data of dynamic characteristics of structures under control. The use of information fusion techniques can further improve the correction of damage detection.