مطالعه شبیه سازی بر روی محلی سازی آسیب پرتو با استفاده از تئوری شواهد

Structural damage localization plays an important role in structural damage identification. An cumulative structural damage localization method based on evidence theory is studied in this paper. An algorithm for cumulative fusion of primary localization information is presented. In the numerical study, a fixed-end beam is inspected. Two damage cases are studied, one with single damage and the other with multiple damages. White noises are added to the beam mode shapes of the first five orders to simulate the measurement noises. First, Modal Strain Energy Change Ratio (MSECR) method is used to localize the damage elements using the noise-contaminated mode shapes. Then, the locations determined by MSECR are taken as primary information and fused cumulatively. In both the single-damage-case and the multiple-damage-case, the fused results show that damaged elements are accurately identified. Furthermore, the damage elements are distinguished much more clearly than those identified from MSECR.

With the development of structural engineering, modern structures are becoming more and more complex, e.g., aircrafts, long span bridges and off-shore platforms. Since the working environments of these structures are usually critical, structural damages will inevitably occur during the service life. The damages will lead to terrible consequences if they are not identified in time. Therefore, structural damage identification becomes necessary for these structures. In the past decades, varieties of damage identification methods [1] have been proposed to identify the occurrence, the location and the qualification of damage. Also, many damage localization indicators have been suggested. Based on the eigen-equation, Ojalvo et al. [2] proposed to locate the model error using modal force residue. The damaged elements are identified by their relevant degrees of freedom. Messina et al. [3] proposed to use a multiple damage location assurance criterion to locate the damaged elements. Their method is formulated on the same basis as the modal assurance criterion. Shi et al. [4, 5] proposed to locate the damaged elements by comparing the modal strain energy of each element before and after the damage. The modal strain energy change ratio (MSECR) of each element is defined to be a damage indicator. In their study, the indicator is proved to be effective in locating the structural damage.Though there are several available damage localization methods and indicators, engineers find it still difficult to determine the damage location in practice. They do not know which method to trust since the results from different methods may not be in good agreement, and are sometime discordant. Hence, it becomes necessary to research on methods to fuse structural damage identification results from multiple sources. Guo [6] proposed a method using information fusion technique. The method is able to fuse local decisions from multiple damage location assurance criterion (MDLAC) method and the frequency change damage diction method (FCDDM), where frequency data and mode shape data are used as information sources respectively. This paper studies structural damage localization using data fusion. Different from the method of Guo, which use frequency data and mode shape data as two information sources, this study takes mode shapes of different orders as information sources. In section two, the algorithm for cumulative fusion is studied. In section three, a fixed-end beam is studied as a numerical example. Two damage cases are studied: one is single damage-case and the other is multiple-damage-case. In both cases, MSECR is used to locate the damage element using mode shapes of different orders.

cumulative structural damage localization method based on evidence theory is studied in this paper. Different from the method which uses frequency data and mode shape data as information sources, this study uses mode shapes of different orders as information sources. Simulation cases constructed by using fixed-end beam with single damage and multiple damages are studied. Results show that the proposed method is efficient under the noisecontamination condition. This provides a new technique to adopt information fusion in structural damage identification.