IDENTIFICATION OF STRUCTURAL MULTIPLE DAMAGED LOCATIONS BASED ON GENETIC ALGORITHMS AND DEMPSTER-SHAFER FUSION THEORY

Guo Huiyong  Zhang Ling

(School of Architecture Engineering and Mechanics, Xi’an Jiaotong University, Xi’an 710049)

Abstract: Information fusion technology is applied to identify structural multiple damaged locations. It is considered that the multiple evidence from different information sources of different importance or reliability are not equally important when they are combined according to Dempster-Shafer theory, which is seldom considered in the existent combination methods. A new method, i.e. weighted evidence balance method, is presented to solve this problem. The method first searches for the optimal weighting coefficients of different evidences using genetic algorithms, then balances the considered evidences according to the weighted average of all and the preferred evidence, and finally combines them. Thus, It is guarantied that the balanced evidences won’t change the weighted average of all and the preferred evidence. The simulation results demonstrate the excellent performance of the weighted evidence balance method to identify multiple damage locations as compared with other methods, such as those methods based on mode shape change or frequency change, basic evidence theory and other weighted evidence combination methods.

Key words: Damage identification  Genetic algorithms  Information fusion  Evidence theory

CLC No: TB123

陕西省自然科学研究资助项目(2002E206). Received 20031003, received in revised form 20040410