The Structure Damage Identification Research Based On The Structure Modal Parameters And SVM

House or bridge structure in service period has the damage more or less, but the accumulation of damages has possibility to cause the disastrous accident and brings the serious loss. Therefore, to ensure the safety of life and property, how to establish the damage identification methods which can accurately position location and extent is the focus of academic and engineering research.After described the research status and significant of structural damage identification, this paper did a systematic exposition about the damage identification based on structural modal parameters and the theory of Support Vector Machine(SVM),then combined the method based on modal parameters with the way of SVM and, applied it to the damage identification of truss structures. The results show that this method has achieved good results in identification of the damage location and the extent of damage.(1) Summarize the background and research significance of structural damage identification and do a systematic exposition about the support vector machine, neural networks, genetic algorithm and wavelet analysis method.(2) Discuss damage identification methods based on modal parameters, during this process, analyze the elementary theory of many damage identification methods based on the frequency shift and based on the mode of vibration change.(3) Describe the theory and algorithm of SVM in detail.(4) Introduce the step to establish support vector machines algorithm in the structure damage identification.(5) Take a cantilever beam and a truss structure as research objects, to do the numerical experimentation simulation, and compile the APDL (the re-development language of ANSYS) procedure, to achieve structure mode automatically solved and the result standardization output under those different damage conditions. Then according to the needs of damage identification to construct sample collection composed of kinds of eigenvectors. Take the cantilever beam as an example, it verified that damage identification techniques based on natural frequencies and support vector machine can locate the simple structure's damage position exactly and, the use of support vector classification machine and regression machine to identify respectively the damage position and the extent of damage of truss also obtained the desired results.Finally, summary the prime task and the research results comprehensively, and point out further deep research topic.