| High-rise tower structures such as television tower, transmit electricity tower etc are a kind of configuration shape in wide use. In this kind of configuration, joints are couple up by bolt or welding line. It's necessary for us to diagnose joint damage in structures in time because bolt and welding line will become flexible or appear crack caused by construction vibration suffering from perennial wind load. If don't find in time, the joint damage will expand and then the structure will collapse acted by violent wind load. For simple structure, the damage detection approach has grown up, but for large and complex structures, one-step detection approaches are difficult to use in practice project. So it is essential for us to research two-step detection approaches for large and complex structures. Therefore, the main content of the paper is as follows:(1) The structural damage identification methods are classified. Mostly as direct model diagnoses methods and indirect model diagnoses methods. Subsequently, the two-step detection approaches that had been researched are reviewed respectively.(2) For tower structures, bolt' looseness or welding line' desquamation can be described by concretion coefficient that is osculating to the joint status of element-end. The element stiffness matrix of plane beam element about joint semi-rigid connection has been presented. In this paper, an additional kind of modeling method: three-element combination modeling method is put forward direct according to the defining of joint damage. Then the element stiffness matrix is reckoned. It is convenient to set up the joint-damage model of dimensional structure on this method.(3) The residual force theory and strain mode theory are analyzed in this paper. The calculation analytical method and experiment mode analytical method of mode especially strain mode are educed and summarized. It establishes the academic basic for acquisition of academic data and practical measure data.(4) The theory foundation and characteristic of NN are analyzed in this paper. Aiming at damage diagnoses problem, the selection of NN structure, selection of training function and selection of training sample are analyzed andsummarized. The basal theory and characteristic of genetic algorithm are analyzed in this paper. Aiming at the simple genetic algorithm's bad astringency, a kind of hybrid genetic algorithm, which colligates the excellence of ameliorative genetic algorithm and bfgs is put forwarded. The astringency of the hybrid genetic algorithm is improved largely. At last, the basal principle and characteristic of fuzzy mode identification are summarized.(5) Two lands of two-step detection approaches are presented: two-step detection approach of delaminating neural network and two-step detection approach of index analysis and reverse analysis. The first layer NN of two-step detection approach of delaminating neural network is trained by using the area residual force vector as the import force and the sub area grade limens value as the export force. The second layer NN is trained by using the strain mode vector of pole end in damage area as the import force and the damage degree of pole end in damage area as the export force. The first step of two-step detection approach of index analysis and reverse analysis identifies the location of the possible damaged poles by analyzing the joint residual force vector. The second step looks on the damage factor of the possible damaged poles as design vector, the 2 norm of the difference of the rod end strain mode vector between the intact and damaged structure as the target function, And uses the hybrid genetic algorithm to do reverse analysis, gains the damage degree of specific damaged joint at last. Performance of the two kinds of proposed damage detection approach is demonstrated with analysis of a plane tower structure. The result turns up trumps.(6) In order to put theory into practical project, xiangjiang transmit electricity tower is taken for example in this paper. At first,...
|
PDF Full Text Request