The Model Updating Study Of Bolted Structure Issues Under High Temperature Environment

In recent year, model updating technique which aimed at improving the accuracy of simulation has been widely put into practice in both scientific and industrial department. Finite model optimization is very necessary considering complex high temperature which occurs commonly in real structures.Traditional dynamic model correction mainly revises design variables directly.However, high temperature model correction will be harder due to the design variables influenced by temperature when high temperature environment is involved. Thus, two grade correction is introduced here to do the job of structure dynamic correction on the basis of steady temperature profile.First in this article, basic principles and ways of construction of three common surrogate models are introduced which are classic RSM responding surface, RBF radial basis function and SVM support vector machine. Advantages and disadvantages as well as applicability under different conditions of these are studied. Two kinds of optimization algorithms, genetic algorithm and weighted self-adaptation method are described systematically. Genetic algorithm searches the optimum solution from different areas, which avoids the local optimum solution. Being the extension of traditional weighted method, the self-adaptation one utilized here mainly focuses on multi-objective optimization, which can extend the optimum solution set to nonconvex area of Pareto frontier and then present evenly distributed Pareto frontier.For the revise of models under high temperature, hierarchical revise is applied here. In this way, thermal and structural dynamic parameters are constructed respectively. Then, this question of revise is transformed into one of optimization as a result of the combination of surrogate model and optimizing algorithm.Then a bolted connection model is validated to prove the feasibility of revise process mentioned above. The final results show that the revise method is feasible.