Optimal Design, Based On Variable Cross-section Cantilever Curvature Of The Modal Analysis Of Dynamic Stress Constraints

The structural dynamic optimum design is an advanced subject in the field of engineering structure design at present. The structural optimization problem under dynamic stress and displacement constraints is attached importance. In this paper, the method combined with curvature mode shape, finite element method(FEM), artificial neural networks(ANN) and genetic algorithms(GA) is applied. The optimization of a taper trunk beam of variable cross section under impact load can be obtained on the base of considering the objectives of minimum gross mass and the condition of constrain of dynamic stress.Firstly, with the curvature modal superposition method, the dynamic stress responses is calculated. The selection of design variables is discussed, to avoid choosing design variables that may result in unrealistic or undesirable design. Mathematical model for the optimization has been established in this paper. The design variables, range of design variables, target function and constrain function are confined.Secondly, a non-linear mapping function from multiple inputs(design variables) to output (maximum dynamic stress) is constructed within BP neural networks, in order to obtain the constraint function values that are necessary in optimum design of structures using genetic algorithms and get the optimum solutions. The generation ability of the BP network after training can be used to replace the calculation of FEM, which makes the constrained optimization to be possible.Thirdly, genetic algorithm is a kind of search and optimization method simulating the life evolution mechanism, which has the advantages of global optimization and implicit concurrency. Considering the constraint, the method is applied, by which the invalid individual will be changed into the valid one. The optimal results are tested with the FEM.