| With the fast development of global industrialization and the crisis of world energy sources, currently how to utilize the sources efficiently is a key problem to be resolved. Therefore, aiming at the base frame, which is a kind of heavy box section plate's structure, of the hydraulic press, it is very important to carry out the structural topology optimization with the lightest weight as the objective. At the same time, that research on the practical, reliable and effective topology optimization method has a promote effect and theoretical value for the development of the design theory of the structure of the hydraulic press. Currently, topology optimization research on the structure of the hydraulic press has not been found with the best knowledge of the author. Therefore, with the ground structure method as a basic research method and with the press'base frame as the research objective, a systematical research on the topological optimization of the box section plate's structure was carried out in this work.The ground structures of press base frame were firstly set up to be used to topology optimization. On the base of analyzing function, load and structure features of press base frame, the basic principles of building ground structure was proposed. The mathematical models for the topology optimization were proposed as well.To increase effective of structural topology optimizing a whole research on the approximate analysis method with three layers BP neural network was underway. A rational method to decide the nodal number of the hidden layer of the three layers of BP neural net based on the genetic algorithm was firstly proposed. The D-optimization method of BP network teaching sample design was firstly proposed. The method is that using a quadratic regression equation without cross items approximate express the relation of plates'thickness and flexibility of press base frame. Under the condition of the sample quantity was given, the optimization problem was set up which aim is to minimum the regression variance of flexibility of base frame. By solving the problem the optimum teaching sample was obtained with most minimum regression variance of flexibility.A genetic algorithm was proposed to solve the problem of structural topology optimization of press base frame using structural approximate analysis with three layers BP network. In that algorithm, a rejecting strategy for treating constrains was used, which means that all individuals'displacement constrains are treat by BP network and only the optimum individuals'stability constrain was check by FEM buckling analysis.Two discontinuous phenomena in the topology optimization of the box section plate's system structure were firstly proposed. It was pointed out that the two mutational phenomena were the reason inducing that the structural approximate analysis method based on the BP neural net was difficult to be applied on the topology optimization. Therefore, the D-optimization design strategy of the teacher's sample was set up, which makes the trained BP neural net can be applied on the approximate quick computation of the structural displacement constraint. The method, which combined the mechanical initial check and the finite element buckling analysis, was utilized to deal with the stability constraints.The structural topology optimization was firstly implemented for the base frame of the hydraulic press in this work. Six optimum topological base frame structures light in weight and novel in layout were obtained.Finally, two 1:5 metal models, which technical parameter and performance is same, but topology and weight is different, as well as one of which is optimum base frame, were manufactured to do the experimental comparison, The experiments showed that the algorithm that was proposed by this work is practical, reliable and valid. |
