Auxiliary Expert System For Large Span Spatial Structure Selection With Genetic Neural Network

With the evolution of society,the advance of science and technology,the enhanced of the country’s overall strength.Airport terminals,sports venues,exhibition centers and aircraft hangars and other large space complex buildings came into being so that the construction industry in a great opportunity to challenge,at the same time.This challenge is a complex design of large spatial structure,which makes it possible to increase the structure type of improperity,which puts forward higher requirements for the design of large span space structure.Now,the design specification of the selection study of domestic large-span spatial structure can not fully meet the design requirements.large-span spatial structure design has some problems like unreasonable selection,high cost and so on.Therefore,the designers pay more and more attention to the problem of structural selection of large space structure design.At the same time,because structural selection has many uncertainty factors.It is very difficult to model and solve this problem with traditional methods.Therefore,to find the scientific method of structural selection,establish an intelligent system,you can solve the cost,design cycle and other issues.Artificial neural network can be realized by learning and training.Artificial neural network combines the knowledge of acquisition,representation and reasoning into one.Artificial neural network can use the important characteristic knowledge which relatively easy to obtain.Therefore,it is necessary to focus on the research of structural selection design system which based on the artificial neural network.In this paper,an improved genetic neural network method used to solve some problems that common BP neural network limitations far and ommon BP neural network may fall into local optimal solution.This method not only uses artificial neural networks highly nonlinear,self-learning characteristics and real-time processing,high fault tolerance and robustness but also utilizes the global optimization ability of genetic algorithm,storage and use of expert experience,so that it can be easier to solve the structural selection problem as long as ordinary designers enter some basic related information.In this paper,a large number of cases of spatial structure of the building are collected,theexample library is established,the main influencing factors of the structural selection are extracted,the model of structure choice of genetic algorithm integrated neural network is built by MATLAB neural network toolbox.The human-computer interactive interface is selected in MATLAB language make the selection process is simple and clear.