Research On Neural Network Algorithm For Rapid Design Of Overhead Traveling Crane

At present,many experts and scholars design crane metal structure by using optimization algorithm.The result of the theoretical optimal design is to seek the lightest weight.But,there is no emphasis on whether it can be used in engineering practice.The enterprise is not satisfied with the theoretical optimal design outcome.Therefore,it is necessary to study the inherent problem of optimal design of the overhead traveling crane metal structure.And,we have explored a new idea for developing CAD software to solve the design of overhead traveling crane metal structure problems rapidly.First of all,this paper studies a method for improvement of PSO(particle swarm optimization)by using the dynamic penalty function,get the improved PSO.It is used to the overhead traveling crane metal structure optimization design.Design scheme generator software based on PSO is developed by Visual C++6.0.This software is used to study essential problems of overhead traveling crane metal structure optimization design and to product the training samples of neural network.Then,this paper studies essential problem of overhead traveling crane metal structure optimization design by using design scheme generator software,to explore the reason of the theoretical optimal design cannot be used in engineering practice.It is found that the theoretical optimal design is obtained by using limiting stress and limiting depth-width ratio.And,the size of the beam is high and narrow.The thickness of the plate is the thinnest about 6mm.According to the requirements of the engineering practice for overhead traveling crane metal structure,it is found that,we need a better scheme to solve the actual problem of engineering in all aspects,not the theoretical optimal scheme.Computer cannot design a metal structure directly,but,it can optimize a metal structure using optimal algorithm.We can use optimization to replace the design,optimization play the role of design scheme generator,the dominance of the design and decision-making power lies in the designer.So the process of optimization should be controllable,and the optimal scheme must be adjustable.The training samples for neural network is obtained by using design scheme generator which generate design scheme for different design parameters consider of light weighted and meets the requirements of enterprise,and essential problems of overhead traveling crane metal structure optimization design.The training samples are stored in database.Based on Artificial Neural Network,the mapping relationship between the design parameter and design scheme is trained.Based on this mapping relationship,the rapid design mapping model of the traveling crane metal structure is built,and the rapid design of the metal structure is carried out by using mapping.As long as the design parameter is inputted,the design scheme is outputted immediately and meets the requirement.And the fast result is achieved.A new idea is developed for the development of traveling crane structure metal CAD software.Finally,combined with the essential problems of overhead traveling crane metal structure optimization design,we found that the rapid design method needs to be further improved.In order to solve the shortcomings in the rapid design method,two ideas to improve the rapid design method are proposed,which are: the method of "multi-model mapping" and the method of transforming the constraints into the design parameters.Finally,the method of transforming constraints into design parameters is found,which is easy to implement and can be used to improve the rapid design method with better results.In this paper,under the Windows system,using development tools by Visual C++6.0 software,the overhead traveling crane metal structure optimization design software is developed.And visualization design,design scheme generator,output design specification,rapid design function are achieved.The innovation of this paper lies in: exploring the essential problem of overhead traveling crane metal structure optimization design;using optimization as the generator of the design scheme,proposing the rapid design method of neural network,exploring a new idea for developing the traveling crane metalstructure CAD software.The method of converting constraints into design parameter is used to improve the rapid design method and get better results.