Study On The Structural Form Selection Of High-rise Building With GNN

The rapid development of the economic and technological development, The civilengineering and structural engineering are rapiding development and has made greatachievements in the21st century. The World especially in developed countries have plannedto establish a higher, greater and longer variety of large complex buildings. Although China isa developing country, because of the rapid development in recent years, economicdevelopment and overall national strength, the construction of a number of taller, bigger,longer building will soon become a very real demand, which is the field of China'sconstruction industry faced with the great opportunities and challenges.However, because of large and complex structure design of the complexity of the problemitself, The increase of choice also means that the structure of the system and the choice of theinappropriate type is aslo to greatly increase, the selection of undeserved can bring seriousconsequence and difficult to repair. Thus, the problem of structure selection in the design oflarge and complex structures is very important. Structure selection is increasingly being civilengineers and the majority of designers more and more attention.But the structure selection isa very complicated decision problem, has strong comprehensive, including much uncertainty(the fuzziness and the randomness, not the unascertained characteristic) information, using thetraditional standards or mathematical programming method to set a model to solve thisproblem is difficult to work. So, if we can find the civil engineering structure, especially largecomplex building's scientific decision-making method for structure design and set up acorresponding intelligent selection system, can to ensure that the design quality, shorten thedesigncycle,andi mprove the comprehensive performance of structure.The artificial neural network has the characteristics of knowledge intensive processingis knowledge acquisition, representation and reasoning are combined into a whole, theyare all through the study of the neural network training, which can make full use ofsample knowledge, and sample knowledge is relatively easily available,Therefore,Based on artificial neural network of large and complex structure selection of intelligent design system is the focus of the future research questions. the commonly used BPalgorithm of neural network can guarantee network learning process the final converg ence, there are obvious limitations, neural networks may not get the global optimal solution, but fall into the local optimal solution.So by using a kind of improved neural network-genetic neural network, and can solveabove problems. Based on genetic neural network method to carry on the choice of structuralsystem of high-rise building.It is the full use of the artificial neural network's highly nonlinear,high fault tolerance, robustness, self learning, real time processing and other characteristics,and use of the genetic algorithm for global optimization ability, also use the neural network tostore the expert experience in the design, so that the general designer put a number ofhigh-rise building structure basic information into the artificial neural network, Thus they caneasily solve the high-rise building structure system selection problem.Avoid throughcalculation and experience to determine the structure caused by the time waste and selectionerror.It has very vital significance for promoting the whole process of the civil engineeringstructure design 's automation, integration, intelligence, the Internet and digital. This papercollects a large number of domestic and foreign high-rise built architecture examples, buildhigh-rise building 's case data base, extract the the Structural Form Selection of High-riseBuilding's main influence factors, and application of MATLAB toolbox neural networkestablishing genetic artificial neural network model of the selection of high-rise buildings,And the application of MATLAB neural network toolbox to establish genetic artificial neuralnetwork model of the Structural Form Selection of High-rise Building, And with thetraditional neural network methods are compared, get their advantages and disadvantages.