Application In High-rise Building Deformation Prediction Based On Neural Network Of GEP Optimization

With the rapid development of the society and human beings continue to converge to the city.The land resources of city are becoming less and less.A large number of high-rise buildings are generated.Deformation of high rise buildings may lead to the occurrence of disasters so that the safe problem seems to be particularly important.If it is possible to accurately predict the deformation of high-rise buildings,we could reduce the damage caused by the disaster effectively.However,there are some shortcomings in the deformation prediction of high-rise buildings which result in the prediction accuracy is not enough.Therefore,how to predict the deformation of high buildings accurately is significant to reduce the occurrence of disasters.This thesis describes the present situation of research on artificial neural networks and Genetic Expression Programming(GEP),and introduces the related theory of deformation monitoring of high-rise buildings and the current common prediction model.Secondly,Using GEP principles with a simple gene encoding and powerful global search ability to solve the selecting initial weights of the Radial Basis Function(RBF)Neural Network and the number of the hidden layer center vector(neuron),it’s also improved the convergence and accuracy of the network.Then,apply the advantages of GEP to the RBF neural network and optimize network.We establish a RBF neural network model based on GEP optimization.Finally,taking a high-rise building in Shanxi as an example,the prediction model based on RBF neural network and the prediction model of RBF neural network based on GEP are applied to predict the settlement deformation of high-rise buildings,A comparative analysis of the two models of the predicted results shows that the prediction model of RBF neural network based on GEP optimization to improve the accuracy of the predicted value nearly twice as the prediction model based on RBF neural network.And so to illuminate the prediction model of the RBF neural network based on GEP is more meaningful and valuable in the field of deformation prediction of high-rise buildings.