基于MATLAB的RBF神经网络在建筑物沉降预测中的应用  

Application of the MATLAB-based RBF Neural Network in Prediction of the Building Settlement

作　　者：汪洋[1] 李明峰[1] 李恒杨[1] 

机构地区：[1]南京工业大学土木工程学院,南京210009

出　　处：《矿山测量》2006年第3期34-36,共3页Mine Surveying

摘　　要：介绍了RBF神经网络的模型和结构,提出将该网络应用于建筑物沉降预测。运用Matlab工具箱函数建立了沉降预测网络模型,编制了计算程序,通过工程实例验证了该模型的正确性和可行性,并和BP神经网络在收敛速度上进行了比较,结果表明RBF神经网络的收敛速度远远快于BP网络。The model and structure of the RBF neural network are described in the paper and then the network is used for prediction of the building settlement.Based on the Matlab tool-box function,the model of the settlement prediction network is established and the computation program is compiled.Proved by actual project examples,the model is correct and feasible.Furthermore,a comparison is made to convergence rate between the RBF and the BP neural networks,the results show that the convergence rate of the RBF neural network is far faster than that of the BP network.

关 键 词：RBF神经网络 建筑物沉降 Matlab预测 

分 类 号：P207[天文地球—测绘科学与技术]