神经网络在建筑物沉降分析中的应用

Application of neural networks to the analisis of building sedimentation

建筑物沉降的诱因与沉降量之间有一个复杂的非线性相关性,应用回归法对这种复杂的相关性进行分析有较大的局限性。人工神经网络是由许多神经元组成的大规模非线性系统,具有较强的动态处理能力,能对简单的非线性函数进行多次复合,来实现一个复杂的非线性函数。神经网络这些特性满足建筑物沉降分析的需求。实例表明,应用神经网络BP算法可以对建筑物沉降原因进行更客观的分析,对沉降趋势预测效果也较好。

There is a very complicated nonlinear pertinence between the inducements and the quantity of the building sedimentation, and analyzing this pertinence with recursive method has large limitation distinctly. Neural network is a large-scale nonlinear system composed by many neural cells. It has better dynamic disposing capability and can compound a simple nonlinear function to actualize a complicated function. Those characteristics adapt to the requirement of the building sedimentation analysis. Some examples indicate that applying BP algorithm of neural network can analyze the reasons of the building sedimentation more objectively, and can forecast the trend of sedimentation better.