考虑多影响因素的BP神经网络预测软土地基沉降研究

Research on Prediction of Soft Ground Settlement based on BP Neural Network with Multi Factors

为了有效地预估软土地基产生的沉降量,针对地基沉降受多种因素影响和制约的特点,采用神经网络方法,建立了综合考虑地基沉降影响因素和处理方法影响因素的水泥搅拌桩法和竖向排水体法地基处理沉降预测BP神经网络模型。结果表明,所建网络输入矢量不仅考虑了影响地基沉降的共有因素（路堤剖面形态、软土地基工程特性、施工期等）,还考虑了水泥搅拌桩和竖向排水体处理软基的沉降影响因素（施工方式、加载方式、桩身强度、置换率、有效排水直径、当量直径等）,且训练样本选自不同的试验工点,模型适用范围广,可以较好地用于软土地基工程的沉降预测。

In order to effectively estimate the settlement of soft soil ground, according to the characteristics of ground settlement affected by many factors, the BP neural network model is established to predict settlement for soft soil foundation treatment method of cement mixing pile and vertical drainage body. The results show that the input vector network of the BP neural network model not only considers the influence of foundation settlement factors （embankment profile, soft soil engineering characteristics, construction period,etc. ） ,and the influence factors for soft foundation method of cement mixing pile and vertical drainage body （construction method, loading means, pile strength, replacement rate, effective drainage diameter, equivalent diameter）are also considered, and the training samples are selected from different models, the BP neural network model is applicable in wide range,so it can be used for settlement prediction of soft foundation.