基于动态模糊RBF网络的基坑沉降预测研究

本文通过构建动态模糊RBF网络对基坑监测中的累计沉降数据进行处理和预测。采用误差下降率和高斯函数中心宽度的调节方法，使网络模型的结构紧凑，呈现动态变化，避免传统网络模型所出现的过拟合现象，提高网络的泛化能力，对基坑监测点的累计沉降量进行预测，通过对比时间序列模型，得出本文所采用的方法具有很高的预测精度。

Based on the dynamic fuzzy RBF networks of foundation pitsubsidence prediction research

Dynamic Fuzzy RBF are network by building on the foundation's total settlement monitoring data processing and forecasting. Error rate of decline in the use of center width of the Gaussian function and regulation of approach, the network model, a compact, dynamic, avoiding the traditional network model over-fitting phenomenon emerged to improve the network generalization ability of the pit monitoring points predict the total settlement, by comparing time-series model to arrive at the methods used in this paper has high prediction accuracy.