高坝基岩多点变形监测的GRNN模型研究

为有效地进行大坝基岩多测点变形分析预测，在既有的大坝变形安全监测数学模型结构基础上，利用广义回归神经网络（GRNN）良好的非线性拟合能力建立变形预测模型，并针对高坝基岩多点位移计监测的实际情况，以多个测点的变形量为分析对象，在利用历史变形资料进行训练后实现多点变形预测。实例计算与比较结果表明，GRNN模型计算快、精度高，是进行多测点非线性变形监测预报的有效工具。

GRNN Model Study on Multi-point Deformation Monitoring in the Bedrock of High Dam

In order to forecast multi-point deformation of the bedrock of high dam more accurately,the General Regression Neural Network(GRNN) has been applied here because of its nice ability of nonlinear fitting.On the basis of the dam monitoring model frame,and using the multi-point survey deformation as output objects,the GRNN monitoring model has been formed.By the training with history data samples,the model can do comprehensive forecast of multi-point deformation.Example with survey data and compare with BP both showed that the GRNN model makes an effective way to forecast multi-point deformation with rapid calculation and high accuracy.