基于探地雷达属性预测路基含水率的模型实验研究

Model test study on moisture content prediction of subgrade based on GPR attributes

从探地雷达属性分析入手,研究探地雷达属性分析和BP神经网络相结合的路基含水率预测的方法。根据铁路路基模型含水率试验数据,优选出最大峰值振幅、总能量、主频带能量、百兆带宽能量百分比、峰值频率、平均瞬时相位、能量半衰时等7种探地雷达属性作为铁路路基含水率预测的基本参数,结合含水率测试资料,建立路基含水率BP神经网络预测模型,预测含水率与实际含水率的相关系数,反映铁路路基含水率与探地雷达属性之间的非线性关系。

Based on the analysis on ground penetrating radar(GPR)attributions,a method for predicting the water content level of railway subgrade was proposed by combining the attribute analysis of GPR and BP artificial neural network.Seven usable GPR attributes,including the maximum peak amplitude,total energy,dominant frequency band energy,energy percentage of hundreds megabytes bandwidth,peak frequency,average instantaneous phase and energy halftime,were selected as the basic analysis parameters of prediction models for the moisture content of railway subgrade,based on model test of moisture content of railway subgrade.According to the test data,a BP artificial neural network prediction model of the moisture content of railway subgrade was established.The trained network model was used to predict the moisture content of the railway subgrade model.The correlation coefficient between the predicted and actual moisture content values was predicted,which can well relate the moisture content of railway subgrade with GPR attributes.