激光图像检测压实土密实度的方法研究

Detection Method of Compacted Soil Compactness by Laser Image

为了检测压实土密实度,搭建了激光图像检测密实度装置,并研究了该检测方法.选取压实土吸收系数、散射系数、激光光斑周长、图像灰度梯度作为神经网络模型的输入特征参数;最后,利用BP神经网络预测密实度.结果表明:(1)BP神经网络经过85次学习后达到了要求的误差;(2)利用系统预测的密实度数值与环刀法检测值相比较,平均绝对误差为0.115和平均相对误差为12.57%.因此,该检测方法检测压实土密实度是可行的.

In order to test soil compactness, a laser image measurement system of soil compactness was established and the nondestructive measurement method of soil compactness was investigated. The absorption coefficient and scattering coefficient, laser spot perimeter, the gray gradient were selected as the classifier input feature values. Finally, soil compactness was predicted by BP neural networks. The experimental results show that BP neural networks reached the required error after 85 loops, compared with the measurement values by using round knife method, the average absolute error is 0.115 and relative error is 12.57%. Therefore, the measurement of the system is feasible.