摘要
水平移动系数是开采沉陷预计的重要参数,对于确定地表变形值、从而准确判定建筑物的破坏程度具有决定性的作用,然而其影响因素十分复杂,很难用一定的理论公式来描述。利用神经网络自学习、自主对复杂的非线性关系的拟合原理,通过对大量样本的学习训练,获得了基于神经网络的水平移动系数求取方法,并通过样本测试,取得了较好的效果,为水平移动系数的求取提出了一种新的方法。
As an important parameter of mining subsidence prediction, the horizontal displacement factor, has a decisive role to confirm ground deformation value so as to judge the destructiveness accurately. But its influence factors are complex and difficult to describe with some theoretical formula. This paper uses neural network serf-learning, independent of complex nonlinear relation fitting principle, and training on a large number of samples, obtains the access to horizontal displacement factor based on neural network, moreover, gets through of sample tests and achieve good effect. Thus it opens up a new way to gain displacement factor.
出处
《地矿测绘》
2012年第3期30-32,共3页
Surveying and Mapping of Geology and Mineral Resources
关键词
水平移动系数
神经网络
样本训练
模型测试
horizontal displacement factor
neural network
sample training
model test