灰色关联分析与BP神经网络的概率积分法参数预测

在综合分析地表沉陷概率积分法参数与地质采矿条件关系的基础上,提出运用灰色关联分析法找出影响概率积分法参数的主要因素,进而利用BP人工神经网络模型预计参数。在对实测数据灰色关联分析后得出:覆岩平均坚固性系数、采厚、倾角、采动程度与各个参数关联程度较高,表土层厚度和采深次之。在此基础上,建立BP人工神经网络模型,并对预计结果与实测数据进行对比分析。结果表明:该方法预计最大相对误差15.78%,最小相对误差1.92%,考虑到个别参数实测值较小,造成相对误差较大,而绝对误差很小,即模型预计效果较好,是一种预计概率积分法参数的有效方法。

The prediction of probability-integral method parameters based on grey relational analysis and BP neural network

On the basis of analyzing the relationship between probability-integral method parameters of surface subsidence and geologic and mining conditions,grey relational analysis is proposed to find out the main factors affecting probability-integral method parameters.And then the parameters were predicted by using the BP neural network model.After analyzing by the gray relational analysis in this paper,the results show that the average consistent coefficient of overburden rock,mining thickness,dip angle and mining extent had a higher relationship degree with parameters,followed by thickness of topsoil and height of mining.Establishing the BP artificial neural network model,the predicted results and the observed values are analyzed and compared with each other on this basis.The results show that the maximum relative error of the method is 15.78％,and the minimum relative error is 1.92％.Considering the smaller individual parameters measured which resulting in a larger relative error,the absolute error is very small.The result shows that the method estimate is accurate.It is an effective method for predicting the parameters of the probability-integral method.