粗集神经网络在煤与瓦斯突出预测中的应用

结合粗集理论的属性约简功能和人工神经网络的非线性映射特性,提出了煤与瓦斯突出的一种预测方法.首先用粗集理论对训练样本进行属性约简和降噪,然后将经过预处理的训练样本代入神经网络进行训练,获得稳定的网络结构,最后用训练好的神经网络对待测样本进行预测.实际应用表明:瓦斯压力、瓦斯放散速度、地质构造、煤的坚固性系数和开采深度是煤与瓦斯突出预测的必要指标;粗集神经网络模型具有较高的预测精度和良好的实用性,是一种十分有效的煤与瓦斯突出预测方法.

Application of rough set and artificial neural network to prediction of coal and gas outburst

A prediction method of coal and gas outburst was presented based on the combination of attribute reduction function of rough set theory and nonlinear mapping characteristics of artificial neural network. Firstly, attribute reduction and denoising were executed. Secondly,the neural network was trained,and a steady network structure was obtained. Finally,the testing samples were predicted by using the efficient neural network. Practical application demonstrates that:(1) gas pressure,gas emission rate,geological structure,protodyakonov coefficient of coal and mining depth are the indispensable indexes of coal and gas outburst;(2) the prediction model based on rough set and artificial neural network has high precision and good practicability,and is a very efficient method for predicting coal and gas outburst.