瓦斯涌出量的人工神经网络预测的研究

Prediction of Gas Emission with Artificial Neural Network

基于反向传播神经网络、径向基函数神经网络和广义回归神经网络对采煤工作面瓦斯涌出量进行预测,比较和分析了瓦斯涌出量的预测值和实测值,并选定精度评价体系对预测结果进行评定。结果表明:反向传播神经网络、广义回归神经网络和径向基函数神经网络都能够较好的预测瓦斯涌出量,其中径向基函数神经网络的预测更精准。径向基函数神经网络的后验差检验比值c=0.07,小误差概率P=1.0,预测误差精度达到1级。

The outflow volume of gas on the coal mining surface is predicted in the present paper based on reverse propagation neural network,radial basis function neural network and general regression neural network,the prediction and measured value of gas excitability is compared and analyzed,and the accuracy evaluation system is selected to evaluate the results.It shows that the reverse propagation neural network,the generalized regression neural network and the radial function neural network can all predict the gas outflow well,but the prediction of the radial function neural network is more accurate.The posterior difference test ratio of RBFNN is 0.07 with small probability error of 1.0.Therefore,the precision of prediction reaches the level of 1.0.