基于PCA-Elman的煤层小断层预测

Prediction of Small Faults Based on PCA-Elman Neural Network

针对煤层小断层发育的复杂性及其预测参数间的相关性,提出了基于主成分分析(PCA)与Elman网络的煤层小断层预测方法。该方法首先利用灰色关联分析确定小断层密度预测参数,然后利用PCA降维提取主成分,消除参数间的相关性,最后以主成分为输入样本,建立Elman网络预测模型。应用实例表明,煤层小断层PCA-Elman预测模型的预测效果较好,平均预测精度达94.1%。

Aimed at the complexity of the small faults in coal seams and the correlations among small faults prediction parameters, puts forward a method based on principal component analysis (PCA) and Elman neural network model for predicting the small faults. In this model, the prediction parameters were determined by using grey correlation analysis, and then principle component analysis was used to eliminate the correlations of the prediction parameters, the prediction model of small fault density was finally built through taking the results of PCA as inputs of the Elman neural network. The research result shows that the PCA- Elman neural network model for predicting small fault density has a nice prediction accuracy,the average prediction accuracy reaches 94.1%.