粗糙集与神经网络技术预测煤厚及小断层的方法

Method to predict coal seam thickness and small fault using RS and NN

提出了基于分析三维地震数据的粗糙集(RS)—神经网络(NN)技术,预测采区断层和煤层厚度变化。利用粗糙集对地震数据中所包含的大量干扰数据进行约简,生成低噪音数据;将约简后的数据输入神经网络进行训练获得断层识别和煤层厚度预测。实际数据验证表明,该方法具有较高的精度。

The thesis put forward a new method of Rough Sets（RS） and Neural Network（NN） technique to detect small faults and coal seam thickness by analyzing 3D seismic data.This method uses RS to reduce seismic data noise,and after reduction,low noise seismic data can be hold.After inputting those reduced data to NN,a predicting model which can detect small faults and predict coal seam thickness can be achieved after NN’s training.After this step,this model was used to detect small fault by 3D seismic data.We find that this method has high precision.