The Research On Identification Of Seismic Phase Structure Based On Convolution Neural Network

At present,the seismic facies structure location and identification mainly relies on artificial method.And subjective factors influence too much in the process of distinguishing geometry and reflection characteristics of seismic wave,which reduces the accuracy of seismic facies identification.In order to avoid the influence of subjective factors and improve identification accuracy,the paper tries to research and implement an automated method for seismic facies identification.Paper's main work includes:The first part,establishes data set with a number of seismic facies structure,then compares accuracy of the convolution neural network and the deep belief network in seismic facies structure identification,on the basis of the data set.The second part,get a convolution neural network model with a higher accuracy,by adjusting structure and methods of the convolution neural network.The third part,based on the convolution neural network model,locates and indentifies the seismic phase structure by traversal and regional labeling method,on the seismic section images.Paper finally forms a set of method based on convolution neural network technology,which can locates and identifies seismic facies structure,by using the data set built in the paper.And shows the method is effective and stable through the experiments.