Geological Target Recognition Based On Physical Property Data With Convolutional Neural Network

In the era of geological big data,how to quickly and effectively extract and mine geological target information from a large number of multi-source geological data has become a new research difficulty.In order to solve this problem,the deep learning method is introduced into the geological target recognition in this paper.Taking the identification of iron ore as an example,the convolutional neural network model is selected to carry out experiments,and the general experience of the geological target recognition method based on deep learning is summarized.In this paper,different data set construction schemes are proposed through program simulation and geological model design.By comparing the performance of different models,this paper uses Ushaped convolutional neural network to identify irregular geological targets in complex geological background,establishes the connection between geological structure and geological targets,improves the anti-noise ability of the network,and finally compares the influence of different physical property input on the prediction results of the network.he geological target recognition experiment shows that the deep convolutional neural network has a good effect on the geological target recognition.The effect of network recognition is closely related to the way of label setting in the data set.When some information is missing,the recognition effect of network depends on the degree of differentiation of the input information for geological targets.