Study On Velocity Auto-picking Method Based On The Neural Network

With the rapid development of geophysical exploration and computer technology,high-density,high-precision and multi-component exploration become conventional and the deployment of seismic instruments with 10,000 channels make high-density,high-precision explorations possible.The application of these technologies make seismic data size become larger and larger,from megabytes twenty years ago to gigabytes ten years ago,even a few terabytes or petabytes currently,so geophysical industry has to face great challenge in seismic data processing and interpretation.In the process of conventional seismic data processing,velocity picking needs a lot of manual operation by processing technicians,which is the most manual intervention in the whole process of seismic data processing at present.With the increasing amount of data to be processed,the time and cost of velocity picking work are further improved,and the problems it faces become more and more prominent.In order to overcome the problem of time-consuming and low efficiency in velocity picking,this paper proposes an a velocity spectrum auto-picking technology based on artificial intelligence neural network by analyzing the process of velocity picking,and realizes velocity auto-picking by imitating the methods of data processing personnel in velocity spectrum picking.Its core is to transform the problem of velocity picking into the problem of image detection and recognition of energy heaps.Based on You Only Look Once(YOLO)and its improved version of V3,after training,the ‘time-velocity' pairs that needs to be picked up in the input velocity spectrum can be identified,detected and output to the energy group,so as to realize velocity auto-picking,and the results of the two models are compared.The test results of synthetic and real seismic data show that compared with the traditional velocity picking algorithm,the auto-picking method has great advantages,which not only does not need intervention and constraints,but also realizes velocity auto-picking,greatly reducing the manual and time costs,It improves the efficiency of velocity picking and keeps a good precision in a certain range.