Research And Application Of Stacking Velocity Picking Algorithm Based On Particle Swarm Optimization And Residual Network

With the development of stacking velocity picking technology and the increasing expansion of geological signal data,the accuracy and efficiency of velocity picking have become important indexes to evaluate stacking velocity automatic picking algorithms.Traditional speed picking is generally achieved by manual,but the manual picking efficiency is low,and it is susceptible to the complex environment and artificial subjective factors,so the interpretation of the geological data by picking results lacks scientific.Therefore,this thesis designs a stacking velocity picking algorithm based on particle swarm optimization and residual network,discusses in detail the mathematical principles and applications of the algorithm from the field of velocity analysis,and realize the automatic picking process of stacking velocity using actual geological data as a model,which effectively improves the pickup efficiency and accuracy.First of all,Aiming at the low efficiency of manual picking and the deficiency of precision of traditional automatic picking algorithm,constrained velocity solution space technology and the particle swarm algorithm are introduced into the velocity optimization picking link of the velocity spectrum to improve the accuracy automatic picking of the optimal velocity.Its one,according to the signal similarity coefficient criterion,the original velocity solution space P is preliminarily constrained to obtain the velocity solution space’P;Its two,use kd-Tree’s nearest neighbor search to perform peak matching on channel data,and at the same time constrain the velocity solution space’P according to the signal in-phase criterion to obtain the velocity solution space’’P;Finally,by analyzing the particle swarm optimization problem,this thesis proposes a new idea to improve its inertia weights,then combine the velocity solution space’’P and the objective optimization function,and use the improved particle swarm model to automatically search and obtain the optimal solution to achieve the automatic picking of the optimal velocity.Secondly,aiming at there are many energy clusters in the particle swarm search space and some of the optimal energy clusters have abnormal items interfering with the automatic picking of stacking velocity.The residual network model is introduced to classify the slice characteristic graph of 3D data volume,and the diffusion process of individual particle extreme value in the process of particle swarm search is constrained based on the classification result of the residual network.This method effectively reduces the interference of the existence of multiple energy clusters and the abnormal terms in some optimal energy clusters to the automatic search of the stacking velocity,so that the error between the automatic picking results and the real reflection signal is smaller.And finally,in the application of actual geological data,the interpretation and verified of the algorithm picking results of this thesis by constructing 3D imaging maps,and it can be found that the best stacking velocity accuracy picked up by using this thesis’ s algorithm meets the application requirements,is more efficient compared with manual picking,and the calculated reflection interface features are obvious,which can provide scientific proof for the interpretation of relevant geological data.