| Seismic exploration is the main method to get the information about geological structures and distribution of oil and gas underground. Residual static correction is a very important method to improve the quality of seismic section. With the deep learning and understanding about the assumptions and non-linear characteristic of residual statics correction, this thesis focuses on the simulated annealing algorithm and genetic algorithm, and its application in the residual statics correction problem. A hybrid optimization method which combined with the maximum energy method is presented. The main contents are as follow:(1) Study on the basic theory of residual statics correction of seismic data, including its assumptions and non-linear, and indicating that residual statics correction is a large-scale, multi-parameter, multi-extremes nonlinear optimization problem. For these characters of statics, study on the basic theory of simulated annealing and genetic algorithm.(2) Study on the objective function, model trace building. For the huge of the seismic data, A MPI-Open MP parallel processing framework has been designed, in order to improve the speed of algorithm.(3) The maximum energy residual statics correction implemented. Study on the parameter selection and step of simulated annealing and genetic algorithm in solving the problem of residual statics correction. The result of residual statics correction using these methods was tested with both theoretic and real seismic data, and gets a good result. But there are still some problems including the edge of axis is not continuous and false structures.(4) A hybrid optimization methods presented by combined with the advantage of each methods. Core algorithms with C were realized and software modules of the residual statics correction were implemented on QT development platform based on Linux operating system. And we have proved and tested hybrid methods with both theoretic and real seismic data. |
PDF Full Text Request