本文選題：剩余靜校正　切入點：模擬退火算法　出處：《電子科技大學》2015年碩士論文

【摘要】：地震勘探是獲取地質構造和地下油氣分布信息的主要方法,剩余靜校正處理對提高地震剖面的質量和地震資料的后續(xù)處理十分關鍵。本文圍繞地震資料的剩余靜校正問題,從剩余靜校正的理論及其假設條件和非線性特點出發(fā),重點研究了模擬退火算法和遺傳算法,以及其在剩余靜校正問題中的應用,結合最大能量法,形成了一種求解剩余靜校正問題的混合優(yōu)化算法。主要研究內(nèi)容如下:(1)研究了地震數(shù)據(jù)靜校正的基本理論,包括剩余靜校正的假設條件和非線性特點,說明剩余靜校正問題是一個大規(guī)模、多參數(shù)、多極值的非線性最優(yōu)化問題。針對剩余靜校正問題的非線性特點,研究了模擬退火算法和遺傳算法兩種非線性優(yōu)化方法的理論基礎,包括模擬退火算法和遺傳算法的基本原理,算法流程以及特點;(2)研究了求解剩余靜校正問題的目標函數(shù)、模型道構建。針對地震資料數(shù)據(jù)量龐大的問題,設計了MPI-OpenMP并行處理框架,以提高算法的求解速度;(3)實現(xiàn)了最大能量法剩余靜校正,重點研究了求解剩余靜校正問題時,模擬退火算法、遺傳算法的參數(shù)選取,實現(xiàn)了模擬退火算法和遺傳算法求解剩余靜校正的步驟。通過理論模型數(shù)據(jù)和實際地震資料數(shù)據(jù),對三種方法的校正效果進行了測試,取得了較好的校正效果,但是仍然存在邊緣部分同向軸不連續(xù)、出現(xiàn)假的同向軸構造等問題;(4)針對單獨算法求解剩余靜校正問題時存在的不足,結合各個算法的特點,利用不同算法的優(yōu)勢,形成了混合優(yōu)化方法,并在Linux環(huán)境下,基于QT平臺,采用C語言開發(fā)了混合優(yōu)化剩余靜校正模塊。通過對理論模型和實際地震資料的處理,取得了較好的校正效果,消除了單獨算法的不足,驗證了混合優(yōu)化方法和軟件模塊的正確性和可行性。
[Abstract]:Seismic exploration is the main method to obtain information of geological structure and underground oil and gas distribution. Residual static correction is very important to improve the quality of seismic profile and the subsequent processing of seismic data.This paper focuses on the residual static correction problem of seismic data. Based on the theory of residual static correction and its hypothetical conditions and nonlinear characteristics, the simulated annealing algorithm and genetic algorithm are studied, and their applications in residual static correction problems are also discussed.Combined with the maximum energy method, a hybrid optimization algorithm for residual static correction problem is developed.The main contents of this paper are as follows: (1) the basic theory of static correction of seismic data is studied, including the assumptions and nonlinear characteristics of residual static correction. It is shown that the residual static correction problem is a large-scale, multi-parameter, multi-extremum nonlinear optimization problem.According to the nonlinear characteristics of residual static correction problem, the theoretical basis of two nonlinear optimization methods, simulated annealing algorithm and genetic algorithm, are studied, including the basic principles of simulated annealing algorithm and genetic algorithm.In this paper, the objective function for solving residual static correction problem and model trace construction are studied.Aiming at the problem of large amount of seismic data, a MPI-OpenMP parallel processing framework is designed to improve the solution speed of the algorithm. The residual static correction of the maximum energy method is realized. The simulated annealing algorithm for solving the residual static correction problem is studied emphatically.The parameter selection of genetic algorithm realizes the steps of simulated annealing algorithm and genetic algorithm to solve residual static correction.Based on the theoretical model data and the actual seismic data, the correction results of the three methods are tested, and good correction results are obtained, but the edge parts are still discontinuous in the coaxial direction.In view of the shortcomings of the single algorithm in solving the residual static correction problem, combining the characteristics of each algorithm and utilizing the advantages of different algorithms, a hybrid optimization method is formed under the Linux environment.Based on QT platform, a mixed optimization residual statics module is developed with C language.By processing the theoretical model and the actual seismic data, a good correction effect is obtained, the shortcomings of the single algorithm are eliminated, and the correctness and feasibility of the hybrid optimization method and the software module are verified.
【學位授予單位】：電子科技大學
【學位級別】：碩士
【學位授予年份】：2015
【分類號】：P631.4

【參考文獻】

相關期刊論文 前2條

1 茍量,彭真明,王克斌;混合優(yōu)化自動剩余靜校正方法研究[J];石油地球物理勘探;2005年04期

2 彭真明,伏敏,梁國勝,龍雨豐;地震勘探中的非線性智能反演技術[J];自然雜志;2001年02期

，

本文編號：1722962