模擬退火演算法和基因演算法於自動化速度之選取

Abstract

我們系統使用模擬退火演算法以及基因演算法來做反射震測訊號處理中速 度之選取。速度之選取是在速度的 semblance 圖中挑選出 time-velocity pairs 來表 示出地層間速度與時間之間的關係函式。傳統的速度選取方法是由有經驗的專家 透過觀察 semblance 以及參考當地地質勘測資料來進行選取，過程非常耗時。我 們在 semblance 上選取峰點，峰點可連接而成一條折線 (polyline)，使得此折線 所通過的峰點能量和為最大，且滿足一些速度限制的條件。以 maximum filter 決 定 semblance 中極大峰點之所在位置後，在峰點排列編碼而成的龐大有限解中， 利用模擬退火演算法和基因演算法計算找到最佳速度圖。實驗結果顯示使用基因 演算法的表現較好。得到的速度可再進一步用來執行 normal moveout correction 和 stacking。我們的研究結果除了將速度選取自動化，更有助於進一步的震測資 料解釋。

We propose a system of automatic velocity picking in seismic data using simulated annealing and genetic algorithm. Velocity picking is to pick the time-velocity pairs representing the relation of stacking velocity and time based on semblance image. Conventionally seismic velocity picking is processed by geophysical experts through looking at the semblance image and geological surveying, which is time consuming. In the semblance image, we choose peaks to become the velocity plot. This velocity plot passes through the peaks having the maximal summation of energy, and satisfies some velocity constraints. So the velocity picking can be formulated into a combinatorial optimization problem. We determine maximal peaks on semblance by the maximum filter, then use simulated annealing and genetic algorithm to find the global optimal solution in the large finite set of peak combinations. The result of genetic algorithm is better. Then we apply the obtained velocity to do the normal moveout correction and stacking. The whole process is automatic, and the velocity picking can improve the seismic interpretation as well.