井间地震速度反演的遗传算法

A GENETIC ALGORITHM FOR THE CROSSWELL SEISMIC VELOCITY INVERSION

井间地震速度反演具有重要的意义,不仅对于准确圈定地下构造的传统地震勘探是重要的,而且速度的变化可以指示非背斜型的地层岩性圈闭。传统的射线追踪方法一般是先给出入射角,然后根据折射定律修改入射角以达到反演的目的。该方法对每条射线均需通过扫描确定入射角,计算量较大,对初始地质模型的要求较高。遗传算法具有全局收敛性,是一种不用梯度信息的优化方法,特别适用于大型的组合优化问题。采用多项式展开表示界面深度和速度,通过弯曲射线追踪算法来计算射线的旅行时间,利用遗传算法进行迭代优化,可以同时反演出地下复杂构造的界面形态和速度变化。该方法与网格化速度反演相比,减少了未知参数,较好地克服了多解性问题,试验结果表明,该方法可以达到较高的反演精度。

Crosswell seismic velocity inversion is of great significance as it can not only actually determine the undermine construction, but also the change of the velocity can indicate non-anticline stratigraphic-lithologic trap. In the traditional ray tracing technology, the incidence angle must be given at first and then the inversion can be carried out through modifying the incidence angle based on the refractive law. The conventional method determines the incidence angle for every ray through scanning, Its computation is abundant and needs to provide a relative precise preliminary geologic modal. A genetic algorithm with global constringency does not need to acquire the gradient and can be easily adapted to solve the large-scale assembling optimization method. The depth and velocity can be expressed by polynomial, and the traveling time can be calculated by the method of crooked ray tracing. The structure and velocity varying of the complex construction can be computed out through iteration by using genetic algorithm. Contrary to grid velocity inversion, this method has less unknown various, and can preferably overcome the multi-answers problem. The model testing shows that its precision is very high.