团体粒子群与最大能量串行融合算法转换波静校正

Serial fusion algorithm of group PSO and maximum energy method for converted-wave static correction

估计转换波的静校正量是一个复杂的非线性问题,常规的线性静校正方法无法取得好的效果.粒子群算法是一种很好的非线性全局最优化方法,但其缺点是"早熟"现象严重.最大能量法是一种常规求取静校正量的方法,局部寻优能力强且收敛速度快是其优点,但是当地震记录含有大的静校正量时易收敛于局部极值.本文在标准粒子群算法的基础上发展出了一种改进的粒子群算法:团体粒子群算法.并且通过对Rastrigin函数的寻优实验证明了其全局寻优能力优于标准粒子群算法.同时为了解决转换波静校正问题串行融合了团体粒子群算法和最大能量法.最后,建立了含一个水平反射层的模型并合成地震记录,加入随机值作为检波点静校正量.对合成的地震数据分别利用团体粒子群和最大能量的串行融合算法、标准粒子群算法和最大能量法求取静校正量并进行静校正.结果证明串行融合算法得到的静校正量与理论值误差很小,静校正后的叠加剖面连续性较好.

Estimation of converted wave statics is a tricky, nonlinear problem. Those statics methods used in estimation of P-wave statics can hardly solve converted-wave static problem. PSO （Particle Swarm Optimization） is a good nonlinear global optimization method, yet ＂premature＂ is its weakness. The maximum energy method is a conventional statics method. The efficiency and strong local search ability is its advantage. However, when the record contains large statics, this method is easy to sink into the local minima. In this paper we develop a improved PSO method： Group PSO, which is based on the standard PSO. The experiment finding the minima of Rastrigin function using the Group PSO shows it has better global optimization ability than standard PSO. In order to solve the converted wave statics problem, we fuse the Group PSO and maximum energy method serially. Finally, we build a model which contains a horizontal reflector and obtain the synthetic seismogram in which we add a series of random values as receiver statics. Then we attempt to resolve the converted wave statics correction problem using the serial fusion algorithm, standard PSO and maximum energy method. The result shows that the statics from serial fusion algorithm are more accurate and the continuity of stack section using these statics is very good.