BP神经网络的改进及其在初至波拾取中的应用  

THE IMPROVED BP NEURAL NETWORK AND ITS APPLICATION IN SEISMIC FIRST BREAKS PICKING

作　　者：王金峰[1] 罗省贤 

机构地区：[1]成都理工大学信息工程学院,四川成都610059

出　　处：《物探化探计算技术》2006年第1期14-17,1-2,共4页Computing Techniques For Geophysical and Geochemical Exploration

基　　金：中国石油化工集团总公司项目资助(T03045)

摘　　要：在地震资料处理中,正确的初至时间,是解决复杂地表静校正问题所需的关键参数。将BP神经网络引入到初至波拾取中,并针对经典BP神经网络收敛速度慢、易于陷入局部极小的缺点,利用组合函数法、限幅法、动量因子法及自适应学习率法等进行了改进。用模拟试验证明了这些改进方法的可行性和有效性,在应用于实际的地震记录初至波拾取中,取得了良好的应用效果。In seismic data processing,the accurate first break time is the key parameter to solve static correction problems of complex surface.In the paper,the BP Neural Network is introduced to pick up the first break.And aiming at the shortcomings of BP neural network,such as the low speed in training,the tendency to local minimum,the limitation of amplitude,the momentum factor method and the self-adjusting learning rate are used to improve the BP algorithm.Then the feasibility and validity of the improved methods is proved by a series of simulation experiments.The application result by the method in the first break picking is satisfied.

关 键 词：BP神经网络 自适应学习速率 动量因子 初至波 初至拾取 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]