【摘要】 为了降低采集过程中各种噪声干扰以及老数据的磁带掉粉等因素造成的地震数据异常,处理与地层性质无关的抖动和毛刺干扰.采用时序RBF神经网络在线重构的方法.结果表明:在保留实际信息的前提下能有效的去除地震数据中的异常,使地震相与实际的地质体更加贴近.该方法具有很好的灵活性、及时性和低耗性,重构后的数据更加有利于后期解释工作和油气藏的描述.更多还原

【Abstract】 This study aims to reduce the noise interference and the powder dropping in the data collection process which lead to all sorts of abnormalities of seismic data and to handle the jitter and burr interference unrelated to formation properties.We have adopted the online reconfiguration of timing RBF neural network.The result shows that the method is effective in removing the abnormal seismic data without the loss of factual information and makes the seismic facies closer to geological body.The method features flexibility,timeliness and low consumption.The reconfigured data is conducive to the interpretation at the later stage and the description of oil-gas reservoir.更多还原