波形分类技术在川东生物礁气藏预测中的应用

川东地区生物礁主要发育于上二叠统的长兴组，是勘探高产油气藏的理想目标，但由于地质特点的多样性和复杂性，对生物礁的地震预测难度很大。为此，探讨了波形分类技术的原理、波形分类的关键参数及神经网络波形分类的处理流程和关键环节，总结了生物礁的特殊地震反射特征，运用神经网络波形分类技术对川东地区三维地震资料进行了波形分类。根据波形分类结果划分的3个地震相带与前人研究的区域沉积相带，即海槽相、陆棚边缘相和碳酸盐岩台地相，有非常好的对应关系，划分出的沉积相带边界更加准确；在陆棚边缘相（生物礁有利发育相带）内准确地预测了生物礁的分布范围。

Application of waveform classification technique in predicting reef gas pool in east Sichuan basin

In east Sichuan basin,the reefs are mainly developed in Chan- gxing formation of upper Permian and are an ideal target to explore high-production reservoirs.Because of diversity and complexity of geological features,the seismic prediction of the reef is very diffi- cult.The principles and the key parameters of waveform classifica- tion technique were discussed,the processing flows and the key links of neural network waveform classification technique analyzed, and the particular seismic reflectance features of the reefs summa- rized.The 3D seismic data of east Sichnan was classified by the neural network waveform classification technique.The classifica- tion results have a good relationship to the regional sedimentary fa- cies,including trough,margin continental shelf,and carbonate platform.The boundary of sedimentary facies divided by this meth- od is quite accurate.On the basis of this,the distribution of the reefs in the margin continental shelf where reefs are well developed has been accurately predicted.