Frequency dispersion curve automatic picking method based on machine learning

The invention discloses a frequency dispersion curve automatic picking method based on machine learning. The method comprises the following steps: preprocessing seismic data; based on the pre-processed seismic data, utilizing a preset classification algorithm to obtain an area where a frequency dispersion curve is located; and extracting the area where the frequency dispersion curve is located to obtain a final frequency dispersion curve. The frequency dispersion curve can be prevented from being manually extracted, the seismic data processing workload is reduced, and the method has great significance in large-scale seismic data processing. 本发明公开了基于机器学习的频散曲线自动拾取方法，包括：对地震数据进行预处理；基于预处理后的所述地震数据，利用预设分类算法，获取频散曲线所在区域；对所述频散曲线所在区域进行提取，获取最终的频散曲线。本发明能避免人工手动提取频散曲线，减小地震数据处理工作量，对大规模地震数据处理具有重大意义。