针对油砂储层的岩心图像识别算法优选与应用

Optimization and application of core image recognition algorithm in oil-sand reservoir

刘焱鑫１ 黄继新２ 尹艳树 吕一兵 王超 齐建强4

长江大学地球科学学院，湖北 武汉 430100 中国石油勘探开发研究院，北京 100083 长江大学信息与数学学院，湖北 荆州 434023 中国石化中原油田分公司勘探开发研究院，河南 濮阳 457001

School of Geoscience, Yangtze University, Wuhan 430100, China Research Institute of Petroleum Exploitation and Development,PetroChina， Beijing 100083, China School of Information and Mathematics, Yangtze University, Jingzhou 434023, China Exploration and Development Research Institute, Zhongyuan Oilfield Company, SINOPEC, Puyang 457001, China

在地质研究及石油生产开发过程中，岩心资料的识别与表征具有重要意义。加拿大麦凯河油砂区块下白垩统麦克默里组储层岩心中发育数量众多的毫米级泥质夹层，人工识别工作量大，亟需采用图像识别算法自动识别。文中选择OTSU分割、粒子群双阈值分割、FCM聚类分割及神经网络方法，开展针对薄夹层自动识别对比研究。岩心识别结果表明，粒子群算法识别平均准确度达到90.29%，识别准确度最高，识别速度快、可靠性高且易于实现，能够完成大规模的岩心薄夹层识别工作。基于此，开发了一套薄夹层识别软件，服务于研究区及相似区块薄夹层的识别，为油藏勘探开发及地学数字化提供技术支持。

Recognition and characterization of core data are of important significance to geological studies and exploitation of petroleum. The reservoir core of Lower Cretaceous McMurray Formation of McMurray River oil-sand block in Canada is developed with abundant millimeter-scaled muddy interlayers. Therefore, the artificial recognition workload is heavy and automatic recognition based on the image recognition algorithm is needed urgently. In this study, various methods such as OTSU segmentation, particle swarm double-threshold segmentation, FCM clustering segmentation and neural network are selected to carry out the contrastive research on automatic recognition of thin interlayers. The core recognition results show that the average recognition accuracy of particle swarm optimization(PSO) algorithm is the highest, reaching 90.29%. Moreover, PSO algorithm is characterized by high recognition speed, high reliability and feasibility, and can recognize large-scaled core interlayers. Based on PSO algorithm, a set of thin interlayer recognition software is developed for recognition of thin interlayers in the study area and similar blocks. It provides strong supports for oil reservoir exploration and exploitation and the promotion of geoscience digitalization.

10.6056/dkyqt202004011