改进PSPNet的电成像测井裂缝自动识别  

Improved Automatic Crack Identification for Electrical Imaging Logging Using PSPNet

作　　者：申科 肖小玲[1,2] 张翔[2] 林茂山[3] SHEN Ke;XIAO Xiao-ling;ZHANG Xiang;LIN Mao-shan(School of Computer Science,Yangtze University,Jingzhou 434023,China;Oil and Gas Resources and Exploration Technology,Ministry of Education,Yangtze University,Wuhan 430100,China;Tuha Branch,China National Logging Corporation,Hami 839000,China)

机构地区：[1]长江大学计算机科学学院,荆州434023 [2]油气资源与勘探技术教育部重点实验室(长江大学),武汉430100 [3]中国石油集团测井有限公司吐哈分公司,哈密839000

出　　处：《科学技术与工程》2025年第7期2691-2702,共12页Science Technology and Engineering

基　　金：国家自然科学基金(41674136)。

摘　　要：针对裂缝特征提取困难导致裂缝分割精度低、网络参数量计算量大的问题,提出一种改进的PSPNet(pyramid scene parseing network)网络用于自动识别电成像测井图像中的裂缝。首先将PSPNet中的骨干网络替换为优化的MobileNetV3网络,减少网络参数量和计算量;其次,引入渐进特征金字塔(asymptotic feature pyramid network,AFPN),用于增加多尺度信息的交互,增强对细小裂缝的识别能力;接着,引入多深度卷积头转置注意力(multi-depthwise Conv head transposed attention,MDTA)进行全局特征的提取,提升关键信息的提取能力;最后,采用Focal Loss和Dice Loss组合相加作为损失函数,以解决数据集类别占比不平衡的问题。实验结果表明,改进的PSPNet网络对电成像测井裂缝具有较好的分割效果。与PSPNet网络相比,mIoU(mean intersection over union)提升了3.17%,mPA(mean pixel accuracy)提升了6.38%。此外,研究成果的参数量、计算量、权重分别比原模型减少94.3%、95.7%和93.8%。同时,开发了基于CIFLog的裂缝识别系统,该系统能够满足对电成像测井的实际需要。An improved PSPNet(pyramid scene parseing network)network was proposed to automatically identify fractures in electrical imaging logging images,which was difficult to extract fracture features and led to low segmentation accuracy and large calculation of network parameters.Firstly,the backbone network in PSPNet was replaced with the optimized MobileNetV3 network,which could significantly reduce the number of network parameters and the amount of computation.Secondly,the asymptotic feature pyramid network(AFPN)was introduced to increase the interaction of multi-scale information and enhance the recognition ability of small cracks.Then,multi-depthwise Conv head transposed attention(MDTA)was introduced to extract global features and improve the extraction ability of key information.Finally,the combination of Focal Loss and Dice Loss were used as a loss function to solve the problem of unbalanced proportion of data sets.The experimental results show that the improved PSPNet network has a good segmentation effect on the fracture in the electrical imaging logging.Compared with the PSPNet network,mIoU(mean intersection over union)improved by 3.17%and mPA(mean pixel accuracy)improved by 6.38%.In addition,the number of parameters,calculation amount and weight of the proposed algorithm are reduced by 94.3%,95.7%and 93.8%respectively compared with the original model.At the same time,the crack identification system based on CIFLog is developed,which can meet the practical needs of the electrical imaging logging.

关 键 词：PSPNet 裂缝识别 电成像测井图像 MobileNetV3 AFPN 

分 类 号：P631[天文地球—地质矿产勘探] TP391.7[天文地球—地质学]