Scholarlink

<p>Cracks in oil pipelines pose significant risks to the environment, public safety, and the overall integrity of the infrastructure. In this paper, we propose a novel approach for crack detection in oil pipes using a combination of 3D drone simulation, convolutional neural network (CNN) feature extraction, and the dynamically constrained accumulative membership fuzzy logic algorithm (DCAMFL). The algorithm leverages the strengths of CNNs in extracting discriminative features from images and the DCAMFL’s ability to handle uncertainties and overlapping linguistic variables. We evaluated the proposed algorithm on a comprehensive dataset containing images of cracked oil pipes, achieving remarkable results. The precision, recall, and F1-score for crack detection were found to be 96.5%, 97.3%, and 95.6%, respectively. These high-performance metrics demonstrate the algorithm’s accuracy and reliability in identifying and classifying cracks. Our findings highlight the effectiveness of integrating advanced simulation techniques, deep learning, and fuzzy logic for crack detection in oil pipelines. The proposed algorithm holds promise for enhancing pipeline surveillance, improving safety measures, and extending the lifespan of oil infrastructure. Future work involves expanding the dataset, fine-tuning the CNN architecture, and validating the algorithm on large-scale pipelines to further enhance its performance and applicability.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Improving Oil Pipeline Surveillance with a Novel 3D Drone Simulation Using Dynamically Constrained Accumulative Membership Fuzzy Logic Algorithm (DCAMFL) for Crack Detection

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年5月12日

查看全文
  • Omar Saber Muhi,
  • Hameed Mutlag Farhan,
  • Sefer Kurnaz

摘要

Cracks in oil pipelines pose significant risks to the environment, public safety, and the overall integrity of the infrastructure. In this paper, we propose a novel approach for crack detection in oil pipes using a combination of 3D drone simulation, convolutional neural network (CNN) feature extraction, and the dynamically constrained accumulative membership fuzzy logic algorithm (DCAMFL). The algorithm leverages the strengths of CNNs in extracting discriminative features from images and the DCAMFL’s ability to handle uncertainties and overlapping linguistic variables. We evaluated the proposed algorithm on a comprehensive dataset containing images of cracked oil pipes, achieving remarkable results. The precision, recall, and F1-score for crack detection were found to be 96.5%, 97.3%, and 95.6%, respectively. These high-performance metrics demonstrate the algorithm’s accuracy and reliability in identifying and classifying cracks. Our findings highlight the effectiveness of integrating advanced simulation techniques, deep learning, and fuzzy logic for crack detection in oil pipelines. The proposed algorithm holds promise for enhancing pipeline surveillance, improving safety measures, and extending the lifespan of oil infrastructure. Future work involves expanding the dataset, fine-tuning the CNN architecture, and validating the algorithm on large-scale pipelines to further enhance its performance and applicability.