Researchers Develop Novel Framework for Identifying Delamination in BFRP Pipes

Strategic Insights -

Researchers from Alexandria University have created a groundbreaking framework for detecting delamination in basalt fiber-reinforced polymer (BFRP) pipes. According to a study published in the International Journal of Lightweight Materials and Manufacture, the team developed a novel framework that combines convolutional neural networks (CNNs) with capacitive sensors to accurately identify delamination location and size. The research demonstrates a high level of efficiency and accuracy, with a training accuracy of 95.2%, recall rate of 93.7%, and F-score of 90.9%.

Key Takeaways:

  • The researchers employed a failure detection-based Electrical Potential Change (EPC) technique to inspect internal layers delamination in BFRP pipes subjected to long-term fatigue loading.
  • The 3D maps of capacitance array values and EPC distribution of node potential were tested, but found to be less accurate due to the bending and twisted effects of the 3D model.
  • A convolutional neural network (CNN) algorithm was adopted to train and test the EPC maps to evaluate delamination location and size.
  • The training accuracy, recall rate, and F-score of the current technology were found to be 95.2%, 93.7%, and 90.9%, respectively.
  • The proposed method's results converged with traditional methods in the literature, such as response surface methodology (RSM), with an error band from the diagonal line of less than 4.86 and 1.14 degrees for location and size, respectively.
  • The framework's reliability, accuracy, and applicability for relevant structures were validated.
  • A novel framework to identify delamination location/size in BFRP pipe based on CNN algorithm hybrid with capacitive sensors was proposed and demonstrated.

Statistics:

  • Training accuracy: 95.2%
  • Recall rate: 93.7%
  • F-score: 90.9%
  • Error band from the diagonal line for location: less than 4.86 degrees
  • Error band from the diagonal line for size: less than 1.14 degrees

Sources:

  • International Journal of Lightweight Materials and Manufacture, 2025,8(3):393-401.
  • doi-org.sdpl.idm.oclc.org/10.1016/j.ijlmm.2024.12.002
  • Keywords: Alexandria University, Alexandria, Egypt, Africa, Algorithms, Neural Networks, Machine Learning, Convolutional Network, Emerging Technologies.