Forecasting and characterization of composite pipeline based on experimental modal analysis and YUKI-gradient boosting

Abstract

The surging demand for fiberglass pipe materials in the Oil and Gas transportation sectors are driven by their exceptional durability and ease of use. This study combined improved experimental and numerical analyses to assess the behavior of composite pipe and focused on characterizing it, especially in terms of identifying the frequency based on different crack lengths. Various tests, such as impact hammer tests and modal analysis, determine the pipe properties and natural frequencies. The study also investigates the impact of varying fiber volume fractions to refine material properties. To understand crack effects behavior, different crack lengths have been introduced into the pipes and analyzing the modal analysis using frequency and mode shapes. The effect of cracks into frequency based on experimental and developed numerical models is investigated. A predective model is developed based on YUKI-Gradient Boosting algorithms and Deep Artificial Neural Networks. To test the robustness of developed algorithm several tests are provided and the results show the effectiveness of YUKI-Gradient Boosting outperforms YUKI-Deep-ANN in accuracy and consistency.