Fatigue crack identification in jacket-type offshore platforms using a parallel multi-scale convolutional neural network

Vibration-based damage detection techniques are widely applied in the structural health monitoring of offshore platforms. This study employs a parallel multi-scale convolutional neural network (PMSCNN) to analyze the acceleration response signals collected from offshore platforms, achieving the localization of fatigue cracks. The effectiveness of the proposed method is validated through numerical simulations of jacket-type offshore platforms subjected to random wave excitations from different directions. The study focuses on identifying fatigue crack damage in platform components, addressing both single and multiple damage scenarios involving small cracks. To assess the robustness of the method against noise, Gaussian white noise of varying intensities was added to the collected signals. The results demonstrate that the proposed approach effectively identifies and locates fatigue cracks in jacket-type offshore platforms, exhibiting strong noise resistance.