A review of improved machine learning-based reliability analysis methods applied to floating renewable energy structures

In recent years, the global wind energy industry has been growing rapidly, the utilisation of marine energy has been increasing, and the depth and number of applications for offshore renewable energy floating systems have been growing. However, this consideration also raises concerns about the structural safety of offshore renewable energy floating systems, making it crucial to enhance structural reliability through failure mode analysis, reliability modelling, and machine learning (ML)-based data processing and prediction. Using key structural parameters of the renewable energy floating system as input and integrating ML to enhance computational efficiency, an improved response surface model (RSM) can be developed to capture the nonlinear behaviour of the system based on traditional reliability analysis. Thus, this paper firstly reviewed the current state of research on ML-based reliability analysis, and then analysed the applicability of ML methods in the reliability analysis of ocean engineering structures; secondly, it introduced the principles and innovative improvement paths of the traditional RSM in reliability analysis, and discussed the existing challenges. Finally, based on the ML technique, an improved application framework of RSM is proposed in the assessment of the reliability of one of the renewable energy floating systems (offshore floating wind turbines) by introducing ML into the reliability analysis of RSM. This new reliability analysis method can handle complex and variable marine environmental loads and coupled power response problems, providing support for enhancing the robustness of system design and informed operational and maintenance decisions.