Abstract
Condition monitoring (CM) of wind turbine gearbox is one of the key concerns for the reliable operation of wind power generation. With the huge ongoing transition towards renewable energies globally, necessary studies are needed to ameliorate this problem in the wind energy industry. Recent developments in CM of wind turbine gearbox towards improving the system lifecycle have necessitated in-silico modelling, which is cost-efficient and saves time. These developments involve applying vibration analysis through gear fault modelling and dynamic simulations coupled with signal processing or machine learning methods for an efficacious CM system. The finite element method (FEM) is an efficient tool that has been utilized in different studies for structural analysis and failure investigations of this unreliable system. Also, machine learning techniques like the artificial neural networks (ANN) model has been utilized for intelligent fault prediction and classification. This study, therefore, presents the state of research in the use of these computational tools in the CM of wind turbine gearbox. The level of correctness of diverse FEM analyses and different ANN variants utilized in literature were recorded. From our findings, the FEM approach produced accurate results with very minimal deviation from traditional methods of experiments. The ANN approach, which can perform feature extraction and feature learning directly from raw data eliminating the need for conventional extraction methods, also offered high prediction and classification accuracies. Also, hybrid ANN models were more effective in fault diagnosis than the standalone models. Finally, based on their notable accuracies, future research direction synergizing both approaches for intelligent fault diagnosis was presented for improved monitoring and diagnosis.