A comprehensive wind turbine digital model approach for drive-train broken tooth-type fault detection

Abstract

This paper presents a modular Digital Model of a wind turbine drive-train explicitly oriented toward the physical analysis and detection of broken tooth faults in the third gearbox stage. The model integrates aerodynamic, drive-train, generator, and control subsystems under a set of deliberate simplifying assumptions that define a controlled and physically interpretable reference framework. To isolate fault-related mechanisms, the first two gearbox stages are aggregated, while the third stage is modeled in detail using a lumped-parameter formulation with time-varying mesh stiffness and viscous damping. Tooth breakage is represented as a periodic reduction in effective contact width through a square-wave approximation. Under turbulent wind conditions and partial-load operation, simulations are conducted for healthy and progressively degraded scenarios. The results show that frequency-domain indicators, including gear-mesh sideband amplitudes and global spectral energy, exhibit clear and monotonic sensitivity to fault severity. Complementary time-domain statistical features, particularly RMS, variance, skewness, and rectified mean value, demonstrate strong monotonicity, trendability, and prognostic ability once damage reaches moderate levels. Overall, the proposed Digital Model functions as a baseline platform for mechanistically interpretable drive-train fault analysis and supports systematic experimental ablations. Its modular structure enables future extensions toward increased model complexity, experimental validation, and the development of data-synchronized Digital Twins for condition monitoring and Remaining Useful Life estimation.