Multi-Domain Health Indicator Construction for Wind Turbine Gearboxes Situational Awareness under Generation-Grid-Load-Storage Integration

In modern power systems with tightly integrated generation, grid, load, and storage, wind turbine gearboxes endure severe transient shocks from frequency regulation steps to emergency braking, that far exceed steady state loads and accelerate fatigue. Conventional diagnostics, which rely on single signal or quasi-periodic load models, cannot capture high frequency torque reversals or nonlinear coupling among electrical, mechanical and oil temperature signals under strong noise. To overcome these limitations, this paper proposes a dual pronged framework for wind turbine gearbox situational awareness in generation-grid-load-storage coordinated operation. Firstly, a multi-parameter dynamic model that quantifies interactions among grid dispatch commands, load fluctuations, energy storage dynamics, mechanical stresses and environmental factors is developed to provide a robust theoretical basis for fault evolution analysis. Secondly, an improved spatio-temporal difference graph convolutional network (STDGCN) with embedded physical constraints is introduced to fuse heterogeneous signals and derive multi-domain health indicator (HI). Finally, a bidirectional long short-term memory (BiLSTM) enhanced with an attention mechanism employs HI to deliver situational awareness and remaining useful life (RUL) estimation. Experimental results validate that the proposed approach significantly outperforms traditional approaches in fault diagnosis and prognosis under complex coordinated operation conditions.