Intelligent fault diagnosis for offshore wind turbine gearbox: A hybrid framework integrating enhanced VMD and GRU

• Introducing real-time step size and improved formula to improve FOA. • IFOA is proposed to optimize VMD parameters, forming the e-VMD. • e-VMD is adopted to resolve mode aliasing during signal preprocessing. • Eliminate redundant IMFs by correlation coefficients and multiscale entropy features. • Implement GRU to address vanishing/exploding gradients and structural redundancy. Regarding the deficiencies of existing fault diagnosis strategies for wind turbine gearbox (WTG) bearings, such as mode aliasing, redundant decomposition of fault information, difficult feature extraction, and inaccurate fault identification, this paper presents an enhanced variational mode decomposition (e-VMD) and gated recurrent unit (GRU) strategy. The key concepts are as follows: Firstly, VMD is employed to address mode aliasing during signal preprocessing. An improved fruit fly optimization algorithm (IFOA) is proposed to optimize the VMD parameters, thus forming the e-VMD. Secondly, a novel screening approach that combines correlation coefficients and multi-scale entropy features is introduced to eliminate redundant intrinsic mode functions (IMFs), ensuring the significance and richness of fault features. Finally, a GRU network is utilized to overcome the problems of vanishing/exploding gradients and structural redundancy in traditional networks, thereby enhancing the identification accuracy. Validation through simulated data and practical case studies shows that the proposed strategy is more effective and has greater engineering applicability compared to conventional methods.