Vibration Signal Denoising Method for Wind Turbine Blades Based on Adaptive Feature Modal Decomposition

• 1、School of Intelligent Engineering and Automation, Beijing University of Posts and Telecommunications,Beijing

Abstract：To address the issue of low diagnostic accuracy caused by the insufficient denoising capability of the Feature Modal Decomposition (FMD) algorithm in wind turbine blade vibration signals, this paper proposes an Adaptive Feature Modal Decomposition (AFMD) method. The proposed method incorporates a dynamic inertia weight and an elite retention strategy to improve the Particle Swarm Optimization (PSO) algorithm, enabling a global optimal search for decomposition parameters. Furthermore, an index for the periodic envelope signal-to-noise ratio (SNR), attuned to the periodic nature of pulses, has been developed to dynamically modify filter coefficients, thus precisely deriving intrinsic mode functions (IMFs). Furthermore, an IMF screening criterion is constructed by integrating energy distribution, periodic features, and entropy information theory to effectively separate noise from feature components. Experimental results demonstrate that, compared to high-pass filtering, Empirical Mode Decomposition (EMD), and FMD, the AFMD method achieves superior performance in filtering metrics such as SNR, waveform similarity coefficient, and mean square error.

Keywords： Wind turbine blades Fault diagnosis Vibration detection Feature mode decomposition