Efficient Vibration Signal Denoising Techniques for Effective Condition Monitoring of Gearbox

Vibration signals are invaluable in diagnosing the health of mechanical systems, particularly in the context of gear machinery. However, these signals are often contaminated with noise, posing challenges in extracting meaningful information. This paper presents a concise overview of gear vibration signal denoising methods and highlights their significance in machinery condition monitoring and maintenance. Of all the existing methods available, transform-based denoising is widely used in industry due to its capability for multi-resolution analysis and ability to handle non-stationary vibration data quite well. In this paper, gearbox vibration signal is collected for healthy and faulty gear conditions (broken tooth). It is then denoised using Wavelet, Minimum Entropy Deconvolution (MED), Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and Maximum Correlated Kurtosis Deconvolution (MCKD). The Signal to Noise Ratio (SNR) along with Kurtosis is being used as a parameter to evaluate the denoised signal. MED surpassed all other denoising methods, demonstrating superior SNR improvement and concurrently retaining signal information.