Wavelet decomposition and statistical characterization for unbalance detection in rotating systems

Abstract

Vibration analysis is a crucial tool for the early detection of faults in rotating machines, as it allows for the prevention of major damage and avoids significant costs associated with these faults. This study examines the phenomenon of imbalance in rotating machines, using signals generated on a test bench at the Santander Technological Units, where specific fault conditions were replicated. The signals obtained were analyzed using wavelet decomposition, from which key characteristics were extracted, such as root mean square (RMS), peak value, kurtosis, and mean absolute value (MAV). These characteristics were then compared using box plots to evaluate the separation between signals from unbalanced machines and those in a fault-free state. This analysis allowed us to identify significant differences between the two conditions, demonstrating the effectiveness of the approach in detecting faults due to imbalance.