Run-to-failure vibration dataset of a three-stages planetary gearbox - PART 1

Data sourced from real operating conditions constitutes a crucial asset for research in machine diagnostics and prognostics, particularly in the context of predictive maintenance and the increasing adoption of artificial intelligence (AI) methods. For fault detection and prognostic applications, datasets must capture the evolution of degradation phenomena over time, enabling both the assessment of machinery health and the prediction of its remaining useful life (RUL). These objectives require long-term, high-quality run-to-failure data collected under realistic and variable working conditions.

Planetary gearboxes are widely employed in industrial and automotive systems and are subjected to complex loading and dynamic interactions among gears.

The Department of Engineering at the University of Ferrara conducted an extensive experimental campaign aimed at documenting the temporal evolution of vibration signals over the entire operational life of a three-stage planetary gearbox. The experiment was performed on a dedicated back-to-back test bench operating under variable speed and torque conditions. A mono-axial accelerometer was mounted on the gearbox housing to acquire radial vibration signals at hourly intervals. The system was operated continuously until the third-stage sun gear experienced complete mechanical failure after 1006 operating hours.

The resulting dataset provides the complete run-to-failure history of the gearbox and offers a comprehensive record of its degradation progression. This dataset represents a valuable resource for both academic research and industrial applications in the fields of diagnostics, prognostics, and predictive maintenance.
Data is provided in .mat format and, due to repository size limitations, is divided into three separate parts, each containing approximately one-third of the total experiment. Each file contains the radial acceleration signal, the sampling frequency, and the operating condition corresponding to the acquisition.

This dataset provides the Part 1 of data collected during the experimental campaign.