Gearbox Vibration Data for Fault Diagnosis Under Varying Loads

Department of Energy data from a gearbox fault simulator records vibration signals under load variations from 0 to 90 percent. The dataset includes measurements for both healthy gears and cracked gears, with spectral analysis highlighting sideband amplitudes useful for AI-based fault detection. Ten text files are provided for each condition.

Use Cases

• Train AI models to detect gear cracks using the amplitude of sidebands from vibration signal processing.
• Analyze the relationship between load variation (0-90%) and vibration sensor readings for condition monitoring.
• Compare modulation sidebands from healthy gear and cracked gear vibration measurements for fault diagnosis.

Strengths

• Data was recorded under controlled conditions with load varied from 0 to 90 percent.
• Includes comparative measurements for both a normal gear and a cracked gear condition.
• Vibration data was captured using four sensors placed in different directions.

Limitations

• The dataset size is unspecified, with no row count or column details provided.
• Data is limited to a single fault type (cracked gear) and a specific simulator setup, which may not generalize to other failures or real-world gearboxes.

Provenance

Source

Department of Energy

Collection Method

Recorded using SpectraQuest's Gearbox Fault Diagnostics Simulator with four vibration sensors.

Freshness

Last updated on 2025-10-09.