Vibration-Based Fault Diagnosis in Rotating Machinery

Rotating machinery underpins industries from power generation to manufacturing, yet its components are subject to wear, misalignment and imbalance. Vibration-based fault diagnosis exploits mechanical oscillations captured via accelerometers or similar transducers to detect incipient defects before catastrophic failure. The approach comprises four stages: signal acquisition, signal processing, feature extraction and fault classification. Traditional time-domain measures such as rms and kurtosis reveal impulsive events but struggle in noisy environments. Frequency-domain methods including fast Fourier transform highlight steady-state resonances but lack temporal resolution for transient faults. Time–frequency techniques such as wavelet packet transform and improved kurtograms offer enhanced localisation of fault-related impulses. Recent trends integrate data-driven algorithms and machine learning classifiers to automate interpretation and improve sensitivity under variable loads. Advances in narrowband demodulation and envelope analysis have refined the identification of resonant frequency bands where bearing or gear faults manifest. Combined with robust statistical indices and adaptive filters, these developments enable real-time monitoring and predictive maintenance. The global drive for Industry 4.0 and digital twins further stimulates the convergence of intelligent signal processing, enabling more accurate, early and cost-effective diagnosis of rotating machinery defects.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Vibration-Based Fault Diagnosis in Rotating Machinery publication trend

The graph below shows the total number of articles in vibration-based fault diagnosis in rotating machinery across all publications each year (not limited to Nature Index journals).

Technical terms

Envelope analysis: Demodulation technique that extracts amplitude variations from narrow-band signals to reveal repetitive impact features.

Kurtosis: Statistical measure of signal peakedness used to detect transient impulses associated with faults.

Wavelet packet transform: Signal decomposition method that provides adaptive time–frequency representation for isolating fault-related components.

Cyclostationarity: Property of signals whose statistical characteristics vary periodically, often exploited to detect rotating machinery faults.

References

1. A New Improved Kurtogram and Its Application to Bearing Fault Diagnosis. Shock and Vibration (2015).
2. Optimal Resonant Band Demodulation Based on an Improved Correlated Kurtosis and Its Application in Bearing Fault Diagnosis. Sensors (2017).
3. IGIgram: An Improved Gini Index-Based Envelope Analysis for Rolling Bearing Fault Diagnosis. Journal of Dynamics Monitoring and Diagnostics (2022).
4. Vibration Analysis of Shaft Misalignment Using Machine Learning Approach under Variable Load Conditions. Shock and Vibration (2020).
5. Teager Energy Entropy Ratio of Wavelet Packet Transform and Its Application in Bearing Fault Diagnosis. Entropy (2018).