Morphological Signal Processing for Fault Diagnosis in Rotating Machinery

Morphological signal processing applies the principles of mathematical morphology to the analysis of vibration and acoustic signals generated by rotating machinery. By employing basic operators such as erosion and dilation, often in combination as opening, closing and top-hat transforms, this approach extracts transient impulses associated with bearing, gear or shaft defects. Selection of an appropriate structuring element determines sensitivity to fault features, while multiscale implementations probe signals at different resolutions to capture both weak and strong impulses. Recent advances integrate morphology with optimisation algorithms (for example particle swarm optimisation or probabilistic principal component analysis) and feature-selection methods (such as entropy measures or correlation analyses) to automate the choice of scale and enhance diagnostic accuracy. Compared with purely frequency-domain or time–frequency techniques, morphological filters offer robust noise suppression, low computational overhead and suitability for non-stationary signals. They have been adopted internationally for early fault detection in wind turbines, railway vehicles, aerospace engines and general industrial plant, thereby reducing maintenance costs and avoiding catastrophic failures.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Morphological Signal Processing for Fault Diagnosis in Rotating Machinery publication trend

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

Technical terms

Mathematical morphology: A signal-processing framework based on set-theoretic operations, using erosion and dilation to probe and transform signal shapes.

Structuring element: A predefined geometric kernel that slides across a signal to determine how morphological operators extract or suppress features of a particular scale and form.

Top-hat filter: A morphological transform formed by subtracting the opened (or closed) signal from the original, thereby isolating small, impulsive features.

Multiscale analysis: An approach that applies morphological operations at multiple scales to capture both coarse and fine features of a signal, enhancing sensitivity to faults of varying severity.

References

1. A Survey on Fault Diagnosis of Rolling Bearings. Algorithms (2022).
2. Bearing Fault Feature Extraction Method Based on Enhanced Differential Product Weighted Morphological Filtering. Sensors (2022).
3. Bearing Fault Signal Analysis Based on an Adaptive Multiscale Combined Morphological Filter. International Journal of Rotating Machinery (2020).
4. A Morphological Filtering Method Based on Particle Swarm Optimization for Railway Vehicle Bearing Fault Diagnosis. Shock and Vibration (2019).
5. Probabilistic Principal Component Analysis Assisted New Optimal Scale Morphological Top-Hat Filter for the Fault Diagnosis of Rolling Bearing. IEEE Access (2020).