Data Decomposition Techniques for Signal Analysis

Data decomposition techniques partition complex signals into simpler components according to mathematical or statistical criteria. Approaches such as Empirical Mode Decomposition (EMD) and its enhancements (EEMD, CEEMDAN) exploit adaptive basis functions to isolate intrinsic mode functions, enabling effective separation of noise and oscillatory modes in non-stationary, non-linear time series. Variational Mode Decomposition (VMD) employs an optimisation framework to extract band-limited modes, improving mode separation and reducing end effects. Complementary methods—Principal Component Analysis (PCA), Independent Component Analysis (ICA) and singular value decomposition—offer linear projections to capture dominant patterns and remove artefacts. Wavelet-based transforms provide multi-resolution analysis, allowing denoising and feature extraction across scales. These techniques are widely applied in biomedical signal enhancement, mechanical fault diagnosis, power-system monitoring and communications. Advances focus on reducing mode mixing, algorithmic stability and computational efficiency, while promoting real-time implementation and integrative frameworks combining statistical selection metrics and optimisation algorithms.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Research from all publishers

Recent work on rolling bearing diagnostics has illustrated the efficacy of combining piecewise aggregate approximation with complete ensemble empirical mode decomposition. In this approach, vibration envelopes are compressed via adaptive aggregation, enabling efficient CEEMDAN analysis of long-duration signals. Subsequent selection of diagnostic modes yields enhanced fault detection accuracy and reduced computational load. Another study has advanced variational mode decomposition by integrating a genetic algorithm to optimise wavelet threshold parameters. Fault vibration signals are first decomposed into band-limited intrinsic mode functions by VMD, then effective components are identified through statistical and correlation metrics. An adaptive wavelet threshold, tuned by genetic search, ensures continuity and minimises reconstruction error, leading to robust detection of bearing faults under varied operating speeds. A further application has targeted partial discharge detection in power cables using CEEMDAN coupled with wavelet packet denoising. Adaptive noise-assisted decomposition isolates discharge pulses, while cross-component correlation and packet-based thresholding suppress both narrowband and white noise, achieving reliable extraction of weak pulses even in low signal-to-noise scenarios.

Data Decomposition Techniques for Signal Analysis publication trend

The graph below shows the total number of articles in data decomposition techniques for signal analysis across all publications each year (not limited to Nature Index journals).

Technical terms

Empirical Mode Decomposition (EMD): Data-driven method that decomposes a signal into oscillatory components known as intrinsic mode functions.

Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN): Enhanced EMD variant adding adaptive noise to reduce mode mixing and improve decomposition stability.

Variational Mode Decomposition (VMD): Optimisation-based technique that decomposes a signal into a predefined number of band-limited modes via iterative spectral analysis.

Wavelet Transform: Multi-resolution analysis tool that represents signals with basis functions localized in time and frequency for denoising and feature extraction.

Intrinsic Mode Function (IMF): Component extracted by EMD or its variants, characterised by symmetric envelopes and a zero-mean instantaneous frequency.

References

1. Bearing Fault Diagnosis Using Piecewise Aggregate Approximation and Complete Ensemble Empirical Mode Decomposition with Adaptive Noise. Sensors (2022).
2. Fault Diagnosis of Rolling Bearings Based on Variational Mode Decomposition and Genetic Algorithm-Optimized Wavelet Threshold Denoising. Machines (2022).
3. Extraction of Partial Discharge Pulses from the Complex Noisy Signals of Power Cables Based on CEEMDAN and Wavelet Packet. Energies (2019).