Muscle Fatigue Analysis With Optimized Complementary Ensemble Empirical Mode Decomposition and Multi-Scale Envelope...

Area of Science:

• Biomedical Engineering
• Signal Processing
• Computational Neuroscience

Background:

• Surface electromyography (sEMG) signal preprocessing is crucial for accurate feature extraction.
• Complementary ensemble empirical mode decomposition (CEEMD) is effective but suffers from mode mixing in complex signals.
• Existing methods like CEEMDAN have limitations in handling intermittent and near-spectrum components.

Purpose of the Study:

• To address the mode-mixing problem in CEEMD for sEMG signal decomposition.
• To introduce an optimized CEEMD (OCEEMD) method integrating least-squares mutual information (LSMI) and chaotic quantum particle swarm optimization (CQPSO).
• To develop a novel feature extraction method, multi-scale envelope spectral entropy (MSESEn), for muscle fatigue analysis.

Main Methods:

• Developed OCEEMD by incorporating LSMI to reduce IMF correlation and CQPSO to optimize noise parameters for efficient decomposition.
• Applied LSMI to calculate IMF correlation, reducing mode mixing.
• Utilized CQPSO to optimize Gaussian white noise standard deviation, enhancing iteration efficiency.
• Selected and reconstructed useful IMFs to obtain a de-noised signal.
• Extracted MSESEn from the reconstructed sEMG signal, leveraging frequency and envelope information.

Main Results:

• OCEEMD effectively suppressed mode mixing between intrinsic mode functions (IMFs) compared to CEEMD and CEEMDAN.
• OCEEMD demonstrated rapid iteration efficiency.
• The extracted MSESEn showed a clear declining tendency over time.
• MSESEn proved sensitive to muscle fatigue, outperforming approximate entropy (ApEn) and sample entropy (SampEn).

Conclusions:

• OCEEMD offers a robust solution for sEMG signal preprocessing, mitigating mode mixing and improving efficiency.
• MSESEn is a promising new feature for analyzing muscle fatigue from preprocessed sEMG signals.
• The proposed approach holds potential for advancing sEMG-based diagnostics and performance monitoring.