Sparsity-constrained Ramanujan subspace feature tracking method for compound fault diagnosis.

Area of Science:

• Mechanical Engineering
• Signal Processing
• Machine Condition Monitoring

Background:

• Compound fault diagnosis faces challenges due to coupled multi-fault features and varying intensity characteristics, hindering accurate feature extraction.
• Existing methods struggle with the complex interplay of multiple fault signatures in rotating machinery.
• The need for robust and accurate methods for identifying simultaneous faults is critical for industrial applications.

Purpose of the Study:

• To propose a novel method for enhanced compound fault diagnosis.
• To address the limitations of insufficient feature extraction accuracy in complex fault scenarios.
• To improve the decomposition efficiency and separability of multi-fault features.

Main Methods:

• Development of a novel sparse Ramanujan sequence to establish a sparse subspace with inherent periodicity.
• Implementation of sparsity-constrained Ramanujan subspace feature tracking (SRSFT) for signal mapping.
• Integration of an enhanced period estimation technique for improved weak period detection and stability.

Main Results:

• The SRSFT method effectively maps signals into a domain where fault features are more separable.
• Demonstrated marked enhancement in decomposition efficiency and compound fault feature extraction.
• Achieved superior performance in multi-period impulse separation and extraction compared to existing methods.

Conclusions:

• The proposed SRSFT method offers an effective and robust solution for compound fault diagnosis.
• Leveraging periodicity and sparsity, SRSFT excels at separating and extracting compound fault features.
• The method shows significant promise for improving the reliability of machinery health monitoring systems.