Harmonic analysis method based on multi-stage frequency correction

Area of Science:

• Electrical Engineering
• Signal Processing
• Measurement Science

Background:

• Noncoherent sampling in digital storage oscilloscopes causes spectral leakage and picket-fence effects.
• These effects lead to biased harmonic parameter estimates and obscure weak signal components.

Purpose of the Study:

• To develop a robust multi-stage frequency-correction framework for accurate harmonic and interharmonic analysis.
• To enhance the reliability of oscilloscope-resident measurements under noncoherent sampling conditions.

Main Methods:

• A multi-stage framework involving spectral domain screening, fitness-guided frequency refinement, least-squares estimation, and iterative reconstruct-subtract procedures.
• Heterogeneous CPU-GPU implementation for parallel processing of interval extraction and fitness evaluations.

Main Results:

• Achieved mHz-level maximum absolute frequency error, a >37x improvement over windowed FFT baseline.
• Attained accuracy comparable to MUSIC without sensitive parameter selection.
• Reduced weak-component amplitude error from tens of percent to 0.2% in challenging scenarios.

Conclusions:

• The proposed framework offers superior frequency accuracy and robustness for harmonic/interharmonic analysis compared to existing methods.
• Enables reliable oscilloscope-resident measurements with a tunable accuracy-cost trade-off.