A Dual-Stage Multi-Reference Spectral-Subtraction Compensation Denoising Algorithm for Measuring Explosion...

Explosion electromagnetic radiation (EEMR) is important for studying explosive detonation, but environmental interference and system frequency-response characteristics influence its electric-field-strength characteristics. This study analyzes EEMR and proposes a dual-stage multi-reference spectral-subtraction compensation denoising (DMSCD) algorithm for field-strength measurement. DMSCD compares signal and pre-sampling noise spectra to identify and attenuate noise-dominant frequency components, and then applies Wiener-gain suppression to overlapping residual noise. To evaluate DMSCD, six baseline algorithms are selected. For 90 EEMR segments from 10 g Φ 20 mm × 20 mm RDX charges, DMSCD reduced the average peak-to-peak error to 74%, lower than that without denoising. Its reconstructed spectra retained clear signal peaks near 40, 126, 221, 270, 388, and 675 MHz, with 160-337 MHz as the main frequency component. Compared with the baseline algorithms, DMSCD more effectively suppressed noise in the frequency domain while preserving time-domain pulse characteristics. In the processing of EEMR, DMSCD can effectively utilize the signal characteristics of the pre-sampling segment to identify and suppress the influence of broadband noise, achieving good denoising performance in the time-frequency domain.