Acoustic detection methods offer a non-destructive alternative to manual inspection for identifying insect infestations in stored products, but their performance is compromised by ambient noise in operational environments. This study presents an enhanced detection algorithm for the Acoustic Stored Product Insect Detection System (A-SPIDS) that enables reliable single-insect detection in the presence of strong external noise. The platform’s physical noise isolation achieved an average attenuation of 45 dB above 2000 Hz. Spectral analysis revealed that insect signals dominate over ambient noise, generating insect-like impulses in the high-frequency band, enabling optimization of the Normalized Signal Pulse Amplitude (NSPA) detection metric to the 1565 Hz–6000 Hz frequency band, resulting in 99.4% detection accuracy at 80 dBA ambient noise levels. The external microphone was leveraged to identify and remove noise-generated impulses from internal piezoelectric sensor recordings, achieving 100% detection with zero false alarms across the recorded dataset featuring species Callosobruchus maculatus, Tribolium confusum, and Tenebrio molitor in oatmeal, rice, wheat, and corn products at noise levels exceeding 100 dBA.
Kadyrov et al. (Fri,) studied this question.