Исследовательская база

Спектральные компоненты звуковых сигналов медоносных пчёл, записанные внутри и снаружи улья

This study investigates the impact of microphone placement on honey bee audio monitoring for time-of-day classification, a key step toward automated activity monitoring and anomaly detection. A custom apparatus enabled simultaneous audio acquisition from internal brood-frame and external hive locations. Sound signals were represented using power spectral density, and Extra Trees and convolutional neural network classifiers were trained to identify diurnal activity patterns. Extra Trees achieved 98-99% accuracy with internal recordings but only 61-72% with external recordings, while CNNs trained on external data achieved 76-87%. Feature selection identified the 100-600 Hz band as especially informative, with relevant components remaining below 2 kHz. The results support internal microphone placement and suggest that low sampling rates and comparatively lightweight models can be sufficient for baseline acoustic hive monitoring.