Buzz-based honeybee colony fingerprint

Abstract

Non-intrusive remote monitoring has its applications in a variety of areas. For industrial surveillance case, devices are capable of detecting anomalies that may threaten machine operation. Similarly, agricultural monitoring devices are used to supervise livestock or provide higher yields. Modern IoT devices are often coupled with Machine Learning models, which provide valuable insights into device operation. However, the data preparation step for ML models has to be addressed differently for industrial and agriculture cases. Animals are characterized by their circadian rhythms and seasonal dependence, which can bias the accuracy of classifiers. In the presented work, a Design-of-Experiment (DoA) approach for extracting valuable bee colony audio data is described. With the presented methods, it is possible to precisely define the most distinctive bee hours where unique colony sounds are emitted. The first step of the data filtering process is based on identifying the ambient temperatures that are conducive to its operation. The second step provides the unique hours specification based on the hives’ characteristics comparison where dissimilar time periods are being marked. For this comparison, the most noticeable difference between the colonies is calculated with MSE integral and thus the trend’s joint component is removed. A new concept of a bees’ fingerprint was introduced for the identification of the particular bee colony.