The absence of the queen in a beehive is a very strong indicator of the need for beekeeper intervention. Manually searching for the queen is an arduous recurrent task for beekeepers that disrupts the normal life cycle of the beehive and can be a source of stress for bees. Sound is an indicator for signalling different states of the beehive, including the absence of the queen bee. In this work, we apply machine learning methods to automatically recognise different states in a beehive using audio as input. We investigate both support vector machines and convolutional neural networks for beehive state recognition, using audio data of beehives collected from the NU-Hive project. Results indicate the potential of machine learning methods as well as the challenges of generalizing the system to new hives.

The widespread decline of honey bee (Apis mellifera L.) colonies registered in recent years has raised great attention to the need of gathering deeper knowledge about this phenomenon, by observing the colonies’ activity to identify possible causes, and design corresponding countermeasures. In fact, honey bees have well-known positive effects on both the environment and human life, and their preservation becomes critical not only for ecological reasons, but also for the social and economic development of rural communities. Smart sensor systems are being developed for real-time and long-term measurement of relevant parameters related to beehive conditions, such as the hive weight, sounds emitted by the bees, temperature, humidity, and CO 2 inside the beehive, as well as weather conditions outside. This paper presents a multisensor platform designed to measure the aforementioned parameters from beehives deployed in the field, and shows how the fusion of different sensor measurements may provide insights on the status of the colony, its interaction with the surrounding environment, and the influence of climatic conditions.

No abstract

No abstract

The multiple-variance method is a cross-correlation method that exploits input signals with different powers for the identification of a nonlinear system by means of the Volterra series. It overcomes the problem of the locality of the solution of traditional nonlinear identification methods, based on mean square error minimization or cross-correlation, that well approximate the system only for inputs that have approximately the same power of the identification signal. The multiple-variance method permits to improve the performance of models of systems that have inputs with high dynamic, like audio amplifiers. This method is used, for the first time, to identify three different tube amplifiers. The method is applied to a novel reduced Volterra model that allows to overcome the problem of the very large number of coefficients required by the Volterra series by choosing only a proper subset of elements from each kernel. Eventually, the multiple-variance methodology is applied to different real audio tube devices demonstrating the effectiveness of the proposed approach in terms of system identification and computational complexity.