Autonomic Computing in a Beef-Production Process for Precision Livestock Farming

Abstract

Precision livestock farming (PLF) offers farmers real-time monitoring and management system. PLF provides a real-time warning when something goes wrong so that the farmer can take immediate action to solve the problem. PLF introduces many new challenges and questions that must be resolved. Some of these challenges are related to the integration of grazing and animal health into the beef-production process. This article introduces an architecture for the self-managing of a beef-production farm. In particular, the architecture includes three autonomous cycles of data analysis tasks (ACODAT) that allow beef producers to have adequate coordination, optimization and planning of the productive process, which are: (i) circuit preparation, (ii) animal purchase, and (iii) animal fattening. This article also instantiates, in a farm, the autonomous animal-fattening cycle, as the first step towards efficient and effective beef-production processes. The main contributions of this architecture are (i) the ability to use everything mining to improve the knowledge of the system and decision-making processes, and (ii) three ACODAT for real-time analysis for sustainable and environmentally-friendly livestock production. The results are encouraging since the ACODAT allows smart management of the beef-production process, naturally introducing artificial-intelligence techniques to develop these tasks. Particularly, modeling using ACODAT allows an adequate description of a precision livestock process. Likewise, the preliminary results of some of the tasks of ACODAT are stimulating because they allow evaluating the feasibility of the proposal. For example, a first task for the identification of cattle fattening has a Mean Absolute Error (MAE) of 5.4 kg, which will be used by ACODAT to identify anomalies in the fattening process. The instantiation of the animal-fattening cycle shows the viability and robustness of this proposal.