ABSTRACT: The objective of this research was to characterize the efficiency and convenience of sheep identification system and its implementation costs. Therefore we quantified the time for implementation, reading and data transfer to a management program in a manual system (earrings) and electronic system (subcutaneous transponders). Furthermore, it was evaluated the implementation costs of the traceability systems by analysis of the absorption costs and comparison between systems. It was observed that the implementation time for the manual system was shorter (2.2 s animal−1) than the electronic system (4.67 s animal−1). As for reading time, the electronic system presented shorter time (2.89 s animal−1) than the manual system (4.67 s animal−1). As for the data transfer time, the electronic system also presented shorter time (16 s) than manual system (6 min and 42 s). In terms of economic costs, the manual system presented lower implementation costs and lower values per tracked animal to ensure the economic viability. In conclusion, the producer must analyze the context around the production unit to choose the best traceability system.

ABSTRACT: The goat and sheep meat producer chain has developed in last years, thus, it is imperative to organize and structure the supply chain and to adopt reliable policies for products' traceability, as a tool to achieve these requirements. The study aimed to make a management program with a traceability model for goat and sheep meat production, with emphasis on ensuring product origin and management practices' transparency at the animal production unit. For this purpose, it was made a reference model in order to emit an origin certificate which, in turn, provides specific information concerning the final product from each unit. Secondly, the program was developed using Hipertext Preprocessor (PHP) technology and as for the Database Management System, it was used MySQL. The schematic model proposed meets the requirements of a traceability system for goat and sheep meat. Furthermore, the program can work as a tool for farm management, by reports and real-time remote access to information.

RESUMO Este trabalho foi conduzido com o objetivo de investigar a aplicabilidade das redes neurais artificiais (RNAs) para estimar a evapotranspiração do cultivo protegido de pimentão. Os dados utilizados compreenderam o segundo ciclo da cultura, cultivado de setembro de 2013 a fevereiro de 2014, constituindo 135 dias de dados meteorológicos diários: temperatura e umidade relativa do ar, velocidade do vento e radiação solar global, utilizadas como variáveis preditoras e evapotranspiração, determinada a partir de dados obtidos por lisímetro de pesagem, usada como variável resposta. Os dados registrados no período foram divididos em três conjuntos para treinamento, teste e validação. O processo de aprendizado da RNA foi estabelecido com precisão aceitável, com erro médio quadrático de 0,005, em que os padrões de evapotranspiração no ambiente protegido estimado pela rede, em comparação àqueles gerados a partir dos dados de lisimetria, apontaram coeficiente de determinação da ordem de 0,87 para uma arquitetura de rede 4-21-1, de múltiplas camadas, algoritmo de aprendizagem de retropropagação do erro e taxa de aprendizagem de 0,1.

ABSTRACT This study aimed to investigate the applicability of artificial neural networks (ANNs) in the prediction of evapotranspiration of sweet pepper cultivated in a greenhouse. The used data encompass the second crop cycle, from September 2013 to February 2014, constituting 135 days of daily meteorological data, referring to the following variables: temperature and relative air humidity, wind speed and solar radiation (input variables), as well as evapotranspiration (output variable), determined using data obtained by load-cell weighing lysimeter. The recorded data were divided into three sets for training, testing and validation. The ANN learning model recognized the evapotranspiration patterns with acceptable accuracy, with mean square error of 0.005, in comparison to the data recorded in the lysimeter, with coefficient of determination of 0.87, demonstrating the best approximation for the 4-21-1 network architecture, with multilayers, error back-propagation learning algorithm and learning rate of 0.01.