Monitoramento estatístico e predição de mudanças comportamentais em bovinos

• Ano de defesa: 2021
• Autor(a) principal: Martins, Thiago Guedes da Luz
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Maria; Brasil; Engenharia de Produção; UFSM; Programa de Pós-Graduação em Engenharia de Produção; Centro de Tecnologia
• Palavras-chave em Português: Carta controle; ΔP2; Vaca de leite; Detecção de cio; F1-score; Kappa; Control charts; Milk cow; Heat detection; CNPQ::ENGENHARIAS::ENGENHARIA DE PRODUCAO
• Link de acesso: http://repositorio.ufsm.br/handle/1/23413

The Maintaining reproductive efficiency is a major challenge in dairy production, since overall productivity and profit are directly dependent on estrus detection. Different aids have been developed throughout the years to identify estrus behavior in dairy cows. The objective of the present work is to develop a statistical method for the prediction of such behavior. The proposed method is called “carta 2”, which is able to behavioral changes related to estrus in dairy cows using the temporal series of an activity index generated by the cowmed system of animal monitoring. The proposed method was compared against the Shewhart, EWMA and CUSUM control charts. For this, records of artificial inseminations which result in pregnancies in animals monitored by the cowmed system were used. Numerical results show that the “cartaP2” had and overall superior performance that the other 2 “cartas”. It reached the F1 index, a score of 0.783, indicating an balance between sensibility and precision, and a Kappa index of 0.773 which indicates as “very good” the performance of the proposed method.