Modelo paraconsistente de predição de estresse em suínos

Resumo (EN)

The advance of swine farming in Brazil has occurred intensively in recent times due to advances in genetics, nutrition, health, reproduction, and management. In this respect, the animals have gone from extensive to intensive farming, where they remain confined until the slaughter. The environment of the intensive breeding system in pig breeding has a direct influence on the comfort and animal welfare conditions, promoting difficulty in maintaining the thermal balance inside the facilities and in air quality, affecting the productive and reproductive performance of the pigs.With the advancement of information technology, the use of expert systems to predict productive parameters becomes increasingly important in agricultural management and in the decision-making process. The bases for the development of this work were the revision of Cordeiro, A.F.Sdoctoral thesis and the use of the database created by the researcher. In this work the indicator was the level of stress in pigs, under the following conditions: hunger, thirst, pain, cold and normal. When inspecting the database, comprised of 72 piglets, 36 maternity (20 females and 16 males) and 36 nursery (20 females and 16 males), the ranges of values attributed to each stress situation presented a level of high intersection, consider a pig as being cold, pain, hunger, thirst or normal, from a single criterion, the temperature may present an inconsistent or contradictory result in order to mitigate this issue, it was defined as the first step to elaborate a method, to extract the contradiction of the obtained data. We considered 18 propositions, considering the minimum temperature, the maximum temperature and average body temperature, considering the following situations: pain, fear, hunger, thirst and normal (baseline), in farrowing and nursery, for females more males, only females and only males. For the process of extraction of contradiction was used ParaDecision-making Meeting Nott software, which applies in full the concepts of Paraconsistent Annotated Evidential Et. To validate the possibility of stress classification from the temperatures obtained after the extraction of the contradiction, the Fuzzy Logic was used. For the process to be implemented, it was necessary to standardize the resulting temperatures and the use of MatLab software, responsible for implementing the Mamdani method. As a result, it was possible to identify at a preliminary stage whether the animal is in a state of stress or not stress, but the degree of maximum pertinence was 0.39 on a scale between 0 and 1, where one represents the value crisp, situation without a doubt.