PREDICTION OF POULTRY DEEP BODY TEMPERATURES USING ARTIFICIAL NEURAL NETWORKS by

NEURAL NETWORKS by LIN LIU (Under the Direction of Takoi Hamrita) ABSTRACT To understand the relationships among ambient temperature (AT), relative humidity (RH) and broiler deep body temperature (DBT), controlled experiments were conducted for different RH (50 and 80%) and AT (31, 34 and 37C) combinations. The DBT measurements were collected by a radio biotelemetry system. Three types of Artificial Neural Network (ANN) models have been developed to predict broiler’s DBT. Type I models predict DBT responses of birds not used in training to AT×RH combinations used in training. Type II models predict DBT responses of birds used in training to AT×RH combinations not used in training. Type III models predict DBT response of birds not used in training to AT×RH combinations not used in training. These models capture the complex relationship among DBT, AT and RH very well. They could be applied in the future in the development of environmental control system for poultry housing.