AplicaÃÃo de Sistemas Fuzzy e Neuro-fuzzy para PrediÃÃo da Temperatura Retal de Frangos de Corte / Application of fuzzy and neuro-fuzzy systems for rectal temperature prediction of broiler chickens.

AUTOR(ES)
DATA DE PUBLICAÇÃO

2009

RESUMO

The aims of this research was to develop and validate a decision support system using fuzzy system, on the ground of fuzzy sets theory, and a neuro-fuzzy system (neuro-fuzzy network), based on the algorithm LOLIMOT (Local Linear Model Tree ) for rectal temperature prediction of broiler chickens. The systems were developed based on the ground of three input variables: air temperature (T), relative humidity (RH) and air velocity (V). The output variable was the rectal temperature (RT). The fuzzy inference system was performed through the Mamdaniâs method which consisted in elaborating 48 rules. The defuzzification was done by Center of Gravity Method. The fuzzy system was developed using MAPLEÂ 8. Gaussian pertinence functions type with 0.33 standard deviation were used in order to perform the neuro-fuzzy network. Ten input data partitions were adapted by the system. The neuro-fuzzy network was developed using SCILAB 4.1. Experimental results, used for validation, showed that the average standard deviations between the simulated and measured values of RT by the fuzzy and neuro-fuzzy systems were 0.13ÂC and 0.11ÂC, respectively. The proposed fuzzy and neuro-fuzzy systems were found to satisfactorily predict rectal temperature based on climatic variables (T, RH and V). Therefore, they could be used as decision support system on broiler chicken growth.

ASSUNTO(S)

modelagem de sistemas biologicos 1. lÃgica fuzzy. 2. redes neuro-fuzzy. 3. avicultura. 4. conforto tÃrmico

ACESSO AO ARTIGO

Documentos Relacionados