Statistical and fuzzy modeling for accurate prediction of feed intake and surface temperature of laying hens subjected to light challenges

• Published in: Computers and electronics in agriculture Vol. 211; p. 108050
• Main Authors: Bahuti, Marcelo, Yanagi Junior, Tadayuki, Lima, Renato Ribeiro de, Fassani, Édison José, Ribeiro, Bruna Pontara Vilas Boas, Campos, Alessandro Torres, Abreu, Lucas Henrique Pedrozo
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.08.2023
• Subjects: Fuzzy inference systems; Laying poultry; Light intensity; Mathematical modeling; Support smart production; Zootechnical responses predicting; Light intensity; Mathematical modeling; Support smart production; Laying poultry; Zootechnical responses predicting; Fuzzy inference systems
• ISSN: 0168-1699; 1872-7107
• DOI: 10.1016/j.compag.2023.108050

•White LED light intensities were evaluated for rearing modern laying hens.•Feed intake and surface temperature of laying hens are dependent on the interaction between the level of illuminance and the exposure time.•FIS models tended to perform better than statistical models.•The prediction models developed can help in making decisions about bird management. The use of artificial lighting for laying hens in poultry houses allows production optimization. However, the effect of recent lighting technologies needs to be investigated, as their use can affect the welfare of hens, causing biological rhythm, behavior, health, productive and reproductive performance and egg quality changes. Therefore, the objective of this study was to evaluate the physiological responses, productive performance and egg quality of hens under different light intensities. If a significant difference between treatments was found, statistical models and fuzzy inference systems (FISs) to predict such responses were developed and compared. Thus, 76 Hy-Line W-80 laying hens were housed in a controlled environment with temperature, relative humidity and air speed adjusted for bird comfort. The hens were exposed to illuminances of 5, 20, 50 and 100 lx for a period of 21 days. Among the evaluated response variables, only the mean feed intake (FI) and surface temperature (tsurf) of the hens had a significant interaction and interaction sliced by main factors (p < 0.05), allowing statistical model adjustment. For these two variables, FISs were also developed so that the illuminance and the time of exposure to illumination (tillu) were defined as input variables of the systems. As a result, the statistical models and FISs presented significant high-accuracy indexes when predicting the FI and the tsurf (except for R2 values related to the statistical models), yet with an advantage for the FISs models, in which greater generalization capabilities were obtained in the validation step. Therefore, the developed models can be used to support decisions related to the management of hens, supporting the smart production, with emphasis on FISs based on expert knowledge, since they can be used to obtain predicted responses in different scenarios in addition to those evaluated experimentally with high accuracy.