Application Of Infrared Thermography For Assessing Beef Cattle Comfort Using A Fuzzy Logic Classifier
The thermal comfort index in animal housing is determined typically based on ambient variables such as temperature and humidity. Furthermore, the physiological measures of thermoregulation are important indicators of thermal comfort, but this measurement is usually laborious. This work aims to develop a fuzzy logic classifier (FCIR) that integrates both environmental and animal factors to determine the level of thermal comfort to allow the environmental assessment and control. The FCIR was constructed with the inputs air temperature (DBT), relative humidity (RH) and skin temperature (IR) obtained by using infrared thermography. The output of the FCIR is a variable associated with the rectal temperature (RT) and heuristic rules were generated to associate the linguistic input variables with the physiological variable RT. An experiment was performed with eight Nellore steers during eight days where DBT, RH, IR and RT were taken in different periods. The measurements were applied by the specialist to guide the construction of the knowledge base. The FCIR is evaluated in comparison with a traditional temperature-humidity index (THI). The thermal environmental is classified in four levels (comfort, critical, danger and emergency) by THI, while FCIR classify the environment in three levels (comfort, critical, danger). The classification by THI and FCIR are coincident for 28,5% of the data and THI classify the days as emergency while the same days are classified as danger or critical by FCIR. Indeed, the RT measurements show no values that could infer a state of emergency of thermal stress. Common value of RT observed in the same day reinforces this result. Besides, the average of RT value during the days shows greater consistency between the classifier and animal’s response.
The thermal comfort index in animal housing is determined typically based on ambient variables, such as temperature and humidity, and it does not regard the animal physiological factors. On other hands, the physiological measures of thermoregulation, such as rectal temperature, are important indicators of the environment suitability and thermal comfort of animals, but these intrusive measures are laborious and impractical to be performed. This scientific work aims to developed a fuzzy logic classifier that integrate both environmental and animal factors, such as skin temperature, in order to determine the level of thermal comfort to allow the environmental assessment and control for automation system for animal housing. The fuzzy logic theory provides a formal methodology which can be applied to transfer the human experiences to automatic classifiers, thus the fuzzy classifier (FCIR) was constructed having as crisp inputs the air temperature (DBT), the relative humidity (RH) and the skin temperature (IR) obtained by using infrared thermography camera. The output of the fuzzy classifier is a value associated with the rectal temperature (RT), thus the classifier uses two environmental variables and a physiological variable of the animal to estimate the level of comfort associated with the RT of the animals (another physiological variable). Four linguistic variables were defined related of DBT, RH, IR and RT and heuristic rules were generate to associate the environment variables DBT and RH and the physiological variable IR with the physiological variable RT (fuzzy knowledge base). The fuzzy rules are used to compose sets of simple and intuitive conditional statements based on specialist knowledge. To support the research, an experiment was performed with eight Nellore steers during eight days where DBT, RH, IR and RT were taken at four periods in each day. The stored measurements were applied by the zootechnist specialist in other to guide the construction of the fuzzy knowledge base and to validate the fuzzy classifier. The fuzzy classifier is evaluated in comparison with the estimated comfort levels based on the rectal temperature and with some traditional temperature-humidity indexes (THI). The thermal environmental is classified in four levels (comfort, critical, danger and emergency) by THI, while FCIR classify the environment only in the first three levels (comfort, critical, danger). The classification by THI and FCIR are coincident for 28,5% of the data whereas 71,5% are classified as divergent way, i.e., THI classify the days (28,13%) as emergency while the same days are classified as danger or critical by FCIR. In fact, the physiological responses RT show no value that could infer a state of emergency of thermal stress by bovines. Common value of RT (38,8oC) observed for the animals in the same day when the environment is classified as emergency by THI reinforces this result. On the other hand the average of RT value for animals during the days classified as danger by FCIR is 38,9oC, showing greater consistency between this classifier and animal’s response, since RT values above 39oC could be indicative of initial stress for bovines. Thus, fuzzy logic classifier developed using infrared thermography as an input is a promising tool to allow the environmental assessment and control for automation system for animal housing.
