Dr. Faridah and team members in the Integrated Smart and Green Building Expertise Group (INSGREEB) successfully disseminated research in the international journal Building Services Engineering Research and Technology in 2021, with the title Feasibility study to detect occupant thermal sensation using a low-cost thermal sensation using a low-cost thermal camera for indoor environments in Indonesia.
In this study, a non-contact building thermal comfort measurement system was developed. This is motivated by the significant influence of thermal comfort conditions on the health, productivity, and psychological conditions of building occupants. Data from electricity consumption in buildings in Indonesia show that the use of air conditioning (AC) is very important to maintain thermal comfort. Unfortunately, the use of AC is often not optimal and wastes electricity, because the AC temperature setting is set at a constant level that does not consider the thermal comfort of the occupants. For example, suppose the AC temperature is too low, in addition to the occupants will feel cold, but will also be very wasteful of electrical energy consumption. On the other hand, if the AC temperature is too high, the occupants will feel hot. An intelligent AC control system is needed in order to balance the thermal comfort needs of the occupants with the need for energy conservation. Ideally, it is hoped that the AC system can intelligently regulate the room temperature, by reading in real time the condition of the occupants’ thermal comfort.
To detect thermal comfort, we need a non-contact sensor system that can predict the comfort level of occupants. The level of comfort is a subjective thing depending on the perception of each person. Therefore, Faridah, a lecturer at DTNTF UGM, and a joint research team studied the feasibility of using thermal cameras and artificial neural networks to detect thermal comfort . Thermal cameras function as sensors that detect the skin temperature of building occupants by capturing infrared radiation emitted by the human body. The use of the camera as a non-contact sensor is necessary because it can measure skin temperature without disturbing the occupants’ activities. With today’s technological advances, thermal cameras can read infrared radiation emitted by occupants and convert it to temperature readings in real time.
The temperature of the forehead, nose, chin, and cheeks was taken with a thermal camera and entered into the artificial neural network (ANN) system to predict the classification of thermal sensations with 7 levels (hot, warm, moderately warm, neutral, moderately cool, cool, cold). The thermal comfort scale with 7 categories was developed in subtropical countries which have a larger temperature range. It is hoped that from this study it can be learned whether this thermal comfort scale is appropriate for use in tropical countries such as Indonesia. This research was published in the journal Building Services Engineering Research & Technology in 2021 .
To conduct the test, 17 Indonesian male respondents were tested in the climate chamber at Engineering Faculty, UGM. Various temperature ranges were tested on the respondents, and their subjective perceptions were recorded and compared with temperature measurements taken via a thermal camera. To analyze the data more in temperature measurements with the camera compared to contact measurements with a thermocouple to validate the temperature data taken. An ANN system with 4 input neurons (temperature on the forehead, nose, chin and cheeks), one hidden layer, and 7 output neurons (7 levels of thermal comfort sensation) was developed from these data. The number of neurons in the hidden layer is varied to get the optimal ANN architecture. The optimal ANN, or able to predict well, has high recall, precision, and F1-score values.
Based on the test results, it was found that respondents in Indonesia tend to be more sensitive to the sensation of cooling than to the sensation of heating, presumably due to adaptation to the tropical environment. The analysis of variance (ANOVA) test shows that the level of thermal comfort can be strongly correlated with the mean skin temperature (MST), which is a parameter that combines the average temperature of the body part being tested. It was also found that the thermal camera could not measure skin temperature as well as the thermocouple sensor. The 7-level comfort scale is not very suitable for Indonesian respondents because they are not used to distinguishing temperature differences. When a 7-level thermal comfort scale is used, the ANN system has an accuracy of 35.7%. When the number of categories is used 5-level and 3-level, the accuracy of the ANN system is better, namely 52.2% and 68.7%, respectively. This shows the need to use a different scale with a smaller number of levels, to suit the conditions or experiences of respondents in tropical countries such as Indonesia.
From the results of this study, it is hoped that it can be a starting point for the development of a better non-contact thermal comfort measurement system in the future. The number of respondents who are limited and only limited to men can be further expanded by increasing the number of participants and studying responses in all genders. If the accuracy of the system can be improved, we can optimize the use of building ventilation systems that are smarter and more efficient.
For questions or exploration of research collaborations, you can correspond directly to:
Email: email@example.comFaridah, Departemen Teknik Nuklir Teknik Fisika, Universitas Gadjah Mada
Jl. Grafika No.2, Sinduadi, Mlati, Sleman, Daerah Istimewa Yogyakarta, 55281
Faridah, F. et al. Feasibility study to detect occupant thermal sensation using a low-cost thermal camera for indoor environments in Indonesia. Building Services Engineering Research and Technology 42, 389-404, doi:10.1177/0143624421994015 (2021)