Journal of Korean Institute of Architectural Sustainable Environment and Building Systems ISSN:1976-6483(Print) 2586-0666(Online) 한국건축친환경설비학회논문집

The objective of this study is to develop prediction model of indoor carbon dioxide (CO₂) concentration using machine learning algorithm. Indoor CO₂ concentration is one of the indicators of indoor ventilation standard, and indoor air quality and ventilation performance can be checked through CO₂ concentration. The machine learning model is a method of analyzing the relationship between measured input/output data and does not require a high level of theoretical knowledge about the output value to be predicted, making it easy to develop a prediction model. In this study, a CO₂ prediction model was developed using an artificial neural network, a support vector machine, a random forest, and a K-nearest neighbor algorithm based on the existing HVAC system operation data. When comparing the performance of the developed CO₂ prediction model, the ANN model showed high performance. As a result of analyzing the time series data using the developed model, the measured indoor CO₂ concentration and the CO₂ concentration of the prediction model were similar, but on average, a relative error of less than about 5% occurred.