EXPERIMENTAL ANALYSIS OF PID-CONTROLLED HEAT RECOVERY AIR HANDLING UNIT BY MACHINE LEARNING METHODS

Abstract

The need for energy continues to increase rapidly all over the world day by day. The demand for energy is rising in developing economies and industrial production areas in these economies. Energy is one of the essential inputs for business areas. Today, most energy is used in buildings for the purpose of heating, cooling, and air conditioning. The problem with air-conditioning systems is that the airflow remains constant despite the change in the number of people in the interior areas. The interest of this paper is to explore how the thermal comfort and air quality of a cloud-based proportional integral derivative (PID)-controlled plate heat exchanger recovery air handling unit were investigated in a classroom environment. Depending on the variables of the temperature and humidity of the outdoor environment, the number of students in the classroom, and the amount of fresh air sent to the indoor environment, the temperature, humidity, and air quality values of the indoor environment were controlled. In line with the data received from the cloud-based system, indoor temperature and indoor air quality values were analyzed by using the machine learning methods separately, that is, support vector machine (SVM), Gauss process regression (GPR), regression trees (RT), and ensembles of trees (ET). In the experiment set, the class's CO₂, temperature, and relative humidity values were compared with the R² values by machine learning methods when the air handling unit was started. As a result of the comparison, the R² value of the amount of CO₂ was obtained from the GPR method at 99%, the temperature amount from the GPR method at 93%, and the relative humidity amount from the GPR method at 98%.