Thermal sensor analysis for prediction of comfort levels of human body in air-conditioned environment

Substantial amount of energy is spent in air-conditioning systems in the buildings. However, they often result in over-cooling which may lead to huge waste of energy, as well as dissatisfaction of occupants. The desirable air-conditioning systems should not only create a thermally comfortable environment for occupants but also reduce energy cost as much as possible. To achieve this goal, the control of the air-conditioning should take occupants’ thermal sensation into account so that it can bring an optimal balance between thermal comfort and energy cost. Therefore, prediction of thermal comfort is crucial for this purpose. Skin temperature has proved to be an effective indicator of thermal comfort level. In this work, an intelligent system is presented, which can obtain room temperature, relative humidity and skin temperature data from different sensors and predict the thermal comfort level automatically at every time instant. A thermal camera is used here to measure the skin temperature remotely. And MLP neural network and SVM are used as classifiers to predict the thermal sensation which is quantified using thermal sensation scale. Results show that the accuracy of this predictive model is high enough and SVM is preferred over the MLP neural network in this case.Master of Science (Computer Control and Automation