Development of a thermal control algorithm using artificial neural network models for improved thermal comfort and energy efficiency in accommodation buildings

The aim of this study was to develop a control algorithm to demonstrate the improved thermal comfort and building energy efficiency of accommodation buildings in the cooling season. For this, two artificial neural network (ANN)-based predictive and adaptive models were developed and employed in the algorithm. One model predicted the cooling energy consumption during the unoccupied period for different setback temperatures and the other predicted the time required for restoring current indoor temperature to the normal set-point temperature. Using numerical simulation methods, the prediction accuracy of the two ANN models and the performance of the algorithm were tested. Through the test result analysis, the two ANN models showed their prediction accuracy with an acceptable error rate when applied in the control algorithm. In addition, the two ANN models based algorithm can be used to provide a more comfortable and energy efficient indoor thermal environment than the two conventional control methods, which respectively employed a fixed set-point temperature for the entire day and a setback temperature during the unoccupied period. Therefore, the operating range was 23-26 degrees C during the occupied period and 25-28 degrees C during the unoccupied period. Based on the analysis, it can be concluded that the optimal algorithm with two predictive and adaptive ANN models can he used to design a more comfortable and energy efficient indoor thermal environment for accommodation buildings in a comprehensive manner. (C) 2016 Elsevier Ltd. All rights reserved.