Artificial neural networks based prediction for thermal comfort in an academic classroom

A predictive models were developed to determine the thermal comfort level for the academic classroom by using artificial neural networks (ANNs). The paper reports experimental and theoretical analysis on a problem of achieving a desired thermal comfort level. The proposed method focused on the classical artificial (feed forward) neural networks (ANN) and the time-series NARX feedback neural networks to achieve the thermal comfort assessed using the predicted mean vote (PMV). The field measurements were conducted in a selected classroom of the Thai-Nichi Institute of Technology (TNI), Thailand. The predicted PMV agreed well with tested PMV data. Therefore, the results would be further demonstrating the feasibility and performance of the approach to achieve the classroom thermal comfort.