Analysis of thermal comfort with predicted mean vote (PMV) index using artificial neural network

Abstract

The thermal comfort standard can be accessed with a predicted mean vote (PMV) index that has six variables: air temperature, air velocity, relative humidity, mean radiant temperature, clothing insulation, and metabolism rate. Calculation of the thermal comfort of a room issued with a PMV index is quite complicated, considering the determination of the effect parameters can take a long time because it has a non-linear relationship. The purpose of this study is to analyze the sensitivity parameters that affect thermal comfort and to develop an artificial neural network model to predict the thermal comfort of the PMV index with two different input scenarios. Sensitivity analysis of thermal comfort uses the rank Spearman method, whereas the modeling of artificial neural networks uses the backpropagation method using a total of 784 data as data training and testing. Based on the results of sensitivity analysis shows metabolism rate and clothing insulation have a significant effect on the sensitivity of thermal comfort. Prediction of thermal comfort with the PMV index using an artificial neural network is very effective and has higher accuracy. The results are proven by validating the two model scenarios with an R2 value of 0.99 and low RMSE value.