Prediction of Indoor Air Quality Using Artificial Neural Networks

This paper described an application of artificial neural networks (ANNs) to predict the indoor air quality (IAQ). Six indoor air pollutants and three indoor comfort variables were used as input variables to the networks. An occupant symptom metric (PIAQ) was used as the measure of indoor air quality, and employed as the output variable.Pollutant concentration, comfort variable, and PIAQ data were obtained from previous studies. Feed-forward networks that employed back-propagation algorithm with momentum term and variable learning rate were used in ANN modeling.Among constructed networks, the best prediction performance was observed in a two-hidden-layered network with the high correlation coefficient and low root mean square error for the test set. Meanwhile, the constructed networks had a better performance than the multiple linear regression analysis. The results showed that the ANN approach can be applied successfully in predicting indoor air quality.