Prediction of indoor air ventilation performance in kindergarten using nonlinear autoregressive neural network / Muhammad Kamil Nazzri ... [et al.]

Indoor air pollution has become one of the major issues that cause health issues for building occupants, especially people from sensitive groups such as the elderly and younger children. However, indoor air pollutants can be reduced by providing adequate ventilation to the building. Effective and adequate ventilation can help to dilute and remove pollutants, providing healthier air for the building occupants to breathe. The adequacy of ventilation can be determined by measuring the concentration of carbon dioxide (CO2 ) in the building, as CO2 is widely used as an indicator for ventilation. Method: To determine the ventilation performance, a method of forecasting through a modelling process using a nonlinear autoregressive neural network (NARNN) is developed. The CO2 concentration data that was collected from kindergarten is used to construct and find the best-fitted model with a suitable number of neurons and hidden layers. This model can help predict the future concentration trend of CO2 in kindergarten and determine the ventilation performance of the building. Result: The concentration of CO2 in the building is decreasing through the operation hours, indicating it has adequate ventilation. The dataset of CO2 concentration is used to develop a prediction model that consists of an artificial neural network (ANN) structure. A model with a 1-9-1 structure with a data division of 80:20 is the best-fit model for forecasting as it has high accuracy and is highly relevant to be used for prediction as it has the nearest R-value near one. Conclusion: Indoor air quality needs special attention from multiple authorities and organisations, especially in buildings that have younger children as occupants. Poor indoor air quality can pose a health risk to the occupants and disrupt their comfort while doing their activities in the building. The modelling technique is one of the most relevant and advanced methods to forecast the quality of a building, as it can help determine the future concentration of pollutants in the indoor environment.