Modelling of AQI related to building space heating energy demand based on big data analytics

According to the annual air quality report released by the Ministry of Environmental Protection of China, the cities that requiring heating reported an average of 168.3 days that could not reach the air quality standard, which was 59.8% higher than warm cities that did not require heating in 2014. Coal-fired energy conversion for building space heating can generate a large amount of haze pollutions during the winter. More seriously, the year-by-year data of district heating capacity increases with an average growth rate of 5.59% published in the China Statistical Yearbook, which means more and more heating associated pollutants will emit into the atmosphere and result in the worse and worse environment quality predictability in the further without taking any effective measure. In this context, considering the factors of energy, environment and human health simultaneously, this paper aimed to develop an air quality index (AQI) associated with energy conversion for building space heating model to guide the energy-saving measures on heating. Principal component analysis and grey relation projection evaluation theory approaches were used to preprocess the historical AQI data. Moreover, a simplified two-dimensionality heating capacity model based on EnergyPlus simulation results was introduced to expand the application range of the AQI-Heating (A-H) model. The A-H model included three components, which were the direct pollutant emission, regenerated pollutant emission and urban self-purifying capacity. Four pollutants related input coefficients including the direct emission factor A, the urban maximum purifying capacity a, the attenuation coefficient b and the regenerated coefficient ξ were solved by homemade code based on Matlab software according to a 10-year observed data (2003-2015). The performance of A-H model was compared with the actual values (data from 2015 to 2016) and the results of conventional linear and nonlinear regression techniques, in terms of the index of mean absolute error (MAE), mean absolute percentage error (MAPE), root-mean-squared error (RMSE) and index of agreement (IA). All performance metrics proved that the A-H model was accurate enough (e.g. IA = 0.983) to analyze the relationship between the AQI and the energy conversion for building space heating.