Indoor heating/cooling represents one of the most significant proportions of domestic buildings’ energy consumption in temperate regions. There is, therefore, huge potentials for energy demand minimisation (and consequently reduction in fossil fuel utilisation) if heating/cooling processes are optimised. The quality of a building’s thermal insulation can be heavily compromised by the presence of cold spots at heat transfer boundaries. One very common situation that could result in cold spots for residential buildings is mould growth within the building’s material, hence controlling/eliminating indoor moulds would contribute to sustainability in space heating/cooling objectives.
This research collects indoor temperature and relative humidity (hygrothermal) data, as well as corresponding building characteristics (heating system, building style/detachment, number of occupants and bedrooms, etc.) data from 15,986 UK buildings and relates this with the measured areas of mould (in square metres) observed in these buildings. The study utilises contemporary database management, AI and process automation techniques to analyse its massive training and testing datasets, which resulted in a deep neural network based mould prediction model for residential buildings.
The study’s social impact is evident in the early warning signalling it can provide for indoor mould and its consequences, viz., fuel poverty, structural/maintenance challenges and mould-related medical/respiratory conditions.