ANN‐based energy consumption prediction model up to 2050 for a residential building: Towards sustainable decision making

Abstract

AbstractThe energy consumption in the residential sector has been increased steadily and occupied approximately 30–40% of overall energy consumption. Recent researches on energy consumption have highlighted the significance of residential building energy consumption forecast for enhanced decision‐making in terms of an energy conservation plan. Therefore, it is essential to predict the energy consumption of a residential building by developing a precise prediction model with 95% coefficient bounds. In this paper, an energy consumption data‐driven prediction model is developed using the artificial neural network (ANN) and TRNSYS software. This ANN model is trained with deep learning by using the Levenberg–Marquardt backpropagation algorithm. A 2BHK single‐story multizone residential building having six zones (two bedrooms, one living room, one kitchen, and two toilets) has been modeled in TRNSYS to estimate the energy consumption based on predicted temperature and humidity. First, the data mining technique is used to discover and summarize the historical weather data for temperature and relative humidity prediction. Secondly, the cooling and heating energy consumption has been estimated based on predicted relative humidity and temperature in TRNSYS. In contrast, the energy consumption of ventilation and lighting system is calculated mathematically based on SP 41 standard.