Investigation of the impact of occupants' behaviour and habits on residential electricity consumption

Residential sector accounts for 30% of the total energy consumption. In Turkey, residential electricity consumption has increased by 27% in last decade. Proportion of residential sector in total electricity consumption has reached 22% in Turkey. This study examines the impact of households behaviours on residential electricity consumption through an information intervention method. An artificial intelligence model called Adaptive Neuro-Fuzzy Inference Systems (ANFIS), which has not been previously applied in electricity consumption behaviour studies, is used to understand factors influencing residential electricity consumption and contribution of behavioural changes to electricity saving potential. A survey was conducted in 100 households in Yalova gathering information about household electricity consumption behaviour, economic, demographic data, and characteristics of electrical appliances in residential buildings. Monthly electricity consumption data of households were obtained from relevant companies. Analyses of monthly electricity consumption of households that received face-to-face energy-saving informational intervention method show a decrease of 10% in electricity consumption, particularly in first month after education. The ANFIS model has a Root Mean Square Error (RMSE) value of 0.038 for the training dataset, and a Mean Absolute Percentage Error (MAPE) value of 22% for the test dataset. These results indicate that model is well trained and prediction performance is at acceptable level. According to the results of the scenarios for behavioural changes applied to the model, it has been revealed that between 7% and 35% can be saved in residential electricity consumption. This study can play crucial role in formulating energy-saving policies and raising awareness in residential sector.