Energy Consumption Basket Optimization and Residential Buildings Clustering by Fuzzy NN Improvement by AHP Architecture and Weights

Abstract

Iran residential buildings are the major energy consumers of this country. Various factors which are effective on energy consumption behavior in residential buildings have converted energy consumption forecasting problem to a great challenge for the institutes of energy consumption optimization. Considering the energy management essence, the present study aimed at modeling, forecasting, and clustering of energy consumption in residential buildings, for energy labeling and auditing of the mentioned buildings. Therefore, in this research by combining Fuzzy Neural Network and AHP, using data obtained from questionnaires, energy consumption behavior of residential buildings has been clustered. AHP weights coefficients and architecture have been used as the initial weights coefficients and architecture of NN. In both modes of with and with out initial weights coefficients and architecture, NN was deployed with the same data. In order to NN training and testing, 270 residential buildings in the city of Shiraz were employed. Comparing the detecting and clustering power of the two mentioned FNN models indicate that FNN with initial weights coefficients and AHP architecture has more accuracy and speed of clustering and prediction compare to the other model.

Keywords

• fuzzy neural network
• ahp
• energy consumption behavior
• clustering
• residential buildings
• auditing