Development of a model for energy management in office buildings by neural networks (case study: Bandar Abbas)

Building optimization measures are implemented to reduce energy consumption and environmental pollution. If energy reduction and optimization in the buildings are not measured, the national economy will be severely damaged. The energy consumption in buildings can be reduced by up to 50% by performing optimization measures in the building sector and applying Article 19 of National Building Regulations. In this study, the effective parameters on energy optimization were identified using questionnaires and expert opinions and then, the energy consumption and carbon dioxide were calculated by entering the parameters into DesignBuilder software. The parameters included types of wall and ceiling, area of windows, type of windows, and insulation of wall and ceiling, each of which contain different modes. In order to limit the problem space, a range of parameters changes in a specified interval was selected. Since it is impossible to model all probable modes, first a finite number of models was tested using the software and then, the interaction of inputs with two important outputs (energy and carbon dioxide) was obtained by training two separate neural networks. The network training facilitates the calculation of the amount of energy and carbon dioxide needed for any desired input needless of DesignBuilder software.