Energy Demand Forecasting in Building using Gate Recurrent Unit

Gemelfour Ardiatus Sudrajad^*Riri Fitri Sari

• University of Indonesia, Depok, Indonesia

Power plants in Indonesia still utilize fossil fuels to create control, especially inside the Java-Bali region, which provides more than 70% of control needs [1]. Isolated from that, from PLN's load profile gathering in prime time, that was almost used for the building sector. Building energy accounts for 33% of world energy utilization and 40% of world GHG spreads clearly and in an indirect way. The problem of electricity waste will increase other impacts, such as bill inflation and greenhouse gas emissions. The target of the Common National Energy Orchestrate (RUEN) in 2025 is to achieve the target of 23% of the energy mix within the power plant. An energy management and forecasting method is needed to create a savings strategy. This paper proposes a significant learning procedure based on Gate Unit (GRU) and a couple of deep learning algorithms to forecast energy. This procedure examines components tallying energy utilization plans in buildings and the non-linear associations between these parameters on an hourly, daily, weekly, and monthly basis inside. In this examination, a building dataset was utilized that recorded energy utilization (W/h) for 3 years, from 2012 to 2015. The foremost vital energy utilization inclination is inside the prime-time run of 19.00-22.00, and the least energy utilization is inside the 04.00-07.00 run. The outcomes approximately outline that the proposed procedure can assess building energy risk and energy supply with a high level of exactness, showing that a GRU regressor incorporates a MAPE regard of 26.23% good run in hourly data figures for one month ahead. GRU is able to have a good level of accuracy and lighter computation than several other forecasting methods. GRU forecast model can also be connected to RE generation to decide the characteristics and potential of RE within the building zone. Separate from that, an optimization demonstration can be created to balance supply and request for electrical energy. And it can also be utilized to mechanize the utilization of electrical gadgets within the building.