Investigation of Performance of Electric Load Demand Forecasting with New Architecture Realized in Long Short-Term Memory Deep Learning Network

摘要

Electric load demand increases tremendously especially for a Metropolitan city like Delhi due to climatic conditions, population growth, local area development, industries expansion, air pollution, thermal devices usage, etc. Hence, the accuracy of electric load demand forecasting (ELDF) is a deciding factor for the power distribution network to retain as an efficient and consumer friendly network. The proposed day-ahead ELDF models are created on a new architecture with an addition of new input features such as electricity tariff rate and consumer type as compared to existing input features considered already for load demand forecasting prevailing literatures. The models are developed in one of the latest deep learning techniques called standard LSTM (Long Short-Term Memory) networks.