Optimizing Power Systems through Tarif Design and Regional Trade Networks

Supplying the electricity demand for a whole region requires a series of sophisticated steps in order to meet the power needs of consumers at all times in a financially optimized manner. One of the most important requirements for a utility company is to always have an accurate forecast of the demand for several days into the future. This helps them schedule the operation of different generators in order to match the demand. In this project, we develop a tool we have coined “E-cast”, a forecasting simulator that is capable of performing short-term and long-term predictions of the electricity demand to enable electricity providers to optimize performance of their facilities and service delivery. Various machine learning algorithms, e.g. deep neural networks such as Long Short-Term Memory (LSTM) and Gated Recurrent Units (GRU) are used to produce the prediction which is then deployed in a web-based tool that can be used by engineers and business managers at electric companies to drive more efficient operations with greater accuracy and precision.