Short-term Load Forecasting System for Smartgrids based on Personal Power Units

Popularization of personal low-power renewable energy sources and storages application led to the appearance of flexible energy markets. When implementing a high-efficiency model of electric power trading the demand for proper assessment of the balance between consumed and produced energy becomes the biggest and the most important challenge for research. The proposed structure of microgrid system is based essentially on the combined use of personal power units and hardware-software complex built on Internet of Things (IoT), MultiAgent (MAS) and Machine Learning (ML) technologies. The main objective of that complex is an adjustment of each consumer's expected daily load curve in accordance with a number of external factors. The data obtained are used for the further formation of the agent's logic, which are carrying out the buy & sell process of electric energy for the benefit of every consumer. The transaction control role falls on each participant in the market. This solution makes it possible to organize the distributed self-regulated market of electric energy and computing resources to ensure the high-effective and reliable power supplying of microgrid participants with minimal involvement from both outer power grid, and the consumers.