Otimização do bem-estar global em programas de resposta à demanda baseados em incentivos

Abstract

The increasing use of distributed resources based on renewable energy resources has triggered transformations in the electricity distribution systems. The growth of distributed resources causes the need for a reorganization in the operation of distribution networks and, consequently, the development of microgrids. The microgrids appear as practical solutions for the large-scale integration of distributed resources. Microgrids are small electrical systems that can operate in parallel with or disconnected from the power grid. Due to the intermittency of renewable sources, it is necessary to forecast generation as well as to manage demand, in order to better equalize the energy balance of the microgrid. This project aims to simulate methods for forecasting generation and to optimize incentive-based demand response methods. Different methods for short-term forecasting of photovoltaic solar generation were studied, such as neural networks, time series and persistence methods. Based on computational effort and precision, neural networks were chosen. For Demand Response, incentive-based programs were evaluated to seek to change the initial demand curve for a set of loads to a final curve that takes into account the effects of energy supply and demand. The benefits of Demand Response methods are noticeable for consumers, who can receive financial rewards for adapting to programs, and for utilities, who can have their costs reduced due to scaling the system for peak hours to be less costly if the peak demand is less. Finally, the incentive value offered every hour was optimized, generating Pareto fronts that represent points of optimized global cost for both the utility and consumers.