Artificial Intelligence for Smart Energy Management within Energy Transition

Umart grid (SG) is an electric system that includes heterogeneous and distributed electricity generation, transmission, distribution, and consumption components. It is the next generation of power grid able to manage electricity demand (consumption/generation/distribution) in sustainable, reliable and economic manner through the penetration of renewable energies (solar, wind, etc.). Therefore, SG includes also an intelligent layer that analyzes the data produced by the consumers as well as by the production side in order to optimize the consumption and the production according to weather conditions and the consumer profile and habits. Moreover, it can enhance the use of green energy through the penetration of renewable energy and the demand response.

SG presents several research problems and challenges that need to be addressed in order to enhance the energy efficiency of traditional/renewable power generators through user participation, to facilitate the penetration (integration) of distributed/centralized renewable energy systems into electric grids, to reduce the peak load by the use of efficient demand response strategies, to balance and optimize generation and consumption, to reinforce the grid protection (grid resilience, fault diagnosis and prognosis, grid self-healing and recovery, etc.) as well as cyber security and privacy issues, etc.

This survey paper treats the problem of smart energy management within the context of energy transition (demand side management, load monitoring, micro grids, energy consulting (residents, utilities), energy saving, dynamic pricing revenue management, smart meters). It presents the motivation, impacts and challenges related to this hot topic. Then, it focuses on the use of Artificial Intelligence (AI) techniques and tools to address the challenges related to the smart energy management. A global scheme (data acquisition and treatment, feature extraction, learning, inference, visualization etc.) presenting the general principals of these techniques is provided. Then, these techniques are compared according to some meaningful criteria such as the used features (consumption/energy features, statistical features, temporal features, civic, house or building features, etc.) and type of data (active power consumption, reactive power, power factor, etc.), and the level of intrusiveness. The goal is to compare these techniques by showing their advantages and drawbacks according to the application conditions and constraints within the context of energy transition. Several examples are used throughout the paper in order to illustrate the presented methods and concepts.