The vision of a low-carbon energy future requires the power system to undergo rapid changes, that it was not designed for. On the supply side, an increasing share of our energy mix is coming from intermittent renewable generators, often distributed across the energy grid. On the demand side, new demands such as electrification of transportation could place unprecedented loads on the power grid. Moreover, there is an increasing decentralization and large-scale availability of data (for example, both the UK and EU aim to have a smart meter deployed in nearly every home). Artificial Intelligence is increasingly seen as one of the pillars for enabling a truly “smart” power grid.
This talk will aim to provide an overview for how emerging techniques in Artificial Intelligence, machine learning and distributed optimization techniques can help us achieve this vision. Specific topics covered will include: how electric vehicle charging can be better coordinated to allow the system to deal with an ever-increasing number of vehicles, or allow them to charge at different points, to minimize the risks for the power grid.
Second, we will talk about how emerging concepts such as virtual power plants could help us to better integrate renewable energy resources into existing grids and reduce curtailment, and the role game theoretic analysis can play in this. We will also cover the increasingly important topic of distributed demand response, looking at challenges and algorithms to coordinate and incentivise a large number of devices to provide flexibility for the grid in real time. Finally, I will outline a future where the power system can use real-time data from distributed smart meters and network monitoring devices to automatically dispatch loads, detect network faults and “self-heal”. I also discuss the challenges in a scenario where generation comes from large offshore wind turbines, monitored and maintained by autonomous robots.