Deep Reinforcement Learning is a branch of Artificial Intelligence that allows machines to learn control and to take actions. Asset Management and Optimization are the natural application of DRL due to its ability to capture complex dependencies and choose optimal actions in real time.
 
Our current energy transition to distributed generation drastically increases the number of actions available, with the optimal action changing throughout the day. Resolving the Operator's Dilemma in the environment with a lot of volatile variables (such as electricity price and energy consumption) is hard.  The deterministic set of rules and abstract models used today to guide operations can account for such changes, but they cannot guarantee optimal performance. Abstract models are also limited by the skills of the modeler. Reinforcement Learning agents learn from their experience of the environment, which helps them to see patterns we may miss and removes the need for modeling.
 
Deep Reinforcement Learning can be deployed both centrally and on the edge at any level of the control chain: economic dispatch, battery dispatch, inverter control, cooling/boilers/combustors control, wind farm optimization and even outage detection. Such flexibility makes it the essential tool for integration and optimization of the complex energy systems containing millions of devices. It enables cost reduction, reduces environmental impacts, and improves safety without major capital investments.

This webinar will provide a window into one of the fastest growing potential applications of AI on the power grid.

Presenter:
Pavel Romashkin
Artificial Intelligence Consultant
VOLITANT AI