The stochasticity of power injections from weather-dependent renewable energy resources challenges power system operations and increases the risk-exposure of electricity market participants. Defining safety regions for which any operative or market decision must be viable can be an effective and tractable tool to overcoming these challenges. Yet, designing such safety regions from historical data with respect to their impact on the quality of the decision risk is tricky. This seminar discusses two proposals for data-driven robust decision tools in renewable-dominant electric power systems that explicitly internalize probabilistic performance metrics of their solution. First, we discuss an approach to learn prescriptive safety regions for wind power generation. We leverage recent results from differentiable programming to create individual safety intervals for each wind generator that adapt to current grid situations such that they optimize an operator-defined risk target. Second, we discuss a method to internalize adversarial uncertainty sets in electricity market clearing such that consumer risk is minimized. We define these sets using a factor stressing approach inspired from financial engineering and show that consumer risk in terms of excess payments can be effectively reduced at a low overall system cost. Speaker(s): Robert Mieth Agenda: – Talk by Robert Mieth at 11:00 am – Lunch box after the talk at 12:00 pm – You don’t have to be an IEEE member to attend this meeting. Room: 202, Bldg: ECE, New Jersey Institute of Technology, 141 Warren St, Newark, New Jersey, United States, 07103