Secondary control activation analysed and predicted with explainable AI

The transition to a renewable energy system challenges power grid operation and stability. Secondary control is key in restoring the its reference following disturbance. Underestimating necessary capacity may require emergency measures, such that solid understanding of predictability driving factors needed. Here, we establish an explainable machine learning model for analysis secondary Germany. Training gradient boosted trees, obtain accurate ex-post description activation. Our demonstrates strong impact external drivers as forecasting errors generation mix, while daily patterns reserve activation play minor role. prototypical model, identify forecast error estimates crucial improve predictability. Generally, input data training have be carefully adapted serve different purposes either or sizing.