A Distributionally Robust Control Strategy for Frequency Safety based on Koopman Operator Described System Model
Qianni Cao, Chen Shen
TL;DR
The paper tackles frequency safety under high renewable penetration by introducing a distributionally robust emergency frequency control (DREFC) framework. It combines Koopman-operator based predictions with a Gaussian Mixture Model (GMM) ambiguity set, governed by a Wasserstein-type distance, and an analytical VaR approximation to yield a tractable min-max control that guarantees safety with tunable conservatism. The approach supports both one-shot load shedding and online DC power reference regulation, with online ambiguity-set updates to maintain accuracy. Case studies on the CEPRI-FS system show that DREFC achieves high safety probabilities while reducing computation time and control costs compared with robust or scenario-based methods.
Abstract
As the proportion of renewable energy and power electronics in the power system increases, modeling frequency dynamics under power deficits becomes more challenging. Although data-driven methods help mitigate these challenges, they are exposed to data noise and training errors, leading to uncertain prediction errors. To address uncertain and limited statistical information of prediction errors, we introduce a distributionally robust data-enabled emergency frequency control (DREFC) framework. It aims to ensure a high probability of frequency safety and allows for adjustable control conservativeness for decision makers. Specifically, DREFC solves a min-max optimization problem to find the optimal control that is robust to distribution of prediction errors within a Wasserstein-distance-based ambiguity set. With an analytical approximation for VaR constraints, we achieve a computationally efficient reformulations. Simulations demonstrate that DREFC ensures frequency safety, low control costs and low computation time.