Information Based Data-Driven Characterization of Stability and Influence in Power Systems

TL;DR

This paper introduces a data-driven approach using Koopman operators and information transfer to classify stability types and identify influential states in power systems, moving beyond traditional linear models.

Contribution

It presents a novel framework combining Koopman operator theory and information transfer for stability analysis and influence characterization in power networks.

Findings

Successfully classified voltage and angle stability in test systems.

Identified influential generators and states contributing to instability.

Validated framework on 3-bus and IEEE 9-bus systems.

Abstract

Stability analysis of a power network and its characterization (voltage or angle) is an important problem in the power system community. However, these problems are mostly studied using linearized models and participation factor analysis. In this paper, we provide a purely data-driven technique for small-signal stability classification (voltage or angle stability) and influence characterization for a power network. In particular, we use Koopman operator framework for data-driven discovery of the underlying power system dynamics and then leverage the newly developed concept of information transfer for discovering the causal structure. We further use it to not only identify the influential states (subspaces) in a power network, but also to clearly characterize and classify angle and voltage instabilities. We demonstrate the efficacy of the proposed framework on two different systems, namely the 3-bus system, where we reproduce the already known results regarding the types of instabilities, and the IEEE 9-bus system where we identify the influential generators and also the generator (and its states) which contribute to the system instability, thus identifying the type of instability.