Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Rapid and Accurate Changepoint Detection of Power System Forced Oscillations

Luke Dosiek, Akaash Karn, Frank Liu

TL;DR

The paper tackles the challenge of rapidly locating forced-oscillation (FO) start and end times in power-system signals. It advances CPD-based FO localization by using a single call to PELT with manually chosen $\beta$, eliminating the prior need for CROPS-driven threshold tuning and the CP cap, and introduces a data-driven minimum FO segment length via a local-SNR criterion. The approach derives a robust pipeline to convert detected CPs into refined FO intervals $$(\tilde{\epsilon}_i,\tilde{\eta}_i)$$ and integrates this with an FO-robust mode-meter workflow, demonstrating a 98% reduction in computation time while maintaining high accuracy in FO timing and in estimating FO amplitude, frequency, and phase. The results suggest near real-time FO time localization with reduced parameter dependence, applicable to mode-meter diagnostics and potentially extendable to nonstationary FO scenarios.

Abstract

This paper describes a new approach for using changepoint detection (CPD) to estimate the starting and stopping times of a forced oscillation (FO) in measured power system data. As with a previous application of CPD to this problem, the pruned exact linear time (PELT) algorithm is used. However, instead of allowing PELT to automatically tune its penalty parameter, a method of manually providing it is presented that dramatically reduces computation time without sacrificing accuracy. Additionally, the new algorithm requires fewer input parameters and provides a formal, data-driven approach to setting the minimum FO segment length to consider as troublesome for an electromechanical mode meter. A low-order ARMAX representation of the minniWECC model is used to test the approach, where a 98\% reduction in computation time is enjoyed with high estimation accuracy.

Rapid and Accurate Changepoint Detection of Power System Forced Oscillations

TL;DR

The paper tackles the challenge of rapidly locating forced-oscillation (FO) start and end times in power-system signals. It advances CPD-based FO localization by using a single call to PELT with manually chosen , eliminating the prior need for CROPS-driven threshold tuning and the CP cap, and introduces a data-driven minimum FO segment length via a local-SNR criterion. The approach derives a robust pipeline to convert detected CPs into refined FO intervals and integrates this with an FO-robust mode-meter workflow, demonstrating a 98% reduction in computation time while maintaining high accuracy in FO timing and in estimating FO amplitude, frequency, and phase. The results suggest near real-time FO time localization with reduced parameter dependence, applicable to mode-meter diagnostics and potentially extendable to nonstationary FO scenarios.

Abstract

This paper describes a new approach for using changepoint detection (CPD) to estimate the starting and stopping times of a forced oscillation (FO) in measured power system data. As with a previous application of CPD to this problem, the pruned exact linear time (PELT) algorithm is used. However, instead of allowing PELT to automatically tune its penalty parameter, a method of manually providing it is presented that dramatically reduces computation time without sacrificing accuracy. Additionally, the new algorithm requires fewer input parameters and provides a formal, data-driven approach to setting the minimum FO segment length to consider as troublesome for an electromechanical mode meter. A low-order ARMAX representation of the minniWECC model is used to test the approach, where a 98\% reduction in computation time is enjoyed with high estimation accuracy.

Paper Structure

This paper contains 12 sections, 20 equations, 5 figures.

Table of Contents

  1. Introduction
  2. Background
  3. The Small-Signal Model
  4. Changepoint Detection
  5. Detecting When a Forced Oscillation Starts and Ends
  6. An Improved FO Changepoint Detector
  7. Manual calculation of $\beta$
  8. Converting changepoints to $\epsilon$ and $\eta$
  9. Minimum FO segment lengths
  10. What if no CPs were detected?
  11. Simulation Study
  12. Conclusions

Figures (5)

  • Figure 1: Improved usage of PELT to estimate FO start/stop samples.
  • Figure 2: Means and regions of $\pm1$ std. dev of time to estimate $\epsilon$ and $\eta$.
  • Figure 3: Means and regions of $\pm1$ std. dev of estimates of $\epsilon$ and $\eta$.
  • Figure 4: Means and regions of $\pm1$ std. dev of estimates of $A$, $f$, and $\theta$.
  • Figure 5: Means and regions of $\pm1$ std. dev of mode estimates.