Robust state estimation and online outlier detection using eccentricity analysis

Kyuman Lee, Eric Johnson

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Both state propagation and sensor measurements are often corrupted by unmodeled non-Gaussian or heavy-tailed noise. Without dealing with such outliers, the accuracy of a estimator significantly degrades, and control systems that rely on high-quality estimation lose stability. To estimate the states of dynamic systems in which both types of outliers occur, we propose a novel approach that combines a real-time outlier detection technique with an extended version of an outlier robust Kalman filter (ORKF). Unlike the ORKF for only measurement outliers, the technique, the extended ORKF (EORKF), also handles situations in which propagation outliers arise; that is, to approximately compute the optimal precision matrices of process outliers, we derive equations and algorithms using the variational inference method. Hence, the EORKF does not restrict noise at either a constant or Gaussian level. Furthermore, for lower computational effort and memory uses, our approach employs the typicality and eccentricity data analysis (TEDA), which provides information about the time when outliers occur and runs the EORKF whenever the TEDA detects outliers. The results of Monte Carlo simulations show that our approach leads to greater improvement in robustness and lower computational complexity than existing methods.

Original languageEnglish (US)
Title of host publication1st Annual IEEE Conference on Control Technology and Applications, CCTA 2017
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1350-1355
Number of pages6
Volume2017-January
ISBN (Electronic)9781509021826
DOIs
StatePublished - Oct 6 2017
Event1st Annual IEEE Conference on Control Technology and Applications, CCTA 2017 - Kohala Coast, United States
Duration: Aug 27 2017Aug 30 2017

Other

Other1st Annual IEEE Conference on Control Technology and Applications, CCTA 2017
CountryUnited States
CityKohala Coast
Period8/27/178/30/17

Fingerprint

Outlier Detection
Robust Estimation
Eccentricity
State Estimation
State estimation
Kalman filters
Outlier
Computational complexity
Kalman Filter
Dynamical systems
Control systems
Data storage equipment
Sensors
Data analysis
Propagation
Low Complexity
Dynamic Systems
Computational Complexity
Monte Carlo Simulation
Control System

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Hardware and Architecture
  • Software
  • Control and Systems Engineering

Cite this

Lee, K., & Johnson, E. (2017). Robust state estimation and online outlier detection using eccentricity analysis. In 1st Annual IEEE Conference on Control Technology and Applications, CCTA 2017 (Vol. 2017-January, pp. 1350-1355). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/CCTA.2017.8062646
Lee, Kyuman ; Johnson, Eric. / Robust state estimation and online outlier detection using eccentricity analysis. 1st Annual IEEE Conference on Control Technology and Applications, CCTA 2017. Vol. 2017-January Institute of Electrical and Electronics Engineers Inc., 2017. pp. 1350-1355
@inproceedings{2a1855dc71cf45c898331e112ad2bd05,
title = "Robust state estimation and online outlier detection using eccentricity analysis",
abstract = "Both state propagation and sensor measurements are often corrupted by unmodeled non-Gaussian or heavy-tailed noise. Without dealing with such outliers, the accuracy of a estimator significantly degrades, and control systems that rely on high-quality estimation lose stability. To estimate the states of dynamic systems in which both types of outliers occur, we propose a novel approach that combines a real-time outlier detection technique with an extended version of an outlier robust Kalman filter (ORKF). Unlike the ORKF for only measurement outliers, the technique, the extended ORKF (EORKF), also handles situations in which propagation outliers arise; that is, to approximately compute the optimal precision matrices of process outliers, we derive equations and algorithms using the variational inference method. Hence, the EORKF does not restrict noise at either a constant or Gaussian level. Furthermore, for lower computational effort and memory uses, our approach employs the typicality and eccentricity data analysis (TEDA), which provides information about the time when outliers occur and runs the EORKF whenever the TEDA detects outliers. The results of Monte Carlo simulations show that our approach leads to greater improvement in robustness and lower computational complexity than existing methods.",
author = "Kyuman Lee and Eric Johnson",
year = "2017",
month = "10",
day = "6",
doi = "10.1109/CCTA.2017.8062646",
language = "English (US)",
volume = "2017-January",
pages = "1350--1355",
booktitle = "1st Annual IEEE Conference on Control Technology and Applications, CCTA 2017",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
address = "United States",

}

Lee, K & Johnson, E 2017, Robust state estimation and online outlier detection using eccentricity analysis. in 1st Annual IEEE Conference on Control Technology and Applications, CCTA 2017. vol. 2017-January, Institute of Electrical and Electronics Engineers Inc., pp. 1350-1355, 1st Annual IEEE Conference on Control Technology and Applications, CCTA 2017, Kohala Coast, United States, 8/27/17. https://doi.org/10.1109/CCTA.2017.8062646

Robust state estimation and online outlier detection using eccentricity analysis. / Lee, Kyuman; Johnson, Eric.

1st Annual IEEE Conference on Control Technology and Applications, CCTA 2017. Vol. 2017-January Institute of Electrical and Electronics Engineers Inc., 2017. p. 1350-1355.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Robust state estimation and online outlier detection using eccentricity analysis

AU - Lee, Kyuman

AU - Johnson, Eric

PY - 2017/10/6

Y1 - 2017/10/6

N2 - Both state propagation and sensor measurements are often corrupted by unmodeled non-Gaussian or heavy-tailed noise. Without dealing with such outliers, the accuracy of a estimator significantly degrades, and control systems that rely on high-quality estimation lose stability. To estimate the states of dynamic systems in which both types of outliers occur, we propose a novel approach that combines a real-time outlier detection technique with an extended version of an outlier robust Kalman filter (ORKF). Unlike the ORKF for only measurement outliers, the technique, the extended ORKF (EORKF), also handles situations in which propagation outliers arise; that is, to approximately compute the optimal precision matrices of process outliers, we derive equations and algorithms using the variational inference method. Hence, the EORKF does not restrict noise at either a constant or Gaussian level. Furthermore, for lower computational effort and memory uses, our approach employs the typicality and eccentricity data analysis (TEDA), which provides information about the time when outliers occur and runs the EORKF whenever the TEDA detects outliers. The results of Monte Carlo simulations show that our approach leads to greater improvement in robustness and lower computational complexity than existing methods.

AB - Both state propagation and sensor measurements are often corrupted by unmodeled non-Gaussian or heavy-tailed noise. Without dealing with such outliers, the accuracy of a estimator significantly degrades, and control systems that rely on high-quality estimation lose stability. To estimate the states of dynamic systems in which both types of outliers occur, we propose a novel approach that combines a real-time outlier detection technique with an extended version of an outlier robust Kalman filter (ORKF). Unlike the ORKF for only measurement outliers, the technique, the extended ORKF (EORKF), also handles situations in which propagation outliers arise; that is, to approximately compute the optimal precision matrices of process outliers, we derive equations and algorithms using the variational inference method. Hence, the EORKF does not restrict noise at either a constant or Gaussian level. Furthermore, for lower computational effort and memory uses, our approach employs the typicality and eccentricity data analysis (TEDA), which provides information about the time when outliers occur and runs the EORKF whenever the TEDA detects outliers. The results of Monte Carlo simulations show that our approach leads to greater improvement in robustness and lower computational complexity than existing methods.

UR - http://www.scopus.com/inward/record.url?scp=85047634005&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85047634005&partnerID=8YFLogxK

U2 - 10.1109/CCTA.2017.8062646

DO - 10.1109/CCTA.2017.8062646

M3 - Conference contribution

AN - SCOPUS:85047634005

VL - 2017-January

SP - 1350

EP - 1355

BT - 1st Annual IEEE Conference on Control Technology and Applications, CCTA 2017

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Lee K, Johnson E. Robust state estimation and online outlier detection using eccentricity analysis. In 1st Annual IEEE Conference on Control Technology and Applications, CCTA 2017. Vol. 2017-January. Institute of Electrical and Electronics Engineers Inc. 2017. p. 1350-1355 https://doi.org/10.1109/CCTA.2017.8062646