Adaptive Unscented Kalman Filter using Maximum Likelihood Estimation

Research output: Contribution to journalConference articleResearchpeer-review

529 Downloads (Pure)

Abstract

The purpose of this study is to develop an adaptive unscented Kalman filter (UKF) by tuning the measurement noise covariance. We use the maximum likelihood estimation (MLE) and the covariance matching (CM) method to estimate the noise covariance. The multi-step prediction errors generated by the UKF are used for covariance estimation by MLE and CM. Then we apply the two covariance estimation methods on an example application. In the example, we identify the covariance of the measurement noise for a continuous glucose monitoring (CGM) sensor. The sensor measures the subcutaneous glucose concentration for a type 1 diabetes patient. The root-mean square (RMS) error and the computation time are used to compare the performance of the two covariance estimation methods. The results indicate that as the prediction horizon expands, the RMS error for the MLE declines, while the error remains relatively large for the CM method. For larger prediction horizons, the MLE provides an estimate of the noise covariance that is less biased than the estimate by the CM method. The CM method is computationally less expensive though.
Original languageEnglish
JournalIFAC-PapersOnLine
Volume50
Issue number1
Pages (from-to)3859-3864
Number of pages6
ISSN2405-8963
DOIs
Publication statusPublished - 2017
Event20th IFAC World Congress 2017 - Toulouse, France
Duration: 9 Jul 201714 Jul 2017
Conference number: 20
http://www.ifac2017.org/

Conference

Conference20th IFAC World Congress 2017
Number20
CountryFrance
CityToulouse
Period09/07/201714/07/2017
Internet address

Keywords

  • Control and Systems Engineering
  • Adaptive filtering
  • Continuous glucose monitors
  • Covariance estimation
  • Covariance matching technique
  • Maximum likelihood estimation
  • Unscented Kalman filter

Cite this

@inproceedings{4e0f56f1b70f4ef094be1d42338b73b6,
title = "Adaptive Unscented Kalman Filter using Maximum Likelihood Estimation",
abstract = "The purpose of this study is to develop an adaptive unscented Kalman filter (UKF) by tuning the measurement noise covariance. We use the maximum likelihood estimation (MLE) and the covariance matching (CM) method to estimate the noise covariance. The multi-step prediction errors generated by the UKF are used for covariance estimation by MLE and CM. Then we apply the two covariance estimation methods on an example application. In the example, we identify the covariance of the measurement noise for a continuous glucose monitoring (CGM) sensor. The sensor measures the subcutaneous glucose concentration for a type 1 diabetes patient. The root-mean square (RMS) error and the computation time are used to compare the performance of the two covariance estimation methods. The results indicate that as the prediction horizon expands, the RMS error for the MLE declines, while the error remains relatively large for the CM method. For larger prediction horizons, the MLE provides an estimate of the noise covariance that is less biased than the estimate by the CM method. The CM method is computationally less expensive though.",
keywords = "Control and Systems Engineering, Adaptive filtering, Continuous glucose monitors, Covariance estimation, Covariance matching technique, Maximum likelihood estimation, Unscented Kalman filter",
author = "Zeinab Mahmoudi and Poulsen, {Niels Kj{\o}lstad} and Henrik Madsen and J{\o}rgensen, {John Bagterp}",
year = "2017",
doi = "10.1016/j.ifacol.2017.08.356",
language = "English",
volume = "50",
pages = "3859--3864",
journal = "IFAC-PapersOnLine",
issn = "2405-8963",
publisher = "Elsevier",
number = "1",

}

Adaptive Unscented Kalman Filter using Maximum Likelihood Estimation. / Mahmoudi, Zeinab; Poulsen, Niels Kjølstad; Madsen, Henrik; Jørgensen, John Bagterp.

In: IFAC-PapersOnLine, Vol. 50, No. 1, 2017, p. 3859-3864.

Research output: Contribution to journalConference articleResearchpeer-review

TY - GEN

T1 - Adaptive Unscented Kalman Filter using Maximum Likelihood Estimation

AU - Mahmoudi, Zeinab

AU - Poulsen, Niels Kjølstad

AU - Madsen, Henrik

AU - Jørgensen, John Bagterp

PY - 2017

Y1 - 2017

N2 - The purpose of this study is to develop an adaptive unscented Kalman filter (UKF) by tuning the measurement noise covariance. We use the maximum likelihood estimation (MLE) and the covariance matching (CM) method to estimate the noise covariance. The multi-step prediction errors generated by the UKF are used for covariance estimation by MLE and CM. Then we apply the two covariance estimation methods on an example application. In the example, we identify the covariance of the measurement noise for a continuous glucose monitoring (CGM) sensor. The sensor measures the subcutaneous glucose concentration for a type 1 diabetes patient. The root-mean square (RMS) error and the computation time are used to compare the performance of the two covariance estimation methods. The results indicate that as the prediction horizon expands, the RMS error for the MLE declines, while the error remains relatively large for the CM method. For larger prediction horizons, the MLE provides an estimate of the noise covariance that is less biased than the estimate by the CM method. The CM method is computationally less expensive though.

AB - The purpose of this study is to develop an adaptive unscented Kalman filter (UKF) by tuning the measurement noise covariance. We use the maximum likelihood estimation (MLE) and the covariance matching (CM) method to estimate the noise covariance. The multi-step prediction errors generated by the UKF are used for covariance estimation by MLE and CM. Then we apply the two covariance estimation methods on an example application. In the example, we identify the covariance of the measurement noise for a continuous glucose monitoring (CGM) sensor. The sensor measures the subcutaneous glucose concentration for a type 1 diabetes patient. The root-mean square (RMS) error and the computation time are used to compare the performance of the two covariance estimation methods. The results indicate that as the prediction horizon expands, the RMS error for the MLE declines, while the error remains relatively large for the CM method. For larger prediction horizons, the MLE provides an estimate of the noise covariance that is less biased than the estimate by the CM method. The CM method is computationally less expensive though.

KW - Control and Systems Engineering

KW - Adaptive filtering

KW - Continuous glucose monitors

KW - Covariance estimation

KW - Covariance matching technique

KW - Maximum likelihood estimation

KW - Unscented Kalman filter

U2 - 10.1016/j.ifacol.2017.08.356

DO - 10.1016/j.ifacol.2017.08.356

M3 - Conference article

VL - 50

SP - 3859

EP - 3864

JO - IFAC-PapersOnLine

JF - IFAC-PapersOnLine

SN - 2405-8963

IS - 1

ER -