- 195 views
- 315 downloads
Performance Analysis and Design for Abnormality Detection Systems
-
- Author / Creator
- Xiong, Ying
-
The performance-orientated abnormality detection design problem has drawn more attention nowadays. This thesis focuses on analytical performance analysis and the corresponding detector design problem for linear filters, the Kullback-Leibler divergence (KLD) and Renyi divergence.
First, sensitivity analysis and sensitivity-based design are conducted for linear alarm filters. Analytical expressions are derived to quantify the sensitivity of a linear alarm filter with unknown data distributions, based on which a new design scheme is formulated to minimize the weighted sum of detection errors subject to upper bounds on the system sensitivities. The second work is on the KLD based detection for independent and identically distributed (i.i.d.) data under generalized Gaussian distributions with shape parameters greater than 1. The false alarm rate (FAR) is analytically obtained and two detection algorithms with constant and adaptive thresholds are proposed. The third work studies the Renyi divergence based detection for i.i.d. multivariate Gaussian data, where the divergence order is between 0 and 1. The off-set and scaling faults are considered under the abnormal condition. The FAR and missed alarm rate (MAR) are derived analytically, based on which a detection scheme is proposed with an adaptive divergence order.
Intensive case studies with both simulated and experimental data are conducted to verify the analytical results and to show improvement of the proposed detection schemes over existing ones. -
- Graduation date
- Fall 2019
-
- Type of Item
- Thesis
-
- Degree
- Doctor of Philosophy
-
- License
- Permission is hereby granted to the University of Alberta Libraries to reproduce single copies of this thesis and to lend or sell such copies for private, scholarly or scientific research purposes only. Where the thesis is converted to, or otherwise made available in digital form, the University of Alberta will advise potential users of the thesis of these terms. The author reserves all other publication and other rights in association with the copyright in the thesis and, except as herein before provided, neither the thesis nor any substantial portion thereof may be printed or otherwise reproduced in any material form whatsoever without the author's prior written permission.