Chapter 6. Change Detection Based on Sliding Windows

  1. Fredrik Gustafsson

Published Online: 16 OCT 2001

DOI: 10.1002/0470841613.ch6

Adaptive Filtering and Change Detection

Adaptive Filtering and Change Detection

How to Cite

Gustafsson, F. (2001) Change Detection Based on Sliding Windows, in Adaptive Filtering and Change Detection, John Wiley & Sons, Ltd, Chichester, UK. doi: 10.1002/0470841613.ch6

Author Information

  1. Linkoping University, Linkoping, Sweden

Publication History

  1. Published Online: 16 OCT 2001

ISBN Information

Print ISBN: 9780471492870

Online ISBN: 9780470841617

SEARCH

Keywords:

  • data segmentation;
  • change detection;
  • jump Markov model;
  • piecewise linear regression model;
  • Matlab;
  • Matlab toolbox;
  • autotuning;
  • signal processing;
  • adaptive signal processing;
  • adaptive systems;
  • fault diagnosis;
  • monitoring;
  • surveillance;
  • signal detection;
  • decision making;
  • statistics;
  • stochastic processes;
  • modeling;
  • estimation;
  • identification;
  • parameter estimation;
  • state estimation;
  • observers;
  • stochastic systems;
  • time-varying systems;
  • discrete time systems;
  • sampled data systems;
  • filters;
  • communication systems;
  • adaptive control

Summary

Model validation is the problem of deciding whether observed data are consistent with a nominal model. Change detection based on model validation aims at applying a consistency test in one of the following ways:

  • The data are taken from a \emindex{sliding window}. This is the typical application of \emindex{model validation}.

  • The data are taken from an increasing window. This is one way to motivate the \emindex{local approach}. The detector becomes more sensitive when the data size increases, by looking for smaller and smaller changes.

The nominal model will be represented by a parameter vector. This may be obtained in one of the following ways:

  • It is recursively identified from past data, except for the ones in the sliding window. This will be our typical case.

  • It corresponds to a nominal model, obtained from physical modeling or system identification.

Particular approaches include the generalized likelihood ratio test, the asymptotic local approach and the divergence test. Isolation and identification of faults, commonly included in the diagnosis step, is given particular attention.