Detection of Events in Signals Via the Model-Based Fault Diagnosis Approach: Application to Bio-Electrical Signals
Detection and extraction of a particular event is a very common problem in bio-electrical signal processing and can be solved by using the classical model-based fault diagnosis methods. From a statistical point of view, the addressed problem can be seen as a change detection problem when transformed into a stochastic framework where the output is categorical: model change from M θ0 to M θ1 . The purpose of this chapter is to present some detection methods based on a two-stage strategies: residual generation and evaluation. The first part is devoted to a theoretical background presentation of such methods with a special attention to robustness problems. In the second part, three different applications of bio-electrical signal processing are reported.
KeywordsResidual Evaluation Generalize Likelihood Ratio Test Generalize Likelihood Ratio Auto Regressive Residual Generation
Unable to display preview. Download preview PDF.