A COMPARISON AMONG DIFFERENT NOTIONS OF ROBUST DIAGNOSABILITY AGAINST SENSOR FAILURES
Wesley Rodrigues Silveira1; Públio Macedo Lima1; Marcos Vicente de Brito Moreira1
1 Universidade Federal do Rio de Janeiro
Baixar PDF doi:10.20906/CPS/CBA2018-0831
Resumo
Recently, different notions of robust diagnosability of discrete event systems (DES) against permanent and intermittent sensor failures have been proposed in the literature. In these works, different assumptions are considered regarding event observation losses, and different strategies for verifying the diagnosability of the language of the system subject to uncertainties in the event observations are presented. In this paper, we present the relation between different notions of robust diagnosability proposed in the literature, and we show that a system is robustly diagnosable against intermittent loss of observations if, and only if, it is robustly diagnosable against permanent loss of observations. We also show that observation masks can always be replaced with projections in the case of non selective sensor failures to model the diagnoser observation of the system, and that even in the case of non selective sensor failures, the two definitions of robust diagnosability of DES against permanent sensor failures proposed in the literature are not equivalent.
Palavras-chave: Fault diagnosis; Discrete event systems; Robust diagnosability; Sensor failures