Residual evaluation for fault diagnosis based on adaptive fuzzy thresholds

The goal of residual evaluation for FDI (failure detection and identification) is to decide whether and where a fault has occurred under the condition that wrong decisions causing false alarms are avoided. For residual evaluation, one can use simple threshold logic, statistical decision theory, pattern recognition, fuzzy decision logic or neural networks. In this paper, we focus on the adaptive threshold logic test based upon fuzzy modelling of the process and fuzzy decision making.