Reliability Models for Sensor Fault Detection with State-Estimator Schemes
Concerning fault detection reliability, several state-estimator-based schemes for a fast detection of incipient instrument faults in dynamic systems are analysed. Based on the theory of Markov these schemes and a classical hardware-realised majority voting system are modeled. Provided with real values for the fault and repair rates, these models facilitate a comparison of the different schemes for deterministic and stochastic instrument fault detection and iso lation (IFDI) by calculating their survival probability, their mean life, their availability or other characteristic quantities of reliability. To allow a uniform comparison, an intelligent measuring system (IMS) with three different input variables to be supervised is taken as a basis. For this example, numerical results and simulation results show the advantages and disadvantages of each estimator-scheme also compared with the hardware-realised majority voting system. In order to demonstrate special characteristics of certain schemes, such as robustness against parameter uncertainties or unknown inputs, the influence of the quality of estimators will be investigated and discussed.
KeywordsFault Detection Estimator Scheme Repair Rate Unknown Input Scheme Life
Unable to display preview. Download preview PDF.