The abundance of residuals and fuzzy decision-making for fault diagnosis and sensor signal validation

The generation of abundant residual signals and the decision-making problems arising from sensor failure diagnosis and signal validation for low direct redundant measurement systems are investigated. The redundancy is supplemented by incorporating the linear interaction equations into the original measurement model, then the abundance of residuals is achieved based on the parity space method. This leads to higher flexibility in the decision-making for fault diagnosis. The fuzzy analysis for pattern recognition is applied together with the threshold test which results in a more reliable sensor failure detection and diagnosis