Fault Detection and Diagnosis using Qualitative Modelling and Interpretation

The Qualitative Modelling and Interpretation (QMi) system translates noisy sensor data into a qualitative description of the underlying behaviorof a chemical plant and uses this information together with qualitative models to identify faults and operating regimes. Qualitative models of normal and faulty equipment are simulated to describe the range of possible behaviors in a chemical plant without the need forexact numeric models which are unavailable for many faults. Sensordata are then used to select between different models. Simultaneously using interpretations from multiple sensors reduces sensitivity to sensor noise, increasing diagnosis reliability. QMI has been implemented on a simulated propylene glycol reactor with good results. ~ Failure detection, identification, leak detection, monitoring, qualitative simulation.