Year
2011
Abstract
Current solution monitoring evaluation systems provide basic auto-correlation and crosscorrelation capabilities that can detect disagreements. When evaluating batch-operated tank data, Tank Monitoring Evaluation Systems (TaMESs) also make limited diagnoses regarding these disagreements, by applying rules to a sub-event representation of the data. This paper extends this sub-event representation to tanks that are fed or emptied continuously. The incorporation of model-based reasoning and Bayesian updating into a TaMES is also discussed. Model-based reasoning reduces the number of unresolved events, in general, and in particular improves reasoning in continuous operated tanks. Bayesian updating is introduced to guide inspectors in their resolution of unresolved sub-events.