RIA: A Bayesian Approach for Adaptive System Level Diagnosis
Tein-Hsiang Lin Principal Investigator
MetadataShow full item record
This research is on developing a probabilistic diagnosis model for adaptive system level diagnosis. In this model the faulty probability of a unit (calculated by a Bayesian approach) is used to determine whether a diagnosis is successful. This limits tests to only those that contribute to the detection of faulty units. Research tasks are: (1) deriving an optimal algorithm for selecting the next test in the adaptive diagnosis based on current post-diagnosis fault probabilities; (2) determining a fixed sequence of tests which minimizes the average number of possible fault patterns; (3) find the minimal subset of tests that lead to a successful diagnosis, assuming all tests pass without detection; and (4) explore the relationship between the diagnosis time and the coverage.