Fault Diagnosis Method Based on D-S Theory of Evidence and AHP

摘要

A new approach to fault diagnosis is presented, with a combination of D-S theory of evidence and AHP. The method for constructing assignment function of basic probability based on AHP with 5-unit preference scale is proposed in detail and its comparison is different from the pairwise comparison in the standard AHP. In addition, the weight of each sensor is taken into account in the process of combination. An illustrative example is given to expatiate on its application to fault diagnosis. After theses pieces of evidence from multi sensors are combined, the reliability of the diagnostic result is improved significantly and its uncertainty is reduced remarkably, which shows the proposed approach to fault diagnosis is reasonable and effective.