Study of Elevator Fault Diagnosis Method Based on FPN-RSVM

摘要

Fuzzy Petri Net (FPN) is regarded as a tool of model building, knowledge presentation, diagnosis and reasoning of the rule-production system widely used in the field of fault diagnosis. Due to a lack of regulating and learning capacity, FPN cannot adjust to variation of the expert system. Accordingly, a kind of fault-diagnosis method that integrates FPN with a regressive support vector machine (RSVM) is proposed and applied to elevator fault diagnosis. First, the building elevator fault-diagnosis network by FPN, and then Expert diagnosis data of elevators in different running periods are trained through RSVM, so groups of weights corresponding to different running periods are obtained. When the elevator is in trouble, weights corresponding to the elevator's running period were chosen. Thus, the reasoning result could be more accurate and match actual circumstance.