Mixed-expert-system-based fault diagnosis for heavy machine
Jun Wu · Computer Integrated Manufacturing Systems · 2010
Considering the complexity of heavy machine structure and fault,the faults of heavy machine was divided into three levels,and the concepts ofrule diagnosis bodyandcase diagnosis bodywere introduced.The knowledge organization mode ofstructure fault treewas evolved based on the product structure tree.And then mapping between rule diagnosis bodies and frame representation were used to describe fault association and fault case respectively.Also,the organization of database was put forward.According to rules characteristics of heavy machine,the rule-based diagnosis was implemented by Back Propagation(BP)nural network trained by the information of rule diagnosis bodies.Besides,the case-based reasoning diagnosis based on case diagnosis body converted to fuzzy membership was proposed as well,in which matching retrieval of to-be-solved case was performed through the nearest neighborhood algorithm.Case reuse and case adjustment were introduced.Aiming at case-base management,new cases were stored by Knearest-neighbor algorithm based on TC similarity.The construction process of original sample cases was introduced throughclassification of associations-verification by clustering-storing after adjustment.Finally,prototype system was developed.