Fault Diagnosis Method Based on Adaptive Weight Adjustment and CBR Theory
Abstract
This paper proposes a fault diagnosis method of automatically adaptive weight adjustment case-based reasoning used for aircraft component fault. This method introduces multi-stage weight assignation method and entropy method to correct weight of fault feature. This paper applied actual-measurement fault data to establish multi-type & multi-fault CBR case library; and conducted local and global similarity analyses of fault features of the target case and historical cases. It also combined KNN neighbor algorithm for the weight assignment of local similarity, used multi-stage weight assignment method to confirm weight of fault feature, and then applied entropy method to further correct weight. At last, comparative analysis of the three CBR fault diagnosis methods including method of mean value was made. Simulation experiment suggested the effectiveness and accuracy of the new method.
Keywords
case-based reasoning; entropy method; multi-stage weight assignation; K-Nearest Neighbor algorithm
DOI
10.12783/dtcse/iccae2016/7184
10.12783/dtcse/iccae2016/7184
Refbacks
- There are currently no refbacks.