Research on classification of partial discharge of switchgear cabinets based on a novel association rule algorithm

In order to assess switchgear insulation status, a novel association rule mining (ARM) algorithm is presented. It is used to recognize the severity of switchgear cabinet partial discharge. The algorithm uses fuzzy C-means clustering (FCM) to divide partial discharge feature interval, candidate sets meeting minimum support and minimum confidence are sought based on an improved Apriori algorithm. Multiple recursions and scans are performed on candidate sets to generate association rules library for classification. Fuzzy reasoning based on association rules are performed over multiple needle corona partial discharge signals sampled in 10KV switchgear cabinets. The results show that partial discharge classification rate using association rules is high and classification conclusions are accurate. It has provided theoretical basis and practical value for insulation status assessment of switchgear cabinets.