基于自适应FP_Growth算法的电能表故障分析

2013～2018年,天津市电力公司共拆回14×104台故障电能表,并构建了电能表故障信息库。对信息库进行数据挖掘,对提高电力行业服务能力、保障用户用电稳定性具有重要意义。使用FP_Growth算法对故障电能表故障数据库进行关联性分析。针对FP_Growth算法需要人为设定阈值的缺点,引入自适应的方法进行阈值设定,按照权重调整支持度。对电能表故障类型与故障影响因素进行关联性分析,得出结论:13版电能表在多方面性能上有所改进,可考虑逐渐替代09版电能表;时钟电池电压低与时钟故障组合是电能表的最高发故障。因此,电能表电池性能仍是需要改进的主要方向。

Fault Analysis of Watt-Hour Meters Based on Adaptive FP_Growth Algorithm

Tianjin Electric Power Company recalled 140,000 faulty watt-hour meters from 2013 to 2018 and constructed a fault database. Data mining of this database was of great significance to improve the service level and stability of power industry. In this paper, FP_Growth algorithm was used to analyze the correlation of the fault database of the fault watt-hour meter. Considering that FP_Growth algorithm need to set threshold artificially, an adaptive method was introduced to set the threshold and the support degree was adjusted according to weight. Analyzing the correlation of the fault types and the influencing factors, it was concluded that most of the fault watt-hour meters were version 09, while version 13 were relatively few. The low voltage fault of clock battery and the fault of clock were the most frequent fault of the watt-hour meters. Obviously, the quality of version 13 watt-hour meter is obviously better than that of version 09, while the battery performance of watt-hour meter is still the main direction to be improved.