金属氧化物避雷器缺陷诊断的反距离加权改进KNN算法

针对现行避雷器在线监测系统中缺陷诊断规则不完善而导致大量漏报和误报事件发生的情况，通过分析避雷器三相全电流和阻性电流、三相电压和阻性电流的Pearson相关系数，并考虑环境因素影响，提取环境温湿度、三相阻性电流和三相电压作为避雷器缺陷诊断的特征参数。提出了一种基于反距离加权改进KNN算法的避雷器缺陷诊断方法，通过实例验证所提方法较其他方法具有更优的诊断正确率（97.28%）和泛化能力，为避雷器缺陷诊断提供了新思路。

Inverse distance weighted improved KNN algorithm for defect diagnosis of MOA

The defect diagnosis rules in the current metal oxide arrester on-line monitoring system are still incomplete, resulting in a large number of false alarms and missing alarms. By analyzing the Pearson correlation coefficients of the three-phase full current and resistive current, three-phase voltage and resistive current of the arrester, and considering the influence of environmental factors, the ambient temperature and humidity, three-phase resistive current and three-phase voltage are extracted as the characteristic parameters of the arrester defect diagnosis. An arrester defect diagnosis method based on the inverse distance weighting improved KNN algorithm is proposed, the case analysis is shown that the proposed method has better diagnosis accuracy (97.28%) and generalization ability than other methods, which provides a new idea for the diagnosis of arrester defects.