偏最小二乘回归法在介损在线监测误差分析中的应用

摘要

On-line monitoring of dielectric loss is an important means for evaluating the capacitive equipment'insula?tion condition.Because the loss value is only a few per thousand relatively under normal conditions,it is often vulnera?ble to fluctuations caused by various factors,resulting in inaccurate measurement.To solve this problem,the error fac?tors in the online monitoring of dielectric loss are analyzd in this paper,and the correlation analysis of error factors is performed based on partial least-squares regression.Moreover,the correlation coefficients between error factors and di?electric loss are calculated.Through the correlation graph between error factors and dielectric loss,it is found that the factors affecting the online monitoring can be arranged in a descending order,i.e.,voltage,humidity and temperature. These results are consistent with those by using gray correlation theory,demonstrating its practicality and effectiveness.%在线监测介质损耗因素(简称介损)是判断电容型设备绝缘状况的重要手段.由于介损在正常时只有千分之几,往往容易受到各种因素的影响而波动,造成测量结果不准确.针对这个问题,分析了介损在线监测的误差因素,采用偏最小二乘回归法对各误差因素进行相关性分析,计算各误差因素与介损的相关系数.通过误差因素与介损之间的关系图,得出影响介损在线监测因素由大至小依次为电压、湿度和温度,与灰关联结果比对,结论一致,验证了该方法的实用性和有效性.