基于核Fisher判别分析技术的电力变压器DGA故障诊断模型研究

鉴于核Fisher判别分析技术(KFDA)在模式识别问题中表现出的良好性能,提出了基于KFDA的变压器故障诊断模型,该模型首先提出了区分放电及过热两大类故障的特征量,并用KFDA分类器来识别类内故障的具体类别。采用基于网格搜索的交叉验证法来选择模型参数,避免了参数选择的盲目性和随意性。实例分析表明,该模型具有训练时间短、不存在局部极小等优点,与IEC三比值及改良电协研法相比,具有更好的故障识别效果。

Fault Diagnosis Model of DGA for Power Transformer Based on Kernel Fisher Discriminant Analysis Technology

In view of the good performance in pattern recognition using Kernel Fisher Discriminant Analysis(KFDA) method,this paper presented a fault diagnosis model of power transformer based on KFDA.The model extracted a feature to distinguish between discharge faults and hot faults first,and then used the KFDA classifier to distinguish in discharge faults or hot faults.To choose parameters of the model,this paper adopted the method of cross validation based on grid search,avoiding the arbitrary and capricious behavior.Practical analysis indicates that this model has advantages of short training time and no part teeny problem.Compared with three rations method of IEC and the improved three rations method,this model has considerable effectiveness of faults identification.