根据三维谱图识别变压器绝缘典型模型放电的研究

分析了变压器绝缘的主要放电形式，设计了模拟变压器放电的４种试验模型，进行了不同情况下模型的放电试验。使用数字化测量装置，在双层屏蔽试验室内，取得了各种模型放电的放电量－相位信息。采用三维谱图提取放电指纹特征，并用人工神经网络业识别不同的放电类型，分析了不同数据处理方式对放电识别的影响，研究结果表明，人工神经网络对油纸变压器绝缘放电有足够识别能力。

Discharge recognition of typical insulation models of transformers based on 3D pattern chart

The main discharge types in insulation of transformers are analysed, 4 kinds of experimental models simulating discharges in transformers are designed and model experiments under some circumstances are performed. Using digital measuring device, in 2-coat shielding room, quantity-phase information of discharge current of 4 models are obtained. The feature of discharge is extracted using the 3D pattern chart and the artificial neural networks is used to recognize the discharge models. The influence of the different data processing methods are investigated. The investigation shows that ANN has enough ability to recognize different types of discharge of oil-paper insulation in transformers.