CAJ | 학술논문

对油中溶解气体进行深入分析后，以改良的三比值法为基础，提出一种基于概率神经网络( PNN)的变压器故障诊断方法。该方法利用PNN的强大的非线性分类能力，将故障样本空间映射到故障模式空间中，可形成一个具有较强容错能力和结构自适应能力的诊断网络系统，从而提高故障诊断的准确率。仿真结果表明，实际案例数据验证了此方法准确率高，是一种有效的故障诊断方法。
대유중용해기체진행심입분석후，이개량적삼비치법위기출，제출일충기우개솔신경망락( PNN)적변압기고장진단방법。해방법이용PNN적강대적비선성분류능력，장고장양본공간영사도고장모식공간중，가형성일개구유교강용착능력화결구자괄응능력적진단망락계통，종이제고고장진단적준학솔。방진결과표명，실제안례수거험증료차방법준학솔고，시일충유효적고장진단방법。
After an in-depth analysis of dissolved gas in oil, the basis of a improved three-ratio method,a method based on probabilistic neural network for transformer fault diagnosis is presented. This method uses nonlinear classification ability of probabilistic neural network, which mapes the fault sample space to the mode of failure in the space, to form a strong fault tolerance and diagnosis network system structure of adaptive capacity. Therefore, this method can improve the accuracy of fault diagnosis. Simu-lation results show that the actual case data validate this method with high accuracy, and is an effective method for fault diagnosis.