EL-CSO-NN在线监测变压器故障预测

打开文本图片集

中图分类号：TB9 文献标志码：A

文章编号：1674-5124（2025）07-0179-10

Abstract： In the safe operation of the power grid， the safe operation of power transformers plays a very important role，and the early prediction of their failures can avoid serious failures and reduce economic and human losses. In this paper， we predict and diagnose transformer faults based on smallsample dissolved gas analysis （DGA）data. First， an online monitor is used to acquire DGA data， followed by ensemble leaming （EL） using the Bagging algorithm with bootstrap resampling to process the smallsample data， and finally，a crosssectional optimization neural network （CSO-NN） is used for short-term prediction of DGA data to achieve the objective of transformer fault prediction. The method is applied in acase，and the case study demonstrates that the EL-CSO-NN algorithm proposed in this paper can achieve good transformer fault prediction and classification.

Keywords： online monitoring; DGA monitoring; ensemble learning; CSO-NN; fault prediction

0 引言

电力变压器在电网的安全运行中起着十分关键的作用，一旦其发生故障，通常会导致大面积停电等问题，并带来较大的经济损失[1-2]。（剩余12939字）