Box-plot-SA-BP:变压器DGA多参量故障诊断模型

Box-plot-SA-BP: Fault Diagnostic Model for Multi-feature Transformer DGA

油中溶解气体分析(DGA)方法是一种典型的充油电力设备故障诊断方法,广泛应用于电力变压器故障检测与状态评估,但由于样本数据的可靠性和诊断模型的有效性影响,导致DGA诊断方法准确率较低。文中提出了一种Box-plot-SA-BP模型,首先,采用Box-plot数据检测法去除异常数据以解决数据质量的问题,然后,利用自注意力机制(Self-attention, SA)准确捕捉多参量样本数据间的联系,提取更加稳定可靠的特征,最后设计BP网络多分类模型实现变压器故障诊断。对比实验证明了Box-plot-SA-BP模型的良好性能,具有较高的应用价值。

Dissolved gas analysis(DGA)is a typical method for fault diagnosis of oil-filled electrical equipment and widely used for detecting faults and evaluating the condition of power transformers.However,the accuracy of the DGA diagnostic method is often affected by the reliability of sample data and the effectiveness of the diagnostic model,leading to lower accuracy.In this paper,a box-plot-SA-BP model is proposed to address these issues.Firstly,the box-plot data detection method is applied to remove abnormal data and improve data quality.Then,the self-attention mechanism is utilized to accurately capture the correlations among multi-parameter sample data,extracting more stable and reliable features.Finally,a BP neural network multi-classification model is designed to achieve transformer fault diagnosis.Comparative experiments demonstrate the good performance of the box-plot-SA-BP model,which shows high practical value in transformer fault diagnosis.