基于DBN-IFCM的变压器故障诊断方法

Transformer Fault Diagnosis Method Based on Deep Belief Network and Improved Fuzzy C-means Clustering

针对变压器故障边界划分的模糊性和传统模糊方法对变压故障诊断准确率低等问题,提出了一种基于深度信念网络(deep belief network, DBN)和改进的模糊C均值聚类(improved fuzzy C-means clustering, IFCM)的变压器故障诊断方法。该方法首先对故障数据进行归一化处理,然后利用深度信念网络对故障数据进行特征提取,最后利用改进的模糊C均值聚类对提取的特征进行聚类,以达到故障分类的目的。仿真实验表明:相较已有的变压器故障诊断方法,所提方法具有较高的诊断准确率,其准确率为93.3%,能够较为精准地识别变压器的各种故障。

Aiming at the fuzzy classification of transformer fault boundary and the low accuracy of transformer fault diagnosis by traditional fuzzy method, we propose a transformer fault diagnosis method based on deep belief network and improved fuzzy C-means clustering(DBN-IFCM). In this method,firstly,the fault data are normalized;then the features of the fault data are extracted with the deep belief network;finally,the improved fuzzy c-means clustering is used to cluster the extracted fault features to achieve the purpose of fault classification. Simulation experiments show that, compared with the existing transformer fault diagnosis methods, the proposed method has a higher diagnostic accuracy rate, the accuracy rate is 93.3%, which can more accurately identify various faults of the transformer.