改进加权模糊C均值算法的风电变压器故障诊断

Fault diagnosis of wind power transformer based on improved feature-weighted fuzzy C-means algorithm

风电变压器作为风力发电系统中电能传输和转换的枢纽设备之一,其安全稳定运行是电网可靠性的重要保障。为提高变压器故障诊断准确度,提出了基于改进加权模糊C均值(FCM)算法的风电变压器故障诊断方法。相比于一般的加权FCM算法,特征加权FCM算法使用固定的特征权重,该方法通过在训练阶段动态更新特征权重,使得不同比重的特征能够体现样本各维度在聚类效果中所起作用的大小。建立基于改进加权FCM算法的风电变压器故障诊断模型,充分考虑了不同比重特征对聚类结果的不同影响,能有效改善复杂数据集的聚类性能。实例研究结果表明:该方法有效地提高了故障诊断的准确率,弥补了传统FCM固定权重分配的不足。

As the main devices for power transmission and conversion in wind power generation systems,the wind power transformers supply safe and stable running,which is the important guarantee for grid reliability.To improve the accuracy of transformer fault diagnosis,a fault diagnosis method for wind power transformers is proposed based on the improved feature-weighted fuzzy C-means(IWFCM)algorithm.Compared with the general feature-weighted fuzzy C-means(WFCM)algorithm,which uses the fixed feature weights,the present method dynamically updates the feature weights during the training phase.Thus,the characteristics of different specific gravities can reflect the role that each dimension of the sample plays in the clustering effect.The fault diagnosis model for wind power transformers based on the IWFCM fully considers the different effects of different features on the clustering results.It can effectively improve the clustering performance of the complex data sets.The case shows that the method can effectively improve the accuracy of fault diagnosis and make up for the deficiency of the traditional FCM fixed weight distribution.