基于动态聚类的电力变压器故障诊断

Fault diagnosis of power transformer using dynamic clustering algorithm

本文提出了一种新电力变压器故障诊断的动态聚类方法,以人工免疫网络对故障样本进行免疫学习和记忆,提取表征故障样本的有用特征作为核可能性聚类算法的初始聚类中心,再用遗传算法动态选取聚类个数和中心实现故障样本的分类。该诊断方法经大量实例分析,并将其结果与BP神经网络等方法的结果相比,表明该算法具有较高的诊断精度。

A novel dynamic clustering algorithm for power transformer fault diagnosis is proposed. Firstly artificial immune network is used to carry out immune memory and learning of the fault sample ; the useful characteristics that effectively represent the fault samples are extracted and used as the initial clustering centers of kernel-based possibilistic clustering algorithm. Then genetic algorithm is used to dynamic optimize and select the number and centers of clustering to achieve the classification of the fault sanaples. A lot of fault samples were analyzed by this algorithm, and the results were compared with those obtained by BPNN. Diagnosis results indicate that samples are effectively classified using the proposed algorithm and the fault diagnosis precision is improved.