基于粗糙集与多类支持向量机的电力变压器故障诊断

发布时间：2018-08-29 18:14

【摘要】：针对传统变压器故障诊断过程中未能将部分反映变压器故障状态的信息有效利用,以致故障诊断信息不完备、诊断结果不准确的情况,将铁芯接地电流等信息与特征气体相结合,以完善故障特征信息。并在此基础上,构建了一种采用粗糙集的一对一多类支持向量机故障诊断新方法。首先利用一对一多类支持向量机实现故障类别区域的划分;然后根据粗糙集的上下近似这一核心思想对故障类别划分区域进行描述,得出故障分类的上下近似域及边界域的集合,并提取故障诊断分类规则;最后利用分类规则实现故障类别划分。该方法实现了故障信息的综合利用,并将粗糙集在不完备数据与复杂模式刻画方面所具备的优良表现,及一对一支持向量机在分类方面的良好泛化性能进行有效融合,从而有效提高故障分类精度。变压器故障实例分析表明,与传统诊断方法相比较,该方法具有更高的诊断正确率,且其可有效反映故障诊断中所出现的不完备信息。
[Abstract]:In the process of traditional transformer fault diagnosis, some information reflecting the fault state of transformer can not be used effectively, so that the fault diagnosis information is not complete and the diagnosis result is not accurate. In order to perfect fault characteristic information, the information of core grounding current is combined with characteristic gas. On this basis, a new method of fault diagnosis based on rough set for one to one class support vector machine (SVM) is proposed. First, one to one class support vector machine (SVM) is used to partition the fault category region, and then the fault classification region is described according to the core idea of rough set approximation. The sets of upper and lower approximate domain and boundary domain of fault classification are obtained, and fault diagnosis classification rules are extracted. Finally, fault classification is realized by using classification rules. This method realizes the comprehensive utilization of fault information, and combines the good performance of rough set in incomplete data and complex pattern characterization, and the good generalization performance of one-to-one support vector machine in classification. Thus, the accuracy of fault classification is improved effectively. The analysis of transformer fault example shows that compared with the traditional diagnosis method, this method has higher diagnostic accuracy, and it can effectively reflect the incomplete information in fault diagnosis.
【作者单位】：西南交通大学电气工程学院;
【基金】：国家自然科学基金(51177136)~~
【分类号】：TM41

【相似文献】