摘要
高压断路器是电力系统中重要的一次设备,其故障诊断是实现状态检修的前提。在实际应用中,诊断模型的准确度会受到数据干扰而产生误判,严重影响检修效果。为提高诊断模型的鲁棒性,本文从采样数据角度对故障诊断进行了研究,提出了基于KPCA-SVM的断路器故障稳健诊断方法。利用核主元分析分离正常数据样本空间与故障数据样本空间,加大了训练样本间的差异度;再以支持向量机建立故障诊断模型对断路器主要故障进行诊断,极大提升了诊断模型的抗干扰性能,实验证明取得了较好的效果。
High voltage circuit breaker is an important kind of primary equipment in power system with its fault diagnosis being the premise of condition-based maintenance. In practice, the accuracy of the diagnosis model may be influenced by data disturbance, which leads to wrong judgment. To improve the robustness of the diagnosis model, some researches about the fault diagnosis in terms of sampled data have been done and a robust diagnosis method based on KPCA-SVM is proposed in this paper. Using KPCA to distinguish normal data sample space from failure data sample space greatly increases the difference degree of training data space. Besides, the main faults of the breaker are diagnosed by the fault diagnosis model based on SVM, which improves the anti-interference performance of diagnosis model. Satisfactory effects have been proved by experiments.
出处
《电工技术学报》
EI
CSCD
北大核心
2014年第S1期50-58,共9页
Transactions of China Electrotechnical Society
基金
江苏省科技支撑计划(BE2013883)
江苏省产学研联合创新(BY2014127-13)资助项目
