摘要
为了提高三相无桥功率因数校正(PFC)工作时的可靠性与安全性,分析了其开路故障机 理和故障特征,并针对该变换器故障状态种类较多且状态样本稀疏等问题,提出一种基于近邻算 法的无桥变换器开路故障诊断策略。该策略以三相输入电流作为故障特征量,通过坐标变换和相 似性分析对各状态的已知样本进行优化,采用基于标准化曼哈顿距离的 K 最近邻算法进行故障模 式识别。仿真和实验结果显示,该诊断策略能够实现以较高采样速率对无桥变换器进行在线的实 时诊断,诊断的准确率较高,诊断速度较快,且对负载突变、谐波干扰等特殊状态具有一定鲁棒性。
In order to improve the reliability and safety of three-phase bridgeless PFC, open circuit fault mechanism and fault characteristic were analyzed. Considering the problems of numerous fault states and sparse feature samples, a novel method based on K-nearest neighbor algorithm is proposed for open circuit fault diagnosis and location. Three-phase input currents are taken as the characteristic variables, and the fault features are extracted by coordinate transformation and similarity analysis. Then the K-nearest neighbor algorithm based on normalized Manhattan distance is used to identify different fault modes. The feasibility has been verified by simulation and experimental results. The proposed method implements a high sampling rate for online real-time diagnosis with higher diagnosis accuracy and faster diagnostic time. It is also robust to some special conditions such as load mutation and harmonic interference.
作者
陈文博
程红
王聪
Chen Wenbo;Cheng Hong;Wang Cong(School of Mechanical Electronic&Information Engineering China University of Mining&Technology,Beijing 100083 China)
出处
《电工技术学报》
EI
CSCD
北大核心
2019年第18期3873-3883,共11页
Transactions of China Electrotechnical Society
基金
国家自然科学基金项目资助(51577187)
关键词
三相无桥
PFC
在线故障诊断与定位
K
最近邻算法
半实物仿真
Three-phase bridgeless PFC
online fault diagnosis and location
K-nearest-neighbor (KNN) algorithm
hardware-in-loop simulation