模块化多电平换流器的子模块开路故障检测方法  

An Open Circuit Fault Detection Method for Sub-modules in Modular Multi-level Converters

作　　者：李翠[1] 刘振兴[1] 柴利[1] 徐波[1] 张波涛[1] 

机构地区：[1]武汉科技大学,湖北省武汉市430081

出　　处：《中国电机工程学报》2017年第23期6995-7003,共9页Proceedings of the CSEE

基　　金：国家自然科学基金项目(61174107;61471275)~~

摘　　要：针对模块化多电平换流器(modular multi-level converter,MMC)子模块开路故障特点,提出一种基于无监督学习-最小二乘互信息谱聚类和整体最小二乘支持向量机(total least square support vector machines,TLS-SVM)的故障诊断方法,前者用于故障特征信息提取,后者用于故障分类识别。在MATLAB/Simulink环境下,搭建可进行故障设置的201电平MMC仿真系统。对采集到的换流器正常和故障运行时的三相电流信号通过滤波去噪处理后,进行Hilbert包络分解得到包络均值,使用最小二乘互信息谱聚类对包络均值进行二分类并获得标签集,然后将标签集和数据集作为基于整体最小二乘支持向量机的训练集并获得分类模型,最后对MMC故障进行分类和识别。仿真实验结果表明,该方法能有效识别高电平MMC的开路故障,并能实现智能决策。Aiming at sub-module open circuit fault characteristics of modular multi-level converters （MMC）, a diagnostic method was proposed based on unsupervised learning-least squares mutual information of spectral clustering and the total least squares support vector machine. The former was used for the failure feature information extraction and the latter was applied to the fault classification and recognition. A 201-level MMC simulation system, which enables to set faults, was constructed in MATLAB/Simulink. Three phase current signals of converters under the normal and fault operation conditions were collected, then after three phase current Signals through the filter denoising processing, the Hilbert envelope decomposition method was used for the signals. Subsequently the mean envelope was obtained. Using the least squares average mutual information of spectral clustering on the mean envelope, and the tag set was obtained, then put the tag set and data set as a least squares support vector machine （SVM） based on the whole training set and get the classification model, at last. The MMC faults can be classified and recognized through the total least squares support Vector machine. The simulation results confirm that the proposed methods can efficiently recognize the open failures of MMC high levels and realize intelligent decision.

关 键 词：模块化多电平换流器 整体最小二乘支持向量机 故障识别 谱聚类 子模块开路故障 

分 类 号：TM46[电气工程—电器]