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

作者：Li Cui; Liu Zhenxing^*; Chai Li; Xu Bo; Zhang Botao

来源：Proceedings of the CSEE, 2017, 37(23): 6995-7003.

DOI：10.13334/j.0258-8013.pcsee.162355

• Fault recognition
• Modular multi-level converters
• Open circuit faults
• Spectral clustering
• Total least squares

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.

• 出版日期2017
• 单位武汉科技大学