Deep Convolution Neural Network Based Fault Detection and Identification for Modular Multilevel Converters

• Published in: 2018 IEEE Power & Energy Society General Meeting (PESGM) pp. 1 - 5
• Main Authors: Qu, Xiangshuai, Duan, Bin, Yin, Qiaoxuan, Shen, Mengjun, Yan, Yinxin
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.08.2018
• Subjects: "Data-bands"; Capacitors; Circuit faults; Convolution; Deep Convolution Neural Network (DCNN); Fault detection; Fault diagnosis; Feature extraction; Modular Multilevel Converter (MMC); Training
• ISSN: 1944-9933
• DOI: 10.1109/PESGM.2018.8586661

This paper proposed a novel fault detection and identification (FDI) method for modular multilevel converters based on deep convolution neural network (DCNN). According to the failure characteristics of the SMs in the MMC, the voltage signals of the capacitors in all SMs are combined as a multichannel sequence and massive "data-bands" are sampled from the sequence and normalized which are used as the input of the proposed model subsequently. So the FDI in power system can transform into image recognition problem. Then high-level features of the data can be learned automatically through DCNN and determine whether a SM's fault has occurred and the faulty SM's identification. Results show that the proposed method can quickly and accurately achieve fault detection and identification. Compared with some existing methods, the proposed method can obtain state-of-the-art results and has a good application prospect in the online MMC protection.