Deep Learning-Based Fault Diagnosis for Asymmetric Cascaded H-Bridge Multilevel Inverters Under Dynamic Motor Loads
International Journal of Circuit Theory and Applications, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Basım Tarihi: 2026
- Doi Numarası: 10.1002/cta.70665
- Dergi Adı: International Journal of Circuit Theory and Applications
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Aerospace Database, Applied Science & Technology Source, Compendex, INSPEC, zbMATH, Academic Search Ultimate (EBSCO), Materials Science & Engineering Collection (ProQuest), Technology Collection (ProQuest)
- Anahtar Kelimeler: cascaded H-bridge, CNN-LSTM, deep learning, fault diagnosis, FPGA, induction motor, multilevel inverter
- Van Yüzüncü Yıl Üniversitesi Adresli: Evet
Özet
This paper presents a novel deep learning-based approach for open-circuit switch fault detection in cascaded H-bridge (CHB) multilevel inverters. Unlike most studies that use static loads, we employ a squirrel-cage induction motor to emulate realistic industrial operating conditions. A three-phase 9-level asymmetric CHB inverter prototype is driven by an FPGA-based controller. High-resolution current and voltage data are collected across 50 different speed-torque operating points, comprising 25 classes (1 healthy, 24 faulty). The proposed CNN-LSTM hybrid model achieves an average accuracy of 97.4% under variable load conditions (with raw signals only, as an ablation baseline) and 99.0% on a held-out test set (with hybrid features). A hybrid feature selection method (Mutual Information and Random Forest) reduces computational load, enabling real-time deployment on an embedded platform (Raspberry Pi 5) with 15–20 ms response time. Extensive comparisons, ablation studies, and robustness analyses demonstrate the superiority of our approach over state-of-the-art methods.