Fault Diagnosis for Vienna Rectifiers Using Machine Learning

This paper presents a fault diagnosis for Vienna rectifiers using machine learning techniques. In the proposed approach, a convolutional neural network (CNN) is trained using input current waveforms to accurately identify the location of the faulty switch under various operating conditions. The Vienna rectifier is particularly sensitive to open-switch faults due to its use of bidirectional switches and the need to maintain neutral-point stability. The proposed method eliminates the need for additional sensors or hardware by utilizing only the measured three-phase input currents as input features. Feature extraction and classification are performed through multiple convolution and fully connected layers, enabling robust fault detection even under current variation and noise. The effectiveness of the proposed AI-based fault diagnosis is evaluated through simulations.