Inverter Open Switch Fault Detection Using BiLSTM Neural Network in Induction Motor Drive

The inverter open switch fault is one of the faults that mostly occurs in electric motor drives. If this type of fault is not detected and tolerated in time, it can lead to performance degradation, high repair costs and even accidents. This paper presents a fault detection and diagnosis (FDD) method to capture single and double open switch/switches faults in a three-phase, two-level inverter fed V/F controlled induction motor. Three deep neural networks (DNNs) classifiers, namely bidirectional long short-term memory (BiLSTM) network, convolutional neural network (CNN), and fully connected neural network (FCN), were trained and tested using three-phase voltage and current rms and angle. The BiLSTM model performed the best with an overall accuracy of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 9. 1 5 \%}$</tex> due to its ability to capture temporal dependencies over time. The proposed FDD method can detect faults and exact faulty switches in less than one fundamental period at different speed levels and during varying speed conditions. The results confirm the performance of the proposed FDD method, making it highly reliable for real-time applications.