Diagnosis and Pattern Recognition of Open Circuit Faults with Inverter Reconfiguration in Direct Torque Control of Induction Motor Drive Using Neural Networks

Due to frequent faults occurring in an inverter, it will be important to detect faults early and find a way to control operation and take preventive action to avoid recurring interruptions. In this paper, we use neural networks to provide a pattern recognition based approach for handling open circuit switching failures in voltage inverters driven by direct torque. It is thought that AlexNet CNN can identify one or more open-circuit problems. First, the Clarke-Concordia transformation divides the three current lines into two vectors. The next step a suitable mapping is used to encode these two new vectors into pictures. After that, the networks were trained and tested using these acquired images. Lastly, a simulation result shows that the AlexNet model has excellent fault-diagnosing capabilities for open-circuit faults in Source voltage inverters and can attain complete accuracy in a variety of fault scenarios. The simulation findings show that the neural network-based approach outperforms the other for both single and multiple errors in classification..