High Accuracy Fault Identification for High-Voltage Circuit Breakers using CGAN-1DCNN

High voltage vacuum circuit breakers are crucial for modern power systems in that they ensure the grid's stability. However, they could wear out and eventually fail due to their intricate design, operational, contact or actuator issues. This paper proposes a method to identify the faults for high voltage vacuum circuit breakers using Conditional Generative Adversarial Network-One-Dimensional Convolutional Neural Network (CGAN-1DCNN). The operational principle of the high-voltage vacuum circuit breaker was examined and a platform constructed to simulate fault scenarios. The data sets encompassed a range of fault types, including normal operation, trip upon closure, coil burnout, and others. CGAN was used to generate additional fault data, enhancing the model's generalization ability. 1DCNN was used to train deep learning on acquired data, diagnosing faults efficiently. The proposed method achieved relatively high precision of over 98% for fault identification compared to 86.30-96.34% of the traditional methods, which guarantee safe operation for the power systems.