Fault Diagnosis Method for Open-Circuit Inverter of Three-Phase Voltage Source Type in UPS System Based on 1D-LKCNN and Multi-Sensor Information Fusion

To address the problem of low fault diagnosis accuracy, this article proposes a multi-sensor information fusion-based open circuit (OC) fault diagnosis method for three-phase voltage source inverters. First, three-phase current signals were simulated and collected under healthy condition, single switch OC, and double switch OC faults. The three-phase current data was parallel concatenated after simple preprocessing to complete data-level fusion. Then, one-dimensional convolutional neural network with large kernel (1D-LKCNN) was used for feature extraction and classification to train the model. Finally, the test set data was inputted into the trained 1D-LKCNN to diagnose the OC fault. Simulation results verified that the proposed method performs better than other algorithms, with the ability to identify faults with 100% accuracy and good anti-noise performance.