Real-time Diagnosis and Fault-tolerant Control of Grid-side Current Signal Anomaly in AC Traction System

In order to solve the problem of locomotive failure caused by the fault of grid-side grounding due to misinformation of grid-side current signal in the application of AC locomotive, a real-time diagnosis method for grid-side current signal anomalies based on multi-operating condition nonlinear model is proposed in this paper. Firstly, based on the system mechanism and engineering design experience, the variables strongly related to grid-side current are extracted, and a multi-operating condition nonlinear estimation model of grid-side current is established. Secondly, based on the two redundant grid-side current signals of the traction drive system and the estimated value of the grid-side current, residual and detection statistics were designed, and the effective fault detection and decision-making of the two grid-side current signal anomalies are achieved by combining the CUSUM algorithm. Finally, different fault-tolerant control strategies are adopted according to the fault decision-making results to ensure the normal control and protection functions of the system. Field data tests show that the proposed diagnostic method achieves timely detection and precise positioning of grid-side anomalies 0.3 s after the occurrence of anomalies, avoiding the impact of on-site faults caused by grid-side signal anomalies.