Controlled Switching Strategy for Unloaded Transformers Using Double-Break Vacuum Circuit Breakers With Grading Capacitors

Abstract

In double-break vacuum circuit breakers, postarc plasma extinction, dielectric recovery, and grading capacitors significantly influence the electrical boundary conditions on the transformer side and the resulting magnetizing inrush current. Furthermore, the dynamic variation of residual flux after opening is coupled with mechanical scatter. Consequently, complex stochastic fluctuations are caused in the closing inrush current. To address these issues and avoid the limitations of traditional fixed-peak switching, a robust optimization control algorithm is proposed. A dynamic Jiles–Atherton (J–A) hysteresis model is integrated with a physics-informed neural network (PINN) in this algorithm. First, the transformer terminal voltage after opening is theoretically derived under the influence of grading capacitors. The mechanism of its impact on magnetic flux and inrush current is also analyzed. Subsequently, the proposed control strategy is validated under the condition of mechanical scatter. Statistical results indicate that the standard deviation of the residual flux prediction error is reduced by approximately 8.5%. Even under significant mechanical scatter, the peak inrush current is attenuated by approximately 97%. The inrush current remains close to the nominal no-load excitation level, and the risk of core saturation is substantially mitigated.