A magnetization hysteresis-based power transformer fault detection algorithm

A novel power transformer detection algorithm is proposed based on the shape of magnetization hysteresis curve. And the magnetization hysteresis is described by a Voltage-Differential Current (V-DCurr) curve in this paper. Firstly, the characteristics of V-DCurr curve are studied. The V-DCurr curve is a symmetrical ellipse for a healthy transformer. For an energizing transformer, the V-DCurr curve is asymmetric. The curve is also an ellipse but with different parameter when an internal fault occurs. Secondly, an image processing method, chain code, is chosen to describe the characteristics of V-DCurr curve in order to detect internal fault from inrush current. Thirdly, simulations based on PSCAD are conducted to verify the proposed algorithm. The results show that the fault can be detected and isolated reliably and rapidly.