中文摘要：

基于励磁涌流波形畸变严重,会呈现出尖顶波的凹弧特征,而故障电流则基本保持基频正弦波（凸弧）特征的思想,该文提出了一种利用波形特征快速区分变压器励磁涌流和内部故障的新方法.该方法利用初始5ms内差电流的数据（5ms数据窗）,将其拓展为长10ms的数据窗,然后通过最小二乘算法拟合差电流波形得到其近似表达式,再根据拟合曲线在各点处的曲率和凸凹特性来区分变压器励磁涌流和内部故障.动模实验表明：该方法能够快速地切除变压器内部故障,动作时间约为5ms,所需时间较短,识别灵敏度高,且实现方便,不受电流互感器饱和及非周期分量的影响.

英文摘要：

A new algorithm based on waveform characteristics is presented for fast discrimination between inrush current and fault current of power transformers. The method is based on that the waveform of inrush current （concave） distorts seriously, while the fault current nearly keeps sinusoid （convexity）. Utilizing the sample data within 5ms after fault initiation and extending the date windows from 5ms to 10ms, the approximate current formula can be extracted from fault current or inrush current using least square fitting method. According to the curvature and convex-concave characteristic of the fitting curve, inrush current and fault current of power transformers can be distinguished. Experiment results indicate that this algorithm is not only effective under various fault conditions but also can be realized easily and hardly to be affected by CT saturation and decaying DC components. Moreover, fault clearance time can be limited within 5ms.