中文摘要：

针对小电流系统发生故障时，各线路零序电流的非平稳、非线性等复杂特性，给出一种基于经验模态分解（empirical mode decomposition，EMD）和Duffing振子检测的方法。EMD方法具有很好的自适应性，而Duffing振子系统对于与内驱动力同频的外界信号具有高度敏感性。首先对故障零序电流进行EMD分解，提取出第1个内禀模态函数作为故障零序电流特征量；然后提出特征量中的五次谐波分量，将其输入到Duffing振子中，通过系统相图变化来确定 当前线路是否发生故障。对各条线路进行不同条件下故障模拟，仿真结果证明了该算法的可靠性。

英文摘要：

Aiming at the characteristics of the un-effectively grounded systems that the zero-sequence current is non-stationary and nonlinear, the approach based on empirical mode decomposition （EMD） and Duffing oscillator detection method was given. EMD method has good adaptivity, and the Duffing system presents a high degree of sensitivity when the external signal＇s frequency is the same as that of the driving force. Firstly, the zero-sequence current was decomposed and the first intrinstic mode function was extracted as the amount of the zero-sequence current fault. Then the fifth harmonic component was extracted and put into the Dulling oscillator detecting system. It could be determined by the changes in the system＇s phase diagram whether the current line is in fault. Fault simulations in different conditions were given for every line and the experimental results indicated that the method is reliable.