我国城乡在3~66kV的配电网中，主变中性点一般采取不接地或经消弧线圈接地的运行方式。这类电网在发生单相接地故障时短路电流只能通过对地电容或阻抗形成小电流回路，故称为小电流接地系统。采用小电流接地系统最大的优点是当系统发生单相接地故障时，并不破坏系统电压的对称性，且故障电流较小，对供电设备不致造成很大的危害。规程允许故障线路或设备继续运行1~2小时，不影响正常供电。单相接地故障如果不做及时的处理，由于非故障相的两相对地电压升高，可能引起绝缘薄弱环节被击穿，发展成为相间短路，使事故扩大，影响用户的正常用电；同时，弧光接地还会引起全系统过电压，进而损坏设备，破坏系统安全运行，甚至发展为永久性相间故障。所以当发生单相接地故障时，必须正确及时地把故障线路检测出来并加以切除，将危害减到最低。现有的选线法又可以分为稳态选线法和暂态信号法。但稳态选线法有稳态分量存在时间长但较微小，不易提取，给选线造成困难等缺陷。 论文在分析小电流接地系统故障选线的现状及传统五次谐波选线原理缺点的基础上，分析中性点非有效接地系统的单相接地故障的稳态特性机理和混沌微弱正弦信号的检测原理，研究小电流单相接地的混沌Duffing振子检测模型，并将稳态零序电流的五次谐波作为故障特征量。将故障后的零序电流全波信号，作为周期扰动加入到Duffing振子混沌模型，并将该模型调至混沌临界状态，利用故障线路与正常线路的零序电流中的五次谐波方向相反使混沌系统发生相变后的状态不同来判别故障线路。采用符号序列信息熵法计算出混沌x时域波形的概率分布，从而辨识系统处于混沌状态还是大周期状态，选出故障线路。应用MATLAB软件进行仿真模拟实验，验证该方法的可靠性。

The neutral point of main transformer in our nation usually adopts NUS or NRS for the 3~66kV distribution network. The short-circuit fault current can form a small current loop only through the ground capacitance or impedance when a single-phase ground occurred in such grid, so it is called the small current grounding system.When a single-phase ground fault occurs, the biggest advantages of using small current grounding systems are that the symmetry of the system voltage would not be destroyed and the fault current is too small to cause a great harm for the power supply equipment. The electricity regulation allows fault lines or equipment continues to run 1 to 2 hours and does not affect the normal power supply. If the single-phase grounded fault would not be timely processed, it may cause insulation weaknesses which would be breakdown to become the phase short-circuit, and affect the user's normal use of electricity since the voltage of non-fault phase increases. At the same time, the arc grounding will cause the over-voltage of whole system, and damage the equipment and the safty of system operation, even become a permanent phase-to-phase fault. Therefore, the fault line must be correctly and timely detected to minimize the damage when the single-phase ground fault occurs. The existing methods of fault line selection can be divided into steady-state fault line selection and transient signals fault line selection. Steady-state signals last for a long time, but it is too tiny to be extracted, and creates obstacles to the fault line selection. The present situation of the fault line selection method in the small current system and the disadavantages of the traditional fifth harmonic selection principle is deeply researched in this paper. The steady-state characteristics of the single-phase ground fault in the neutral non-effectively grounded system and the weak signal detection principle based on the chaotic Duffing model are respectively analyzed. The small current single-phase ground model and chaotic Duffing oscillator detection model are analysed, and the fifth harmonic wave of zero-sequence current in the state-state is regarded as the fault characteristic physical quantities. The full-wave signal of fault zero sequences current is took as the periodic disturbance, which is added into the Duffing oscillator model. Then the Duffing system is made into the critical chaotic state, and the different direction of the fifth harmonic of zero-sequence current in fault line and the normal line to let this Duffing model in different state are used to find the fault line. The probability distribution of the time-domain waveform from Chaos’ x wave is caculated by the symbol sequence information entropy method, so that whether the Duffing system is in chaotic state or cycle state is identified to select a fault line. At last, this paper uses Matlab software to simulate the application of experiments to verify the reliability of this method.