Research On Line Detection And Fault Location For Single-Phase-to-Earth Fault In Neutral Ineffectively Grounded Systems

The neutral ineffectively grounded system represents a typical scheme in the 60kV and the following power distribution networks for our country. If a single-phase-to-earth fault happened, the advantages of this kind of system are the symmetrical line voltage and small fault current, so it may not probably excise the fault immediately and enhance the power supply reliability. However, in order to guarantee safe operation of system, it is necessary to distinguish the fault line and definite the fault location as soon as possible, which is the fault line detection and location questions for single-phase-to-earth fault.The neutral ineffectively grounded system is also called small current grounded system. When single-phase-to-earth fault happened, because of its weak fault characteristics, the accuracy of the fault line detection and location has been a difficult problem in power system protection and control domain.After deeply researching the characteristics of single-phase-to-earth fault in neutral ineffectively grounded system, systematically comparing the advantages and disadvantages of existed methods about the fault line detection and location, this paper presents a new method of fault line selection and location by obtaining the step response after injecting DC current incentive. When a single-phase-to-earth fault occurred, a DC current signal has been injected into the system and its step response has been obtained. By analyzing the spectral characteristic of this step response, we can find that the spectrum of fault lines and non-fault lines shows a great difference: at one frequency place nearby the resonance frequency, the peak amplitude of fault line is bigger than that of all non-fault lines and the phase is opposite. Based on this characteristic, it presents a new criterion in fault line detection. At the same time, the information of line parameter and the transition resistance has been contained in the peak amplitude and its corresponding frequency. This is a frequency nearby the line resonance frequency and has relation to the response attenuation time constant. There is a mapping relations existing between the fault distance and frequency response peak value besides its corresponding frequency, which may construct the fault location criterion. Based on this criterion, this paper presents a fault location scheme based on BP neural network. In order to fully study the influence on complex characteristics of the composite load on the above fault line detection and location result, this paper also presents a parallel difference equations load model based on Elman neural network. Using the ATP simulation tool, a large number of simulations has been carried on to test this method. The results show that, the influences of extinguishing coil compensation, transition resistance and load dissymmetry as well as load characteristic on this method is very small. This method has good usability and compatible on fault line detection and location.