Transient Faulty Feeder Selection Tecniques For Non-Effectively Earthed Network Based On Prony Algorithm

In China, non-effectively earthed neutral is widely used in medium voltage distribution networks. When a single-phase earthed fault occured in non-effectively earthed network (usually called single-phase earthed fault for short), the arc self-distinguishing rate is high and the power supply can be maintained for about one or two hours for a permanent fault. Hence the power supply reliability of non-effectively earthed system is very high. However, the weak fault current and intermittent arcing in some faults makes it very difficult in selecting the faulty feeder. Up to the present, many utilities have to select the faulty feeder by manually switching off feeders due to lack of reliable selection techniques, which will result in unnecessary interruption in healthy feeders. With the consumer's increasing demand for power supply reliability and quality, the utilities pay more attention to fault detection techniques. In resent years, significant progress has been made on faulty feeder selection techniques thanks to the efforts of power system academics and engineers. Several kinds of faulty feeder selectinon equipments are invented and proved to have better reliability by the fault trial results. However, the success rate of these new faulty feeder selection equipments are less than 90 percent, therefore further improvement is needed.At present the conventional faulty feeder selection methods can be classified into two kinds in terms of the way of signal utilizing, active methods and passive methods. The former methods mainly include method of signal injection, residue current method, and shunt resistor method and so on. The disadvantages of the active methods are that they all require adding signal injection devices, which is complicated and mighe affect safe operation of the primary system. The passive methods can be sorted into two types: detection methods based on steady-state signals or detection methods based on transient signals. Steady-state methods include fault current amplitude and phase comparing, fault current direction method, harmonic method and so on. The main problem existed in steady-state detection method is that the fault current is too small to be detected. Moreover, the successful selection relies on the sustained earthing at the fault, where the arcing will affect the incorrect selection of the faulty feeder. Hence steady-state detection methods are all invalid in detecting the arcing faults in neutral isolated system and all kinds of faults in Peterson coil earthed system. By contrast, when selecting faulty feeder using transient signals, the influence of arc suppression coil can be neglected since the frequency of transient signal is very high. It is also invalid in case of the intermittent arc earthed fault. In addition, the detection sensitivity is very high as the amplitude of transient signal is much larger than that of steady-state signal. Detection methods using transient signals include the first half-wave method, transient zero modal selected frequency band (SFB) method (including comparing amplitude and phase of zero modal transient currents in SFB between faulty feeder and healthy feeder, detecting the reactive power direction of zero modal transient signals in SFB), energy method, Wavelet transformation method and so on. Particularly, transient zero modal SFB method is a great breakthrough in using transient signal to select faulty feeder. It is simple and can be easily realized. The selection equipments based on transient zero modal SFB method have been widely used. The shortcoming of the method is that the signals must be filtered before performing in order to get the transient signals in SFB. Thus the dominant transient frequency (DTF for short) component of transient signal will be significately attenuated when it is low, which may result in incorrect faulty feeder selection. Therefore, it is necessary to continue the research work on fault detection method based on transient signals.A novel detection principle based on Prony algorithm is presented in the paper, which determines the faulty feeder by comparing the amplitude and phase of DTF admittanceof faulty feeder and healthy feeder. Details of the research works are as follows:1. The transient signal characteristic of single-phase earthed fault in non-effectively earthed system is analyzed accurately to some extent based on approximate fault modal network, and the varying trend of transient zero modal voltage and current when earthed fault with transition resistance is summarized. That energy of DTF signal is the chief component of transient energy is proved. The DTF signal can be used to select faulty feeder and has high detection sensitivity is demonstrated too.2. The principle of Prony algorithm is introduced and the rule of choosing parameters when fitting transient signals generated by single-phase earthed fault based on Prony algorithm is presented. The problem of how to determine sampling frequency, sampling time and model order is discussed by certain examples.3. Transient zero modal voltage of bus bar, transient zero modal current in faulty feeder and healthy feeder generated by single-phase earthed fault are decomposed into exponentially different frequency component by using Prony algorithm. The common principle of how to determine DTF is presented too. The feasibility and validity of Prony algorithm in analyzing transient signals generated by single-phase earthed fault is verified too.4. The phasor of zero modal current and voltage at DTF are defined, and the ratio of them is defined as DTF admittance. It is proved that DTF admittance is approximately equal to the corresponding admittance of sinusoidal frequency at steady-state. When short circuit in positive direction occurred, the amplitude of DTF admittance is large and the phase is close to -90°, whereas when short circuit in negative direction occurred, the amplitude of DTF admittance is small and the phase is close to 90°.5. The faulty feeder selection principle and realization based on Prony algorithm is presented, which determines the faulty feeder by comparing the amplitude and phase of DTF admittance of faulty feeder and healthy feeder. When the admittance amplitude of certain feeder is larger than set admittance and its phase is close to -90°, and then this feeder is selected as faulty feeder. When the admittance amplitude of certain feeder is less than set admittance and its phase is close to 90°, and then this feeder is selected as healthy feeder.6. Influence factors considered when set faulty feeder selection criterions are discussed. The set value of admittance magnitude must consider the influence of arc suppression coil when DTF of system is low, that is, inductive admittance of arc suppression coil must be added to admittance set value. Some secondary factors are neglected when deduce the phase of DTF admittance and the phase error of short feeder is big when length difference among all feeders connected to same bus bar is large, so the set value of admittance phase can not be exact 90°, but an operation angle range.7. Vast simulation and Prony fitting of transient zero modal voltage of bus bar, transient zero modal current in faulty feeder and healthy feeder are performed by building ATP simulation model and writing Matlab programs. The novel faulty feeder selection principle is verified by ATP simulation data and field measured fault data respectively.A single-phase earthed fault detection method based on comparing amplitude and phase of DTF admittance is presented. The DTF admittance is calculated based on DTF voltage and current signals fitted from transient signals by Prony algorithm. The method is based on the identification of the frequency domain characteristic parameter of transient signals. The criteria used are clear and therefore have better reliability and stability than conventional techniques directly utilizing the time domain transient signals.The setting of faulty feeder selection criterion can be easily determined and the performance of presented method is also easy to be analyzed and evaluated as the admittance has clear physical meaning. Moreover, only zero modal voltage of bus bar and zero modal current of the feeder itself are used to select the faulty feeder, hence the method is of self-maintained. Because it does not require filtering to obtain transient signals in SFB, the DTF admittance can also be correctly extracted even when its frequency is very low. Hence, it has better sensitivity and reliabilityfor high resistance fault. The two presented methods are significant improvements to the existing transient techniques of faulty feeder selection methods. Their application will enhance the success rate of faulty feeder selection and will improve power supply reliability and quality. They will also help to promote the developments of transient faulty feeder selection techniques.