输电线路故障测距.doc

1、Travelling Wave Based Fault Location for Teed CircuitsCansn Y. Evrenosoglu, Student Member, IEEE, and Ali Abur, Fellow, IEEEAbstractThis paper describes a fault location algorithm for three terminal lines using wavelet transform of the fault initiated transients. The results presented in 1 are exten

2、ded to the case of three terminal configuration and a new single ended procedure is developed for teed circuits. The algorithm gives accurate results for the case of three terminal lines including series compensated branch, mutual coupled line section and different values of fault resistances. The p

3、erformance of the algorithm is tested on different scenarios by using ATP/EMTP program and MATLAB Wavelet Toolbox.Index TermsElectromagnetic transients simulations, fault location, frequency dependent line model, mutual coupled lines, series compensation, teed circuits, wavelet transform.I. INTRODUC

4、TIONDue to the superimposed reflections of the fault signal from the T-node and the fault point, fault location in teed circuits (Fig. 1) presents unique challenges. In addition to the impedance based fault location techniques, there are various types of fault location methods proposed for teed circ

5、uits using either phasor based or travelling wave based models.The post fault differential currents from each terminal is used in 2 in order to locate the fault in multi terminal transmission lines. A fault location technique using the pre-fault load flow for phase alignment is described in 3 by uti

6、lizing the multi end phasor measurements in order to determine the fault location.The post fault synchronized phasor measurements are used to solve the system differential equations for fault location in multi terminal lines in 4. The use of negative sequence multi-ended measurements for fault locat

7、ion in three terminal lines is proposed in 5. Recently a new fault locator for three terminal lines is described in 6 using phasor measurement units in order to solve the travelling wave differential equations.Travelling wave based fault location for transmission lines is initially formulated in 7 b

8、y defining a discriminator that combines the wave characteristics and its first derivative. A travelling wave technique for teed circuits is introduced in 8 considering the cross correlation between the forward and backward travelling waves and a polarity change criterion in order to determine the f

9、aulted region prior to fault location estimation. A single ended fault location method is proposed in 9 for two and three terminal lines where the voltages and currents are estimated by solving the travelling wave equation and then different criteria are used to determine the fault location. More re

10、cently,in 1, the use of the discrete wavelet transform (DWT) of the modal components of the fault initiated travelling waves is proposed in order to estimate the location of the fault.The presence of parallel transmission lines with mutually coupled line sections makes the fault location problem mor

11、e difficult in transmission lines. Two different algorithms for different types of faults are developed in 10 and 11 by applying the Z-transform to the loop equations and using Newton Raphson method to solve the nonlinear equation. One ended data with the simplified line model is used by neglecting

12、the shunt capacitance.Another phasor based single ended fault location technique is proposed in 12 where post and pre fault data are used with zero sequence current from a healthy line in order to solve the algebraic equation. A similar approach to 6 is used in 13 by introducing the synchronized mea

13、surements. A complex and nonlinear equation is derived from the nodal equations and solved by Newton Raphson iterative scheme in14. The technique is validated for a parallel transmission line with a teed circuit, using a lumped line model and for single phase to ground faults only.Another source of

14、difficulty in fault location problem is the presence of series capacitors which are widely used in power systems in order to improve the transfer capability and increase the stability margins. Metal Oxide Varistor (MOV) is the most popular protection device which is connected across the capacitor.Th

15、e existing fault location techniques have to be adapted in order to cope with the complexity introduced by the nonlinear V-I characteristics of the MOV. Different solutions including the use of Artificial Neural Networks 15, single ended 16 and multi-ended 17 measurements have been proposed.In this

16、paper, a travelling wave based fault location technique which is developed earlier in 1 will be extended to the three terminal circuits with mutually coupled line segments and MOV protected series capacitors. Preliminary results 18, 19 indicate that this approach can overcome the challenges presente

17、d by such topologies. The performance of the proposed fault location algorithm is tested by introducing random errors representing the quantization error introduced by A/D converters, to the simulated signals and using various fault resistances.II. FAULT LOCATION PROCEDUREThe following assumptions a

18、re made in developing the fault location procedure: Three terminal measurements are available The measurements need not be synchronized An open communication channel is available between the terminals There is no injection or load at the tee point.The procedure consists of three stages. In the first

19、 stage, the modal transformation is applied to the measured voltage signals. Clarke 20 transformation matrix is used asWhere , and are phase voltages, is ground mode voltage and ,are aerial mode voltages. In the second stage, the discrete wavelet transform (DWT) is applied to the modal voltages and

20、the squares of the wavelet transform coefficients (WTC) are obtained in order to determine the instant when the energy of the signal reaches its maximum value. Daubechies-4 21 mother wavelet is used for wavelet transformation. Then in the final stage, ground mode WTCs in scale-1 are observed in orde

21、r to determine the fault type (whether the fault is grounded or not) and aerial mode WTCs in scale-1 are processed based on the Bewley lattice diagram22 of the fault initiated travelling waves in order to determine the fault location. In the following sections the last stage of the fault location pr

22、ocedure is described in detail for various possible cases.A. Fault Location in Teed CircuitFault location in teed circuits involves two basic steps. In the first step the faulted line segment is identified and in the second stage the fault location along the faulted line segment is determined.The ae

23、rial mode WTCs are compared at each bus in order to identify the faulted line segment. The magnitude of the first peak of the aerial mode WTCs obtained at the sending end of the faulted line segment will be significantly higher than those obtained at the sending ends of the other nonfaulted line seg

24、ments. Once the faulted line segment is known, the location of the fault is determined by using a modified version of the single ended algorithm proposed in 1 as described below.A grounded fault is assumed to occur at the first half of the line segment A-T, at point in Fig. 2. The first peak of the

25、aerial mode WTC is due to the backward travelling wave arriving at bus A at time . The second peak is due to the reflected backward travelling wave arriving at bus A at time .The fault location is given by 1 (1) Where is the aerial mode propagation velocity in scale-1.Now consider a fault close to t

26、he T-node, shown as in Fig. 2. The backward travelling wave arrives at bus A at time while the forward travelling wave arrives at bus A at time . The fault location can then be determined as in 1 (2)where L is the total length of the line segment A-T and is the aerial mode propagation velocity in sc

27、ale-1. Fault location can also be determined by using the third peak of the aerial mode WTCwhich arrives at bus A at time . The following equation is used for the calculations 18: (3)Nevertheless, as the fault location moves closer to the T-node, the second peak gradually decreases, coming very clos

28、e to the first peak and eventually the two becoming indistinguishable. The difficulty of identifying the second peak when the fault is in the second half, can be overcome by using (3) instead of using (2) to calculate the fault location.The faulted half of the line is determined by comparing the tim

29、e difference , between the arrival time instants of the aerial mode and the ground mode WTCs with the time difference, obtained for a fault located right at the middle of the line.Since the ungrounded and symmetric faults do not produce remote end reflections, (1) will be used in order to locate the

30、 fault independent of the half in which the fault occurs.B. Fault Location in a Teed Circuit With MOV Protected Series CapacitorThe most popular and widely used device for protecting series capacitor against high voltage during faults is the Metal Oxide Varistor (MOV), which is installed directly ac

31、ross the series capacitor as shown in Fig. 3, and has a nonlinear I-V characteristics (4) The nonlinear characteristics given in (4) allows no current to flow through the MOV under normal operating conditions. In case of fault, when the voltage across the capacitor reaches the threshold , MOV clamps

32、 the voltage and starts to conduct. The voltage recorded at the sending end will have a different waveform because of this clamping action compared with the waveform obtained without using the MOV.Assume that a single phase to ground fault occurs on the line segment A-T and voltage transients are re

33、corded at the sending end of the line segments A-T, B-T and C-T in Fig. 3. These signals are not synchronized. The voltage signals of the faulted phase are presented in Figs. 4 and 5 for the cases with or without the MOV in order to show the effect of the varistor. Note that in Fig. 5, after the fau

34、lt occurs the phase voltage is clamped between 200kV thanks to the MOV protection.The simulated voltage signals at each bus are subsequently transformed into the modal domain. Discrete wavelet transform coefficients (WTC) of different scales for the aerial and ground mode signals are then calculated

35、 using the wavelet transform. The WTCs of the aerial mode voltages in scale-1 for both cases with and without the MOV are given in Figs. 6 and 7. It is observed that both the shape and the peak arrival instants of the wavelet transform coefficients of the aerial mode voltages for each case, are iden

36、tical for a certain period after the fault. This period extends well beyond the needed duration for the successful application of the fault location procedure described in the previous section.C. Fault Location in a Teed Circuit With Mutually Coupled Line SectionA partially coupled teed circuit is s

37、tudied as shown in Fig. 8.The detailed interpretation of the lattice diagram in 19 shows that the end point of the mutually coupled section, , behaves like a discontinuity where multiple reflections occur during a fault. Because of this complexity introduced by the coupled section the following situ

38、ations must be studied depending on where the fault occurs: Fault is in the coupled section,A-M In the first half of the coupled sectionIn the second half of the coupled section. Fault is beyond the coupled section, M-TIn order to specify the faulted section (coupled or uncoupled), the difference be

39、tween the arrival time instants of the WTCs peaks of the aerial mode and the ground mode voltages are calculated. Then the calculated value is compared with the time difference obtained for a fault right at the end of the coupled section.In case a grounded fault occurs in the coupled line section, t

40、he algorithm described in the previous section will be valid, and the (1) and (3) will be used in order to locate the fault.When the fault is in the uncoupled section as shown in Fig. 9, the first peak of the aerial mode WTC which is due to the backward travelling wave, arrives at bus A at time and

41、the th peak which is due to the backward travelling wave that is firstreflected from bus A, will arrive at bus A at time . It is observed that, the WTC peak (th) which has the largest magnitude after the first two WTC peaks is due to the backward travelling wave which is reflected from the fault poi

42、nt. As can be seen from the Figs. 10 to 12, the arrival time of the WTC peak due to the reflected backward travelling wave increases (from fifth to tenth) as the fault location approaches the T node. The fault location can be calculated by the following equation: (5)Provided that the fault location

43、is not too close to the T-node, an alternative and simpler procedure can be used as follows 1: (6)where L is the total line length. As a result, either one of the (5) or (6) can be used in order to locate the fault. Alternatively, their average value can be used to minimize any estimation errors.D.F

44、ault Location in a Teed Circuit Using Synchronized Measurements at Three TerminalsNow assume that the measurements are synchronized at each bus and voltage transients are recorded at buses A, B, and C as shown in Fig. 1. After the faulted line segment is identified by comparing the WTCs of each bus

45、as described above, the double ended fault location algorithm described in 1 is used to locate the fault. The aerial mode voltage WTCs obtained at the sending ends of the faulted line segment and one of the un-faulted line segments are used in this algorithm as follows: (7)where is the time differen

46、ce between the arrival times of the aerial mode WTCs of the voltages recorded at the sending ends of the faulted line segment and the chosen un-faulted line segment. is the total line length of the faulted line segment and the chosen un-faulted line segment. Assume that line segment C-T is the fault

47、ed line segment and,and are the arrival time instants of the initial peaks of the aerial mode voltage WTCs at the buses A, B, and C respectively. Then the fault location will be calculated as follows: (8)Where j=A(or B) as the chosen un-faulted line segment.III. SIMULATION RESULTSAll simulations are

48、 carried out by using ATP/EMTP program and MATLAB with a sampling time interval of 3. The fault occurrence time is chosen as 0.02 sec. The tower configuration of 220 kV transmission line is given in 18 and 19. The frequency dependent transmission line model is used throughout the simulations. All the line segments are assumed to be fully transposed. The aerial mode propagation velocity is calculated as mi/sec in scale-1 corresponding to