WO2019232595A1

WO2019232595A1 - A method of estimating the location of a fault on an electrical distribution network and an associated system

Info

Publication number: WO2019232595A1
Application number: PCT/AU2019/050596
Authority: WO
Inventors: Ali Tashakkori JAHROMI; Syed Mofizul Islam; Peter Joseph Wolfs
Original assignee: Federation University Australia
Priority date: 2018-06-07
Filing date: 2019-06-07
Publication date: 2019-12-12
Also published as: AU2019280259A1

Abstract

A method of estimating the location of a fault on an electricity distribution network is disclosed. The method comprises locating a plurality of travelling wave detectors on the network, each travelling wave detector adapted to detect voltage travelling waves (VTWs) or current travelling waves (CTWs) propagating through the network, and providing a data processing system adapted to process data and store at least one data set relating to network parameters. The method also comprises operating the detectors to detect travelling waves propagating on the network and to store the time of arrival of said travelling wave at each detector so as to produce time of arrival data, and using the at least one data set and the time of arrival data to estimate the origin of the travelling wave on the network and thereby estimate the location of the fault.

Description

A METHOD OF ESTIMATING THE LOCATION OF A FAULT ON AN ELECTRICAL DISTRIBUTION NETWORK AND AN ASSOCIATED SYSTEM

Field

This invention is in the field of detecting faults and identifying the location of faults within electrical distribution networks, and more particularly but not limited to, detection of high impedance faults and the location of high impedance faults within overhead and/or medium voltage electrical distribution networks.

Background

Faults occurring within Medium Voltage (MV) overhead distribution lines of electrical distribution networks have ignited a number bushfires and wildfires that have caused significant property damage and loss of life. For example, downed conductors, vegetation contacting

conductors, sparks from clashing conductors or apparatus failure are the common ignition sources.

Rural distribution networks are typically aged long radial systems with large numbers of lateral and tee junctions, resulting in complex network topologies. These complex topologies limit the ability of prior art fault location technologies to identify fault locations with sufficient accuracy for crews to identify the underlying cause of transient and intermittent faults, and for crews to attend and repair faults in a timely manner.

Faults launch travelling waves since sudden changes in state variables of a power system can generate a voltage and current travelling wave, which propagate from the fault location toward the ends of a connected network. In overhead transmission lines, faults travel at close to the speed of light. Summary of the Invention

According to a first aspect of the invention there is provided a method of estimating the location of a fault on an electricity distribution network, comprising:

i) locating a plurality of travelling wave detectors on the network, each travelling wave detector adapted to detect voltage travelling waves (VTWs) or current

travelling waves (CTWs) propagating through the network; ii) providing a data processing system adapted to process data and store at least one data set relating to network parameters;

iii) operating the detectors to detect travelling waves propagating on the network and to store the time of arrival of said travelling wave at each detector so as to produce time of arrival data;

iv) using the at least one data set and the time of arrival data to estimate the origin of the travelling wave on the network and thereby estimate the location of the fault .

The at least one data set may comprise data as to propagation times for travelling waves between a plurality of predetermined points on the network.

The propagation times between a plurality of

predetermined points may comprise the propagation time along predetermined sections of the network, wherein each section is located intermediate consecutive points of the plurality of predetermined points.

The at least one data set may further comprise differential data, wherein the differential data comprises data as to the difference in propagation times for a travelling wave from at least one pre-determined point on the network and each detector in at least one

predetermined pair of detectors.

The differential data may comprise a data set of differential data associated with each one of a plurality of predetermined points on the network. The differential data may comprise differential data for a plurality of predetermined pairs of detectors on the network with each one of said plurality of predetermined points .

The data processing system may further process the time of arrival data to produce differential data as to the difference in the time of arrival of the travelling wave between at least one pair of detectors.

The differential data associated with the time of arrival data may comprise differential data as to the difference in the time of arrival of the travelling wave between a plurality of pairs of detectors.

The processing of at least one data set and the time of arrival data by said data processing system may further comprise comparing the differential data of the at least one data set and the differential data of the time of arrival data so as to produce a first parameter indicative of the origin of the travelling wave from at least one of said pre-determined points.

The method of estimating the location of faults on the network may further comprise the steps of:

producing said first parameter for each one of said pre-determined points; and

forming a set of the first parameters by selecting any of the first parameters within a pre-determined range.

The method of estimating the location of faults on the network may further comprise the step of using the parameters in the set of first parameters to determine a line section of the network likely to contain the fault.

The method of estimating the location of faults on the network may further comprise a cost function value is determined for each pre-determined point associated with the first set of parameters and

the line section of the network associated with the pre-determined point with the lowest cost function value is selected as the line section containing the origin of the travelling wave. The method of estimating the location of faults on the network may further comprise the step of applying the first parameter associated with the selected line section to determine the origin of the travelling wave on the selected line section.

The at least one data set may further comprise transformer propagation delay data that comprises data as to reductions in travelling wave propagation times on the network arising from transformers connected to the

network .

The detectors may be located at geographically diverse locations on the network.

At least some of the detectors may be located on the network independently of any sub-stations on the network.

The detectors may monitor received travelling waves to discriminate between switching events and fault events.

The method of estimating the location of faults on the network may further comprise the step of at least one detector performing Clarke's transform on detected

voltages so as to determine one or more parameters of travelling waves detected by the at least one detector.

The method of estimating the location of faults on the network may further comprise the step of the at least one detector determining modal components of travelling waves from said Clarke's transform.

The method of estimating the location of faults on the network may further comprise the step of performing discrete wavelet transforms on aerial mode components derived from voltage or current signals detected by the at least one detector, and wherein the discrete wavelet transform is used to determine the one or more parameters of travelling waves.

The wavelet transform coefficient energies may be derived from voltages or currents detected by the at least one detector and the wavelet transform coefficient

energies are used to determine one or more parameters of the travelling waves. The one or more parameters may include:

i) time of arrival of a travelling wave at the at least one detector;

ii) aerial mode energies associated with phase to phase faults on said network; and

iii) ground or earth mode energy associated with phase to ground faults on said network.

The detectors may analyse travelling wave data over multiple cycles so as to discriminate between travelling waves originating from switching events and travelling waves originating from fault events.

The method of estimating the location of faults on the network may further comprise the step of processing data associated with travelling waves to identify the occurrence of repeat transient faults originating from substantially the same location on the network.

The travelling wave may be a voltage travelling wave or a current travelling wave.

For voltage travelling waves, the detectors may be

capacitive sensors, or other sensors responsive to the electric field surrounding the conductor, located in close proximity to the network conductors without being in contact with the network conductors.

For current travelling waves, the detectors may be Hall effect or Rogowski coil sensors, or other sensors

responsive to the magnetic field surrounding the

conductor, located in close proximity to the network conductors without being in contact with the network conductors .

The network may be an overhead network, or may be an overhead medium voltage network, or maybe a radial network

According to a second aspect of the present invention there is provided a method of detecting a high impedance fault on an alternating current (AC) power distribution network, the method comprising the steps of:

monitoring the network at one or more locations for the occurrence of travelling waves on two or more consecutive cycles of the baseline AC power of the distribution network;

analysing travelling waves detected over two or more consecutive cycles of the baseline AC power of the distribution network to identify similarities in the characteristics of the travelling waves; and

determining that the travelling waves originate from a high impedance fault where the similarities in the travelling waves repeat, or substantially repeat, for the two or more consecutive cycles.

The method of detecting high impedance faults on an alternating current (AC) power distribution network may further comprise the steps of:

operating travelling detectors at one or more locations on the network to monitor network voltage or current ;

performing a Clarke's transform on voltages or currents detected by said at least one detector so as to identify modal voltages or currents of the network;

analysing the aerial and ground mode voltages or currents to identify phase to phase and phase to ground faults on the network.

The method of detecting high impedance faults on an alternating current (AC) power distribution network may further comprise the step of calculating wavelet transform coefficient energies from the modal voltages and currents and using the wavelet transform coefficient energies to determine phase to phase and phase to ground faults on said network.

According to third aspect of the present invention there is provided a system for estimating the location of a fault on an electricity distribution network,

comprising :

a plurality of travelling wave detectors located on the network, each travelling wave detector arranged to: detect voltage travelling waves (VTWs) or current travelling waves (CTWs) propagating through the network; and

store the time of arrival of said travelling wave so as to produce time of arrival data;

a data processing system arranged to store at least one data set and estimate the origin of the travelling wave on the network based on the at least one data set and the time of arrival data, to thereby estimate the location of the fault.

The plurality of travelling wave detectors may be synchronised by satellite navigation signals.

The travelling wave detectors may be adapted to time stamp signal data detected by sensors and wherein time stamp data for the plurality of travelling wave detectors is synchronised by satellite navigation data.

The data processing system may be arranged to receive the time of arrival data from the plurality of travelling wave detectors and process the time of arrival data and the at least one data set to estimate the origin of the travelling wave.

Also disclosed is a method of estimating the location of a fault on an electricity distribution network, the network comprising a plurality of nodes and line sections therebetween, the method comprising:

locating a plurality of travelling wave detectors on the network, the travelling wave detectors arranged to detect voltage travelling waves (VTWs) or current

travelling waves (CTWs) propagating through the network; formulating at least one data set of network

parameters including data related to wave propagation times between nodes and detectors on the network;

determining arrival times of a travelling wave at two or more of the plurality of travelling wave detectors, wherein the travelling wave is caused by a fault on the network; based on the determined arrival times and the at least one data set of network parameters, identifying one of the line sections from which the fault originated.

The step of determining arrival times of the

travelling wave may comprise detecting voltages at conductors of the network and converting the voltages to aerial mode components. The converted voltage values may then be processed using discrete wavelet transform techniques to obtain Wavelet Transform Coefficients. A first local peak of a derivative of the Wavelet Transform Coefficient may be indicative of arrival time.

The step of formulating at least one data set of network parameters may comprise formulating a data set of travelling wave propagation times along each line section and formulating a data set of travelling wave propagation times from at least one node to at least two detectors. A third data set of time differences between arrival times registered by the at least two detectors may then be formulated .

The method may further comprise estimating a point along the identified line section where the fault

originated .

Brief Description of Drawings

Various embodiments of the invention will now be described with reference to the accompanying drawings of which :

Figure 1 is a schematic representation of an

electricity distribution network;

Figure 2a is a schematic representation of an electricity distribution network with a plurality of travelling wave detectors D1 to D4;

Figure 2b is a schematic representation of an electricity distribution network with a high impedance fault;

Figure 3 is a schematic representation of a

travelling wave detector unit; Figure 4 is a schematic representation of a portion of the travelling wave unit of Figure 3, showing

components of a travelling wave antenna;

Figure 5 is a schematic representation of network phase conductors and the travelling wave antennas of Figure 4 in relation to three phase conductors of an overhead networks;

Figure 6 is a circuit model of the conductors and antennas shown in Figure 5;

Figure 7 is a graph of a balanced three phase voltage signal of the antennas shown in Figure 5;

Figure 8 is a graph of the three-phase voltage of the conductors shown in Figure 5;

Figure 9 is a graph of a current transient of a simulated fault of the network;

Figure 10 is a graph of a voltage transient

associated with the current transient of Figure 9;

Figure 11 depicts various aerial mode voltages for the simulated fault of Figure 9;

Figure 12 shows a further voltage transient arising from the fault of Figure 9;

Figure 13 is the calculated cost function used to determine the most likely location of the fault on the network;

Figure 14 is a flow chart of a process for

determining a location of a fault;

Figure 15 is a block diagram of a voltage travelling wave detector;

Figure 16 is a block diagram of a central data processing system.

Detailed Description

Embodiments of the present invention provide a system and method for estimating the location of faults within an electrical distribution network, such as a medium voltage overhead electrical distribution network, using detectors adapted to detect travelling waves, such as voltage travelling waves (VTW) or current travelling waves (CTW) . The voltage and current travelling waves are inextricably related to the line characteristic impedance. In some embodiments of the invention, VTWs are preferred to be detected as these may be detected with a capacitive antenna system, however the principles for identifying the location of a fault on a network as described herein are applicable to both current and voltage travelling waves.

Embodiments of the present invention also provide a system and method for characterising the nature of a fault detected on an electrical distribution network, such as high impedance faults or clashing conductor faults.

Figure 1 is a schematic representation of a medium voltage overhead electrical distribution network 5

according to an example. The location of a fault can be determined by recording instances at which VTWs or CTWs generated by the fault arrive at various points of the network. The network 5 has a plurality of VTW detectors 10 in disparate locations on the network 5. The VTW

detectors 10 monitor the operation of the network 5 to identify the occurrence of a VTW. Each VTW detector 10 records the time at which it detects a VTW, which may be referred to as the "time of arrival" (ToA) of the VTW at the VTW detector 10 of interest.

Time recording equipment used by the various VTW detectors 10 may be synchronised by use of Global

Positioning System (GPS) equipment, or similar equipment such as Global Navigation Satellite System (GNSS) sensors, embedded within, or otherwise associated with the VTW detectors 10.

Each VTW detector 10 transmits VTW data to a data processing system, which in this example is provided at least in part by a centralised data processing station 15. The data processing system and each VTW detector 10 may comprise any suitable wireless communication means so that the VTW data may be transmitted wirelessly. The

centralised data processing station 15 processes VTW data received from the VTW detectors 10 and determines the origin on the network of any VTW detected and/or whether the VTW originated from a fault or a switching event.

Fault detection method and system

An embodiment of a method and associated system for identifying the location of a fault within an electrical distribution network 5 will now be described with

reference to Figure 2a to Figure 14. In particular,

Figure 2a shows a plurality of nodes "Q", numbered Q1 to Q34 on the network 5. The nodes Q serve to divide the network 5 into a series of line sections "L", numbered as LI to 133. The location of nodes Q may be selected so as to be situated at any convenient location within the network 5. It is not necessary that the nodes Q

correspond to the physical location of electrical

equipment within the network 5. VTW detectors 10,

numbered as Dl, D2, D3 and D4 are located at separate geographic locations around the network 5. In particular, VTW detector Dl is located on line section LI between nodes Q1 and Q2, VTW detector D2 is located on line section L8 between nodes Q8 and Q9, VTW detector D3 is located on line section L20 between nodes Q20 and Q23, and VTW detector D4 is located on line section L31 between nodes Q31 and Q34.

The line sections L of network 5 (i.e 11 to L33) are used in the development of data sets that characterise properties of the network 5, including a first data set comprising VTW propagation times along each line section L (i.e. VTW propagation times between adjacent nodes Q, such as line section L6 between nodes Q6 and Q7) . The first data set may be referred to as the line section

propagation time data set ( LSPT data set ) .

The LSPT data set is developed by estimating the time for a VTW to propagate between adjacent nodes Q, such as along line section L6 between nodes Q6 and Q7. This is based on the length of a particular line segment L and the propagation velocity of the VTW along that line section L. The VTW propagation velocity (v®)for each line section L is provided by:

(Equation 1)

where, _L ⁽ⁱ and ct^iy are per-unit-length inductance and capacitance of the line section L, respectively.

The LSPT data set may be used to develop a second data set which is the set of VTW propagation times from each node Q to each of the VTW detectors D1 to D4. The second data set may be referred to as the detector

propagation time data set (DPT data set) .

For example, the time for a VTW to propagate from node Q5 to detector D1 may be one data point of the DPT data set. The time for a VTW to propagate from node Q5 to detector D2 may be a second data point. The DPT data set comprises propagation times for voltage travelling waves to propagate from each node Q to each VTW detector.

The DPT data set may be used to develop a third data set, which is a set of VTW arrival time differences between each pair of VTW detectors 10 for VTWs originating at each node Q. This third data set may be referred to as the Network Arrival Time Difference Data Set (ATD (network) data set) .

Each of the ATD (network) data set, the DPT data set and the LSPT data sets, are network specific data sets that can be calculated in advance and stored by a VTW detector 10 and/or by the data processing system at the central data processing station 15. These network

specific parameters may be used in conjunction with VTW data measured by the VTW detectors 10 to estimate the location of a fault on a network.

To demonstrate how the location of a fault can be estimated using the ATD (network) data set, the DPT data set and the LSPT data sets, refer now to Figure 2b which shows a portion of a network with fifteen nodes, Q1 to Q15, two VTW detectors 10 (numbered as nl and n2 ) and one fault "F".

Fault F is located between nodes Q4 and Q2. Detector nl is located at node Q1 and detector n2 is located at node Q15.

VTW propagation time for the line section "L" from Q4 to Q2 (i.e. LSPT data for the line section "L" between Q4 and Q2 ) is represented by the variable T_L. VTW propagation times from Q4 to detector nl is represented by Ti,_ni (i.e. DPT data from Q4 to detector nl is Ti, _ΐ) and the VTW propagation time from node Q4 to detector n2 is

represented by Ti,_n2(i.e. DPT data from Q4 to detector n2 is

It, n2 ) ·

The ATD (network) data set between detectors nl and n2 for a VTW originating at Q4 (which is a node at one end of line section "L") may be expressed as:

ATHL⁼T_I-T,_M <^E-^:i^uati°^{n 2}1

For a fault F which occurs on line section l, the location of fault F may be expressed in terms of the propagation time of a VTW along line section "l" from node Q4 to fault F. Hence, the location of fault F from node Q4 can be expressed as "CXT_L", where represents the fraction of the length of line section "l" from node Q4 to where the fault is located. Accordingly, "a" is in the range 0.0 < < 1.0.

Using the above expression, the propagation time from fault F to detector nl is:

" TL, ni - aT_L"

and the propagation time from fault F to detector n2 is:

" TL, _n2 + aT_L"

The difference in arrival time at detectors nl and n2 of a VTW originating at fault F may be expressed as:

ATD (Fault) nl,n2 = (TL,H2 + OCT_L) - (T_L,nl - OCTL)

(Equation 3) A parameter "S" can be introduced so that the VTW propagation time from node Q4 to either of detectors nl and n2 can be expressed in the following generalised form:

T_L,n + (S_l,n * OC^(L)T_L)

Where S = +1 if the path from the relevant node (in this example Q4) to the relevant detector does not include line section "l", and S = -1 if the path from the relevant node (in this example Q4) to the relevant detector

includes line section "L".

Using parameter "S" in the above Equation 3 allows it to be written in the form:

ATD ( Faul t) nl ,n2 = [ Tl,n2 + ( S[.,n2*CXTl) ] — [ Tt,nl + ( Sp nl*(XTl.) ] which can be re-written as

(Equation 3a)

Note that [Tq_n2 - Ti,_ni] is the ATD (network) data set for node Q4 associated with line section l, assuming that

Tl,n ^< Tl,n₂ ·

This allows Equation 3a to be re-written as:

ATD (fault) = ATD (network) + ( S^L _ni ,_n2*aTi.)

(Equation 3b)

If the ATD (network) data set (and associated LSPT data set and DPT data sets) have been estimated accurately then, for the same fault, the ATD (Fault) data set should equal a measured data set (ATD (measured) data set) .

Substituting the ATD (measured) data set into equation 3b provides the following generalised equation:

ATD (network) +

- ATD (measured) = 0

(Equation 4)

Using the alternate notation of

for ATD (network) and ATD (measured) , Equation 4 can be re-written as:

(Equation 4a)

where :

Equation 4 is an over determined system of equations forN>2, where N is the number of VTW detectors. The

ATD (network) data set

, and the "S" parameter data set

are static data sets that can be determined by analysing the network and stored in the data processing system.

The ATD (measured) data set (i.e. DT^_h. ) may be calculated once a fault has occurred and the ToA of the resultant VTW at each VTW detector 10 has been measured.

To allow for inaccuracies in the ToA measurements by the VTW detectors 10, and in the calculation of the ATD (network) data set, Equation 4 can be redefined as the following constrained optimization problem:

By choosing ||*|| as the standard vector norm (Euclidean) , the cost function becomes:

By applying the first-order optimality condition, closed- form expressions will be obtained for d® :

Equation 7 can be rewritten for 5® as

The location of fault F can thus be determined by solving the optimization problem as follows:

(i) Determine optimum values of "a" for all line sections "1" using ATD (network) data,

ATD (measured) data, the S parameter data set and equations 8 and 9 above;

(ii) identify the line sections "1" for which the

value of "a" (as calculated in step (i) ) is in the range 0.0 < < 1.0;

(iii) for each line section identified in

step (ii), determine the minimum value of the corresponding cost function using equation 5; (iv) identify the line section "l" from step (iii) with the minimum cost function value and select this as the line section on which fault F occurs;

(v) use the value of "a" calculated for the line

section "l" selected in step (iv) to determine the location of fault F on the network.

System hardware and modelling

Referring now to Figures 3 and 4, there is shown block diagrams of hardware components of a system for detecting fault locations within the electrical

distribution network 5 according to an embodiment of the invention. In particular, Figure 3 is a block diagram of a VTW detector unit 300 according to a specific example, and Figure 4 is a more detailed block diagram of a receiving end of the VTW detector unit 300.

A plurality of VTW detector units 300 may be provided at various locations on the network electrical

distribution network 5 in order to detect a fault location according to embodiments of the invention.

In this example, three external antennas 305 are capacitively coupled to network conductors in order to detect the electric field of travelling waves. Antenna signals are processed by an analogue receiver and sent to the VTW detector 10 and a data acquisition system 330. The data acquisition system 330 captures raw data as an aid to the system development and evaluation.

The detector 10 in this example is based on the Texas Instruments Delfino 28377 processor. The detector 10 is in communication with a GPS system 310 in order to have access to GPS signals for determining the ToA of VTWs .

The detector 10 is also connected via a USB port to a single board computer system 315, to provide data transfer and control. A power supply for the VTW detector unit 300 in this example requires a 230Vac input and is based on a DC UPS with a 12V sealed lead acid battery 320.

The single board computer 315 uses an Advantech AIMB- 213 motherboard with a LINUX operating system. It has a 200MB hard drive for the operating system and a 1TB external hard drive 325 for data storage. The data

acquisition system 330 is a high-speed PCI based four channel acquisition card with a sample rate of lOMs/s on each channel. The card has 256Ms of storage on board. The data acquisition system 330 can be remotely triggered, and data can be downloaded using a local Wi-Fi transceiver 335 or 3G network connection 340.

Referring to Figure 4, each of the VTW detector 10 and the data acquisition system 330 is connected to an independent wide bandwidth amplifier and antialiasing filter 420, 425. The VTW detector 10 samples VTWs at 1.5Ms/s, while the (high-speed) data acquisition system 330 samples VTWs at lOMs/s.

In this example, the VTW detector 10 and the data acquisition system 330 are connected to the antennas 305 that are typically mounted 1.5m below line conductors of the network 5. The antennas 305 typically have a mutual capacitance of approximately 2pF. The mutual antenna capacitance forms one part of a capacitive voltage

divider. The lower capacitance is approximately 20nF resulting in an attenuation of 1000:1. Voltages at this level are compatible with both the VTW detectors 10 and the (high-speed) data acquisition system 330.

The VTW detector 10 is enclosed, in this example, with a double-skinned enclosure to reduce solar heating, and has thermostatically operated fans that exchange air with the atmosphere through the bottom surface of the enclosure .

With reference to Figure 5, a theoretical model applicable to the system shown in Figures 3 and 4 will now be described. Figure 5 shows three phase conductors A, B, and C, of the network 5, with antennas 305 (in this example, short conductive antennas denoted as D, E, and F) , of VTW detector 10 placed at distance "hi" or "h₂" under the phase conductors A, B and C. Table 1 shows an example of typical 22kV distribution pole dimensions that may be applied to the three phase conductors A, B, C and antennas D, E, F of Figure 5.

Table I

Typical 22kV distribution pole dimensions

A capacitive impedance exists between the conductors A, B and C, and also exists between the conductors and ground .

The pole configuration of Figure 5 can be modelled electrically as a network of twenty-one capacitors, as illustrated in Figure 6. The capacitance C between two conductors with length l is given by

Capacitance to ground of conductors with length l is given by

2D. (Eqn. 11)

r

Where D is the distance between the conductor centres, variables ri and r2 are the conductor radius, and

constants £_Q ^and £_r are free space permittivity and dielectric relevant permittivity respectively.

By using the superposition theorem, the antenna to ground voltage can be determined as the sum of the responses to each phase voltage. The total network thus reacts as a capacitive voltage divider. The voltage dividers are modelled as being frequency independent as the resistance of the air gap is much higher than the capacitive impedance. As each antenna has a significant capacitance, 20nF, physically connected to ground in the receiver circuit, the calculation can be simplified by ignoring the capacitance between antennas. The antenna voltages to ground are therefore given by:

(Eqn . 12)

or

(Eqn .

VDEF — X- V ABC

13)

Each variable X_[j in Equation 12 above is the

capacitive voltage divider ratio that forms between phase conductor Ϊ and antenna j and ground. The impedance ratio matrix X is invertible and the phase voltages can be calculated from the antenna to ground voltages:

(Eqn .

V ABC = X^® V DEF

14)

For the configuration shown in Figure 5, with the example dimensions provided in Table 1, and by considering the capacitance divider of detector input, the X^-1 matrix is equal to:

0.8573 0.7647 -1.7010

X-¹ = -1.6219 1.0967 3.4102 * 10⁶ (6)

0.7647 0.8573 —1.7010-1

For the configuration shown in Figure 5 and Table 1, a simulated voltage for antennas D, E, F, in response to balanced phase voltages, is shown in Figure 7. It is noted that because a distance between the antenna F (a central antenna) and the three conductors A, B, and C, is almost equal, its voltage to ground is close to zero. Figure 8 shows the three-phase voltage of conductors A, B and C reconstructed by multiplying the antenna voltages and the transform matrix X^-1.

Example of fault location detection

As described above, the location of a fault can be determined from a set of times of arrival (ToA) of VTWs at VTW detectors 10 on the network. An example of how the VTW detector 10 can "time-stamp" VTW arrival times, i.e.

detect VTWs and identify their arrival time, is described below.

Voltages on conductors A, B and C detected by the antenna 305 (i.e. measured voltages V_A, V_B and V_c) may be transformed into aerial mode components using Clarke's real transform matrix as follows:

Where V_a , V and V_c are the phase to ground voltages; V₀ is a ground mode; and V_a and

are the aerial-mode voltages.

Discrete wavelet transform (DWT) techniques may then be used to detect, and log, the ToA of VTWs. A discrete wavelet transform (DWT) for a given signal x(k) with respect to a mother wavelet W(k) is given by

Where a and b are scale and translation factors,

respectively. In one example, lower scale factors 'a' give higher frequency or more detailed information of a hidden pattern in the signal, whereas higher scale factors 'a' give lower frequency information. Travelling wave

transients may appear as high frequency components, and therefore travelling waves can be detected by a lower scale DWT. Wavelet Transform Coefficient (WTC²) energies can then be used to determine the ToA of a VTW.

The selection criteria include the wavelet

correlation to the expected transient signals, in that the better the correlation, the higher the Wavelet Transform Coefficient energy (WTC²) . The EMTP-RV software package may be used to simulate a radial feeder electricity

distribution network, and the Wavelet Toolbox incorporated in the MATLAB software package may be used to perform the DWT . For example, in at least one simulation the Reverse Biorthogonal 3.1 (rbio 3.1) wavelets in MATLAB have shown a good correlation with expected fault signals.

The system depicted in Figure 14 can be used to calculate the ToA of a VTW in accordance with the above process .

The system of Figure 14 comprises capacitive voltage antennas 305 which detect the voltage wave form on the conductors A, B and C of network 5. Alternate embodiments may use Hall effect sensors or Rogowski coil sensors to detect current travelling waves.

An analogue to digital converter (A/D converter) 1405 receives the voltage waveforms from the antennas 305 and converts these signals into digital signals.

A modal transform module 1410 receives the digital signals from the A/D converter 1405 and performs a Clarke transform on the digital signal to derive the ground mode voltage Vo and the aerial voltages V_a and Vp.

A discrete wavelet transform module 1415 calculates the WTC² coefficients from the ground mode and aerial voltages V_a and Vp. Electrical faults will produce a VTW which are visible in the three Clarke components.

Electrical faults of different types will produce

different responses in the three Clarke components and this may be utilised to identify the fault type. For example, phase to phase faults as produced by conductor clashing will produce strong VTWs in the aerial components while high impedance faults to the earth will produce strong VTWs in the ground mode voltage. Without loss of generality, the methods described herein can be applied to a plurality of fault types. In one example, the WTC² coefficients of one aerial voltage V_a may be utilised.

The WTC² coefficients are then differentiated with respect to time by a differentiator module 1420.

The first positive peak output from the

differentiator 1420 is recorded as the ToA of the VTW.

To demonstrate the above, a simulation of the above fault detection system was carried out using the EMTP-RV simulation software package and the MATLAB software package .

The EMTP-RV simulations start from a load-flow steady-state solution to shorten the simulation time. EMTP simulations were run from the load-flow simulation for one cycle (20ms) and all faults are applied at 0.005ms into the cycle. Frequency dependent models are deployed to represent the distribution lines. The pole and cross-arms configuration of Figures 5 and 6 were used. The adapted IEEE 34-bus network (as depicted in Figure 2a) was used as the study system. Line lengths and travelling wave (TW) propagation times are provided in Table II below. (Note that: each node 'Q' in Figure 2a is a bus in the 34-bus network; and the line section between nodes Q1 and Q2 from Figure 2a is denoted by 001-002, and between nodes Q2 and Q3 is denoted by 002-003, etc.)

Only spot loads were considered during the

simulation. The spot loads were simulated as constant impedance loads. Parasitic capacitance of power

apparatus, was ignored, however it should be noted that the capacitance at equipment such as distribution

transformers should be accounted for in order to achieve accurate VTW fault location.

Simulations were carried out with time-step of 100ns.

Phase voltages were obtained at the end of each branch.

By using MATLAB, obtained voltages are down-sampled to the equivalent sampling frequency of 2MHz, following the decoupling of the phase voltages into the aerial-voltage quantities, and calculating dWTC²/dt .

Four simulated fault-scenarios are described in

Tables IIIA and IIIB below.

TABLE II-A

DESCRIPTION OF SCENARIOS AND SIMULATION RESULTS

TABLE II-B

DESCRIPTION OF SCENARIOS AND SIMULATION RESULTS

A simulated fault current for the fault at line 006- 007 between phase A and phase B is shown in Figure 9. At an arc burst instant, which is indicative of the simulated fault, the current suddenly increases to 9A. The current magnitude is limited by the arc resistance and line characteristic impedance.

A voltage transient resulting from the fault is created by the sudden current rise at the arc burst instance. The voltage transient is shown Figure 10. The voltage transient magnitude is around 3kV. The line characteristic impedance can be estimated from current and voltage transient magnitude and is around 330W. The current and voltage transients travel along the

distribution lines as travelling waves.

Tables Ilia and Illb also provide the instances when the VTWs are detected via VTW detector 10 at respective buses (i.e. ToAs) . Figure 12 shows the voltage transients recorded at bus 1 for the simulated arc burst instant.

The calculated dWTC²/dt of aerial-mode voltages at buses 1, 5, 12, 14, 29 and 34 for the fault accruing at line 006-007, 2.7km away from bus 6 (i.e. 9061.7m x 0.3) where t = 0.3, are illustrated in Figure 11. The first local peaks of dWTC²/dt illustrate the VTW arrival times on the respective buses.

In order to identify the fault location, the closest VTW detector 10 / bus to the fault is identified by comparing the ToAs registered by the various VTW detectors

10.

From the fault-scenario results depicted in Figure 11, it can be verified that the VTW detector 10 at bus 14 is the closest sensor to the fault location.

Following the identification of the VTW detector 10 closest to the fault, the 6_*1 column ATD (measured) data set "DTⁿ}n¹ _>n ⁿ)" can now be calculated.

Using the ATD (network) data set

and the S parameter data set

" in conjunction with the

ATD (measured) data set, the pertinent "a" can be obtained for each line section of network 5.

Table IV illustrates the calculated results of "a"s for the various line sections of network 5 arising from the fault occurring on line section 006-007. The table demonstrates that just four of the calculated alphas are in the acceptable range of 0 < a < 1 , namely line sections 006-007, 009-010, 010-011, and 013-015.

TABLE IV CALCULATED ALPHAS FOR THE FAULT AT LINE 006-007

The line sections with 0 < a < 1 are candidates for containing the fault. For the lines which S (ί) _ 0 the fault location is not observable and "a" cannot be

calculated. The remaining lines cannot be a candidate for containing the fault as a is outside the range 0 < a < 1.

The next step is to calculate the cost function for the individual candidate line on its optimum calculated "a" values.

Figure 13 illustrates the calculated optimum cost function for selected lines.

The presence of transformers between a line section "L" and a VTW detector 10 can reduce the accuracy of the various data sets that characterise the network 5,

including the ATD (network) data set. In this example, the increase in VTW propagation time along a line section "L" or between a node Q and a VTW detector 10 due to the presence of one or more transformers or other equipment can be added to the calculated propagation time for the VTW using the following formula:

Where qi_n is the number of transformer between the origin of line "l" and the detector "n".

is the total estimated error caused by distribution transformers, and where:

T_p Z_AC_t

and where Z_A is the characteristic line impedance; and where C_t is the capacitance of the transformer.

Fault type distinction example

An embodiment of a method and system that

distinguishes between a switching event and a high

impedance fault is now described with reference to Figures 1, 15 and 16.

With reference to Figure 1, the data processing system provided at, for example, the centralised data processing station 15 is also configured to identify the occurrence of a high impedance fault. According to an example, a method for identifying a high impedance fault involves identifying the occurrence of VTWs with similar characteristics over consecutive cycles of the baseline AC power distributed by the network 5. In particular, where VTWs with similar characteristics are detected at the centralised data processing station 15 for at least two, and typically multiple, consecutive cycles of the baseline AC power distributed by the network 5 (i.e. at least two, and typically multiple, consecutive cycles of the 60Hz or 50Hz power distributed by network 5) , the centralised processing station 15 may determine that the VTWs originated from a high impedance fault rather than from a switching event within an item of network equipment.

A high impedance fault will typically trigger a VTW as the baseline AC power of the network achieves a

particular voltage, and the VTW will be sustained until the voltage of the baseline AC power cycle drops below a second (and typically different) network voltage.

High impedance faults will typically be sustained over two or more consecutive cycles of the networks baseline AC power cycle, and will typically generate VTWs over consecutive cycles that have similar characteristics, at least in terms of initiation voltage, extinguish voltage and duration.

In contrast, a network switching event will typically only trigger VTWs for the duration of the switching event, which typically is of much shorter duration than a

switching event, and these VTWs will typically have different characteristics at different times during the switching event.

Referring now to Figure 15 there is disclosed a functional block diagram of VTW detector 10s. Components that are the same as components shown in Figures 3 and 4 will be given the same reference numeral. Voltages detected by antenna 305 are passed to anti-aliasing buffer and filter 420/425 before being passed to an analogue to digital (A/D) converter 1505. A/D converter 1505 passes its output signals to circular buffer 1510 and event detector 1515. Circular buffer 1510 temporarily stores the data output from A/D converter 1505. Event detector 1515 analyses the data received from the A/D converter 1505 to determine whether a voltage transient has been detected. Where a voltage transient occurs, Event

Detector 1515 triggers Event Recorder 1520 to receive and store a portion of the data stored in circular buffer 1520. In this way, data associated with a VTW is stored.

Data from event recorder 1520 is passed to

communications port 1540, which transfers recorded event data to the central processing unit 15. As in Figure 3, the communications port 1540 may include a Wi-Fi

transceiver 335 and/or a 3G network connection 340.

A functional block diagram of central processing system 15 is shown in Figure 16. In particular, event data output from VTW detector 10 is transmitted by the communications port 1540 of VTW detector 10, and received by communications port 1605 of central processing unit 15.

Communications port 1605 passes received data to Event Classifier module 1610 and Fault Location module 1615. Output from Event Classifier module 1610 and Fault Location module 1615 are passed to Decision Maker module 1620, which determines how to deal with the detected VTW.

Fault Location module 1615 may operate in accordance with an embodiment of the invention described above in relation to Figures 2a to 14.

Event Classifier module 1610 may operate to

distinguish between VTWs generated by switching events, and VTWs generated by high impedance faults. A network switching event will typically only trigger VTWs during the switching event, which may be of relatively short duration compared to a high impedance fault, and the VTWs triggered by switching events tend to have different characteristics at different times throughout the

switching event. In contrast, high impedance faults may extend for a much longer timer period than a network switching event, and will typically generate VTWs over consecutive cycles that have similar characteristics on a cycle to cycle basis .

The difference in characteristics between VTWs generated by the switching events and VTWs generated by high impedance faults can be used by event classifier 1610 to discriminate between switching events and high

impedance faults.

Decision maker 1620 receives output from Event

Classifier module 1610 and Fault Location module 1615 and associates a high impedance fault with corresponding location information. This information may be passed back to communications port 1605 and transmitted to a circuit breaker or relay if the characteristics of the fault are sufficient to require a power outage. The information may also be communicated to a maintenance register for further investigation and/or maintenance and/or repair.

Communication port 1610 is adapted to receive data from multiple VTW detectors 10/10s associated with network 5. Event Classifier module 1610, and Fault Location module 1615, may also process multiple data streams from multiple VTW detectors 10/10s.

Fault Location module 1615 requires multiple data streams from multiple VTW detectors 10/10s in order to determine the ATD (Fault) data sets.

It will be understood to persons skilled in the art of the invention that many modifications may be made without departing from the spirit and scope of the

invention .

It is to be understood that, if any prior art

publication is referred to herein, such reference does not constitute an admission that the publication forms a part of the common general knowledge in the art, in Australia or any other country. In the claims which follow and in the preceding description of the invention, except where the context requires otherwise due to express language or necessary implication, the word "comprise" or variations such as "comprises" or "comprising" is used in an inclusive sense, i.e. to specify the presence of the stated features but not to preclude the presence or addition of further features in various embodiments of the invention.

Claims

1. A method of estimating the location of a fault on an electricity distribution network, comprising: i) locating a plurality of travelling wave detectors on the network, each travelling wave detector adapted to detect voltage travelling waves (VTWs) or current

2. The method of claim 1 wherein the at least one data set comprises data as to propagation times for travelling waves between a plurality of predetermined points on the network .

3. The method of claim 2 wherein the propagation times between a plurality of predetermined points comprises the propagation time along predetermined sections of the network, wherein each section is located intermediate consecutive points of the plurality of predetermined points .

4. The method of any one of claim 2 or claim 3 wherein the at least one data set further comprises differential data, wherein the differential data comprises data as to the difference in propagation times for a travelling wave from at least one predetermined point on the network and each detector in at least one predetermined pair of detectors .

5. The method of claim 4 wherein the differential data comprises a data set of differential data associated with each one of a plurality of predetermined points on the network .

6. The method of claim 5 wherein the differential data comprises differential data for a plurality of

predetermined pairs of detectors on the network with each one of said plurality of predetermined points.

7. The method as claimed in any one of claims 1 to 6 wherein the data processing system processes the time of arrival data to produce differential data as to the difference in the time of arrival of the travelling wave between at least one pair of detectors.

8. The method of claim 7 wherein the differential data associated with the time of arrival data comprises differential data as to the difference in the time of arrival of the travelling wave between a plurality of pairs of detectors.

9. The method of any one of claim 7 or claim 8 wherein the processing of the at least one data set and the time of arrival data by said data processing system further comprises comparing the differential data of the at least one data set and the differential data of the time of arrival data so as to produce a first parameter indicative of the origin of the travelling wave from at least one of said pre-determined points.

10. The method of claim 9 wherein the method further comprises the steps of:

producing said first parameter for each one of said pre-determined points; and

11. The method of claim 10 further comprising the step of using the parameters in the set of first parameters to determine a line section of the network likely to contain the fault.

12. The method of claim 11 wherein

a cost function value is determined for each pre determined point associated with the first set of

parameters and

the line section of the network associated with the pre-determined point with the lowest cost function value is selected as the line section containing the origin of the travelling wave.

13. The method of claim 12 further comprising the step of applying the first parameter associated with the selected line section to determine the origin of the travelling wave on the selected line section.

14. The method of any one of claims 1 to 13 wherein the at least one data set further comprises transformer propagation delay data that comprises data as to

reductions in travelling wave propagation times on the network arising from transformers connected to the

network .

15. The method of any one of claims 1 to 14 wherein the detectors are located at geographically diverse locations on the network.

16. The method of claim 15 wherein at least some of the detectors are located on the network independently of any sub-stations on the network.

17. The method of any one of claims 1 to 16 wherein the detectors monitor received travelling waves to

discriminate between switching events and fault events.

18. The method of claim 17 further comprising the step of at least one detector performing Clarke's transform on detected voltages so as to determine one or more

parameters of travelling waves detected by the at least one detector.

19. The method of claim 18 further comprising the step of the at least one detector determining modal components of travelling waves from said Clarke's transform.

20. The method of any one of claims 18 or 19 wherein said method further comprises the step of performing discrete wavelet transforms on aerial mode components derived from voltage or current signals detected by the at least one detector, and wherein the discrete wavelet transform is used to determine the one or more parameters of travelling waves .

21. The method of any one of claims 18 to 20 wherein wavelet transform coefficient energies are derived from voltages or currents detected by the at least one detector and the wavelet transform coefficient energies are used to determine one or more parameters of the travelling waves.

22. The method of any one of claims 18 to 21 wherein the one or more parameters include:

i) time of arrival of a travelling wave at the at least one detector;

iii) ground or earth mode energies associated with phase to ground faults on said network.

23. The method of any one of claims 1 to 22 wherein the detectors analyse travelling wave data over multiple cycles so as to discriminate between travelling waves originating from switching events and travelling waves originating from fault events.

24. The method of any one of claims 1 to 23 wherein the method further comprising the step of processing data associated with travelling waves to identify the

occurrence of repeat transient faults originating from substantially the same location on the network.

25. The method of any one of claims 1 to 24 wherein for voltage travelling waves the detectors comprise capacitive sensors and for current travelling waves the detectors comprise Hall effect and/or Rogowski coil sensors, and wherein the defectors are located in close proximity to the network conductors without being in contact with the network conductors .

26. The method of any one of claims 1 to 25 wherein the travelling wave is a voltage travelling wave or a current travelling wave.

27. The method of any one of claims 1 to 26 wherein the network is an overhead network.

28. The method of anyone of claims 1 to 27 wherein the network is an overhead medium voltage network.

29. The method of any one of claims 1 to 28 wherein the network is a radial network

30. A method of detecting a high impedance fault on an alternating current (AC) power distribution network, the method comprising the steps of:

monitoring the network at one or more locations for the occurrence of travelling waves on two or more

consecutive cycles of the baseline AC power of the

distribution network;

analysing travelling waves detected over two or more consecutive cycles of the baseline AC power of the

distribution network to identify similarities in the characteristics of the travelling waves; and

determining that that the travelling waves originate from a high impedance fault where the similarities in the travelling waves repeat, or substantially repeat, for the two or more consecutive cycles.

31. A method of detecting high impedance faults on an alternating current (AC) power distribution network, the method comprising the steps of:

operating travelling wave detectors at one or more locations on the network to monitor network voltage and/or current ;

performing Clarke's transform on voltages or currents detected by said at least one detector so as to identify modal voltages or currents of the network;

analysing at least one of the aerial and ground mode voltages to identify phase to phase and/or phase to ground faults on the network.

32. The method of claim 31 further comprising the step of calculating wavelet transform coefficient energies from at least one of modal voltages and modal currents and using the wavelet transform coefficient energies to determine the occurrence of phase to phase and phase to ground faults on said network.

33. A system for estimating the location of a fault on an electricity distribution network, comprising:

a plurality of travelling wave detectors located on the network, each travelling wave detector arranged to:

detect voltage travelling waves (VTWs) or current travelling waves (CTWs) propagating through the network; and

34. The system as claimed in claim 33 wherein the

plurality of travelling wave detectors are synchronised by satellite navigation signals.

35. The system as claimed in claim 34 wherein the

travelling wave detectors are adapted to time stamp travelling wave data detected by travelling wave detectors and wherein time stamp data for the plurality of

travelling wave detectors is synchronised by satellite navigation data.

36. The system of any one of claims 33 to 35 wherein the data processing system is arranged to receive the time of arrival data from the plurality of travelling wave

detectors, and process the time of arrival data and the at least one data set to estimate the origin of the

travelling wave.