METHOD AND APPARATUS FOR DETECTING AN EVENT
FIELD OF THE INVENTION
This invention relates to a method and apparatus for detecting an event.
BACKGROUND OF THE INVENTION
High voltage outdoor insulators, bushings and terminations deteriorate over time due to exposure to pollution, moisture, heat and UV radiation. By way of illustration, the recent prolonged dry spell and hot summers in Australia accelerated degradation of insulators in transmission lines resulting in an increased incidence of structural damage and decreased dielectric strength. These lead to transmission line faults. It would be desirable to detect faulty transmission line insulators in service. Present methods of detection include visual inspection, image intensification, infrared thermography, electric field distribution measurement and measurement of acoustic emission. The disadvantage of these techniques is that they allow only very large defects to be detected. They are not capable of picking up signs of deterioration caused by more subtle changes of the insulating material. They rely only on spot measurements and therefore an impractical amount of effort is needed to observe performance trends over time. Furthermore, some of these techniques (such as visual inspection and infrared thermography) are not able to detect defects located away from direct line of sight of the sensing device. Apart from technical difficulties associated with each method, there does not exist a widely accepted standard set of reliable condition monitoring measurements. The problem is further enhanced by difficulty in inspecting large parts of transmission lines running through rugged terrain.
SUMMARY OF THE INVENTION
The invention is based on the concept of detecting electromagnetic radiation emitted from electrical apparatus and indicative of an event such as a partial discharge. Partial discharge events not symptomatic of substantial breakdown of electrical insulators and other
electrical apparatus frequently precede, by some substantial time, subsequent substantial breakdown of the apparatus. By detecting such events, the apparatus in question may possibly be repaired or decommissioned in advance before substantial damage or inconvenience arises.
In certain aspects, the invention is concerned with detection of such partial discharge events, but it is also applicable to detection of other kinds of event.
In one aspect, the invention provides a method of detecting an event in electrical apparatus wherein detection is effected by detecting electromagnetic radiation from the apparatus characteristic of said event to generate an electrical signal representing the electromagnetic radiation, and subjecting the signal to non-stationary wave signal analysis to generate an output indicative of said detecting.
The invention also provides an apparatus for detecting an event in electrical apparatus, having detection means for detecting electromagnetic radiation from the apparatus characteristic of said event, means for generating an electrical signal representing the electromagnetic radiation, and means for subjecting the signal to non-stationary wave signal analysis to generate an output indicative of said detecting.
BRIEF DESCRIPTION OF THE DRAWINGS
The invention is further described by way of example only with reference to the accompanying drawings in which:
Figure 1 is a front view of a tower supporting lines of a high voltage power transmission system, fitted with an apparatus formed in accordance with the invention, for monitoring faults in the transmission system.;
Figure 2 is a diagrammatic perspective view of towers and associated components of the power transmission system and monitoring apparatus of Figure 1,
Figure 3 is a block diagram of processing steps executed by a processing unit incorporated into the apparatus of Figure 1;
Figure 4 is a block diagram of components of an alternative receiver apparatus useful in the apparatus of Figure 1 :
Figure 5 is graphical representation of experimental results of fractal analysis performed in a method of the invention;
Figure 6 is a diagram of a power substation incorporating apparatus in accordance with the invention;
Figure 7 is a diagrammatic perspective view of a power transmission system incorporating another apparatus in accordance with the invention; and
Figure 8 is a block diagram illustrating processing steps in a receiver apparatus forming part of the apparatus of Figure 7.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
Figures 1 and 2 show a high voltage power transmission system 120 having pylons 160 which support transmission wires 140, 142 144 via connection insulators 146, 148, 150. This embodiment of the invention has apparatus 125 arranged to detect electromagnetic radiation from the insulators, arising from partial discharge at the insulator, in turn characteristic of the beginning of development of substantial breakdown of the insulator.
The phenomenon of partial discharge in electrical apparatus is for example the subject of the publication "IEE Guide to Partial discharge measurement in Power Switchgear", (IEEE Std 1921-1993) published 1993 by the United States Institute of Electronics Engineers,
Inc. and the publication "draft American National Standard guide for partial discharge in
liquid-filled power transformers and shunt reactors" (ANSI/IEEE C57.113) issued January 1998 for trial use by the United States Institute of Electronics Engineers. The disclosures of these publications are hereby incorporated to form part of the disclosure of the present specification. The latter publication defines "partial discharge" in the following terms:
"Partial discharge: An electric discharge that only partially bridges the insulation between conductors".
It is in this sense that the term is used in the present specification. The publications describe methods of measuring partial discharge by detection of pulses in electric current applied to the apparatus under consideration, and arising from the discharge. The determination of whether detected pulse events are characteristic of existence of partial discharge may be made by reference to the energy of detected pulses. The mentioned draft document ANSI/IEEE C57.113 suggests that under described test conditions a tested component should be considered as not being faulty if:
a) The acceptable energised background noise level associated with a measurement does not exceed 50% of the acceptable terminal partial discharge level of the component, in any case being below lOOpC,
b) Any detected increase in partial discharge levels during a 60 minute test period do not exceed 30% of the acceptable terminal partial discharge level, and
c) The partial discharge levels during the 60 minute test period do not exhibit any steadily rising trend and there is no sudden, sustained increase in levels during the last 20 minutes of the test period.
Criteria of this nature may be applied in the present invention, although, the criteria may be varied to suit the particular circumstances. The criteria may be best selected by experimentation, or on the basis of general experience concerning the component in question. However, at least adoption of a threshold energy level of lOOpC or a value
within one or two orders of magnitude of this is likely to be generally applicable.
Apparatus 125 has electromagnetic sensors 152, 154, 156 positioned at each pylon and connected to a respective processing unit 162 on or adjacent the tower.
The sensors 152, 154, 156 may be directional antennas preferably having a wide bandwidth characteristic of eg IMHz to 3GHz. Each antenna may be in the form of a bi- conical, horn, log-periodic, dipole or any other form of broadband antenna which is preferably mounted close to the equipment that is to be monitored. The directional characteristics of the antenna should be such as to allow high gain to be achieved in the main antenna lobe. The design is preferably rugged and corrosion resistance.
The positioning and directional sensitivity of the sensors is such that each sensor 152, 154, 156, is sensitive to electromagnetic radiation largely only from one respective insulator of the associated pylon 160. As shown in Figure 5, the sensors 152, 154, 156 have narrow directional sensitivity envelopes 170, 172, 174. Each sensor is positioned below a respective insulator, with its sensitivity envelope directed upwardly so as to substantially encompass only a single one of the insulators 156, 148, 150.
Electric signals are generated by the sensors 152, 154, 156 when electromagnetic radiation from the respective insulators is incident thereon. The processing units 162 process signal from the associated sensors to determine if the signal exhibits signal characteristics indicative that detected electromagnetic radiation is characteristic of partial discharge at an insulator. If so, a signal is generated indicative that a fault exists. That signal is passed to a high speed transmitter 162a associated with the particular processing unit, so that the transmitter transmits a corresponding signal from an antenna 186 forming part of the processing unit. The transmitter may for example be a radio transmitter or a microwave transmitter. The transmitted signal includes data as to which pylon the partial discharge has occurred at, and which particular insulator of that pylon is faulty. The transmitted signal is received at a base station 188 which generates a suitable alarm signal. On generation of the alarm signal, a maintenance crew can identify the fact that a significant
partial discharge has occurred at a particular insulator on a particular pylon, eg permitting the crew to quickly attend to replacing the affected insulator.
The operation of the signal processing units 162 is now described with reference to Figure 3.
The first processing step performed by a processing unit 162 is the step of data acquisition 200, performed by accepting electrical signals from the sensors 152, 154, 156.
After data acquisition, a step 202 of signal filtering is carried out using a digital or analog adaptive noise filter. The pass band of the filter can be adjusted in accordance with the noise level of the installation site. More advanced signal filtering by use of eg a wavelet filter and/or a Kalman filter may be implemented, to reduce noise. Signal possibly containing pulses due to incidence of radiation indicative of partial discharge is then processed at step 204, at which signal segmentation is performed. This is effected to reduce the processing time and data storage space. The data in the filtered signal derived at step 200 is segmented by implementation of a signal segmentation algorithm based on the energy measurement of the signal, E with respect to the estimated background energy, R. The statistical estimation of the steady background noise from the surroundings may be obtained from a nearby site with relative less interference. Segmentation may be effected in accordance with the following equation:
Energy = E(n) = Σ [ (x(n-l) * w(m)]2 / R
Where: x is data to be analysed, w represents the width of m n represents the number of iterations of E m iterates fom 1 to n, and
R is the average square value of the background noise.
The segmentation process can be used to find out the time instant when a partial discharge occurs. Detailed analysis of each segment may be effected in order to explain the characteristic of the signal. The segmentation process may also eliminate significant amounts of unwanted data from the raw signal and thus improve the speed of computation and transmission.
Next, at step 206, the data describing individual pulse arising from partial discharge events, as derived at step 204, is analysed in the frequency domain by using Fast Fourier Transform. Since the characteristics of certain defects on the high voltage equipment only display on certain frequency bands, this FFT allows identification and discrimination of specific types of signals originating from eg defective equipment and existing communications channels. Particularly, existing communications channels such as broadcast transmissions occupy a specific frequency band, eg FM band, aeronautical communications band, mobile phone band and CB radio band. Removal of these essentially periodic signals, obviously not associated with partial discharge events, is very useful for subsequent processing operations.
Signal having components removed therefrom by the FFT processing last mentioned is then subjected to non-stationary wave signal analysis, to determine if detected signal is indicative of occurrence of a partial discharge event. Two forms of such analysis are fractal analysis and Wigner-Ville Distribution analysis. Implementations of these are next described.
Fractal analysis may performed using the fractal algorithm described below. This is derived from an algorithm proposed by T. Higuchi, "Approach to An Irregular Time Series on The Basis of the Fractal Theory", Physica D., pp. 277-283, 1988, the disclosures of which are hereby incorporated to form part of this specification.
The derived fractal dimension D is determined from the slope of the curve Lm(k) plotted against £ on a doubly logarithmic scale as further described below. The algorithm proposed by T. Higuchi for calculating the fractal dimension has been adapted for use in
this embodiment of the invention because it is regarded as being a stable estimator of fractal dimension. The Higuchi algorithm is able to calculate the highest, lowest and average fractal dimensions, and any one or more may be utilized in practicing this invention, it has been found convenient to use the average fractal dimension. Particularly, discrimination of different type of defects can be performed using the fractal analysis method.
The basis of this fractal analysis is a finite set of time series observations taken at a regular interval, X(I), X(2), X(3), X(4) X(n). From the given time series, a new time series, X™is to be derived as follows (m = 1, 2, 3, ,k ):
~N - m
X™ = X(m),X(m + k),X(m + 2k),....,X(m +
(2) where both k and m are integers and indicate the initial time and the internal time respectively. The next step is to determine the length of the curve, Lm(k)
LJk) l k
The fractal Dimension D is the slope of the curve Lm(k) plotted against & on a doubly logarithmic scale.
In a practical implementation, the fractal dimension may have a value in the range 1 to 2, with a value of 2 being indicative that no partial discharge event has been detected, a value of 1.8 or lower such as in the range 1.3-1.8 being indicative of detection of a partial discharge event. It has been found that it is possible to deduce from the measured fractal dimension characteristics of an insulator beyond that the insulator is at the point of onset of failure. For example, Figure 5 shows a plot of the determined fractal dimension, D, for various conditions of a cracked insulator known to exhibit characteristics evidencing likely onset of substantial failure. The fractal dimension was found to have a value of about 1,4
1.8 under the condition that the insulator was dry and clean, about 1.6 under the condition that it was wet and clean, and about 1.8 under the condition that it was wet and polluted by application of a surface layer having a surface conductance of about 10 μS.
Analysis by the Wigner Ville Distribution (WVD) method is described in the publication by Shie Qian, "Introduction to Time-Frequency and Wavelet Transforms", Prentice-Hall, the disclosures of which are hereby incorporated to form part of this specification.
WVD is a joint time-frequency representations of a signal. WVD is defined as
Wx is the Wigner Ville Distribution of signal x(t) in time domain, * is the convolution with respect to time, and v is the frequency distribution of signal x(t) .
The Wigner- Ville Distribution is always real-valued. It preserves time and frequency shifts and satisfies relevant marginal properties. Analysis by use of the Wigner- Ville Distribution has been found provide a good representation for a non-stationary and time- varying signal like the electromagnetic pulses emission of which is characteristic of the partial discharge at an insulator 152, 154, 156. Using the WVD, information in the time and frequency domain can be visualized in one single graph, and discrimination of different type of defects can deduced from that. Particularly, it is possible to determine whether a partial discharge event has occurred. As with the mentioned fractal analysis, it is also possible to deduce other aspects of the condition of the insulator from the WVD obtained.
The WVD presented for example graphically, differs significantly for articles such as surge arrestors and insulators, depending on for example whether they are wet or dry or clean or polluted. Each also exhibit readily distinguishable patterns in the cases where there is no
substantial partial discharge and the case where there is such partial discharge. A description of differences, including illustrations, is contained in the publication Wong, K.L.; Electromagnetic emission based monitoring technique for polymer ZnO surge arresters IEEE Transactions on Dielectrics and Electrical Insulation Volume 13, Issue 1, Feb. 2006 Page(s):181 - 190 , the contents of which are hereby incorporated to form part of the disclosure of the present specification.
Analysis of the obtained WVDs for purposes of distinguishing the occurrence of a partial discharge event, and other matters as mentioned between may be effected by use of pattern recognition processes such as described in the paper "Statistical Pattern Recognition: A Review", Anil K. Jain, Robert P.W. Duin and Jianchang Mao, IEEE Transactions on Pattern Recognition, VoI 22, No. 1, January 2000, the contents of which are herby incorporated to form part of the disclosure of the present specification.
Signal from the processing step 206 is passed to the RF transmitter 162a for outward transmission via antenna 186 thereof of a radio signal containing the signal data. The signal data may be in the form of an alarm signal when one or a sufficient number of partial discharge events within a predetermined time is detected, as well as identification of the particular insulator with respect to which alarm applies. Additionally, or alternatively, other information about particular insulators may be transmitted, such as information about the wet/dry and polluted/non-polluted conditions mentioned. When received at the receiver, a suitable alarm may be generated, enabling an insulator for which a problem has been identified in this way to be attended to, such as by replacement of the insulator.
In is not essential that all the processing above described be effected at the processing units 162. For example, Figure 4 illustrates an implementation of the base station 188, in which steps shown in the broken line box 227 in Figure 3 are instead performed at the base station. Particularly, the signal sent from the processing units is signal directed directly from the acquisition system 200 to the transmitter 162. Then, at the base station, received data signal is passed to a high speed data acquisition system 240. Resultant data signal from the system 240 is filtered at a filtering step 242. The output from the filtering step
242 is then processed and stored at step 244, with resultant output being passed to a conventional Supervisory Control And Data Acquisition (SCADA) System 246, commonly used in power systems for delivering data for data acquisition, and supervisory purposes.
As mentioned, these embodiments of the invention employ non-stationary wave signal analysis of which two specific forms of analysis, Fractal Analysis and Wigner-Ville Distribution analysis have been described. A single such analysis, such as fractal analysis or WVD analysis may be employed, but it is preferred to employ a plurality of such analyses independently, such as to employ both fractal and WVD analysis, with an event being taken to have been positively identified on the basis of a poll of the results of each. In the case where fractal and WVD analyses are performed independently, an event may be taken to have been identified only in the case where the results of both analyses are so indicative.
Figure 6 illustrates an implementation of the invention in a high voltage substation 220. Here directional antennae 222 like those forming sensors 152, 154, 156 are employed. Each antenna is sensitive to electromagnetic radiation from a separate substation insulator 228 which supports a sub station high voltage power line 230 in the substation. Electrical signal from detected electromagnetic radiation at the antennas is transmitted to a signal digitization and processing unit 224 by interconnecting communications links 226. The unit 224 may operate to process these signals in the same way as the described processing unitsl62.
The embodiments of the invention described above rely on the provision of sensors in the nature of antennas which receive the electromagnetic signals generated at a fault. However, where the fault to be detected is in a component physically associated with an adjacent electrical conductor forming part of the apparatus in which a fault is to be detected, or in another apparatus in physical proximity to the apparatus in which fault monitoring is to be effected, it may be possible to utilize that conductor as an antenna. It is possible to modify the embodiment of Figures 1 to 4, for example, to utilise the power
transmission wires 140, 142, 144 as sensor antennas. ■ Particularly, it is observed that electromagnetic radiation emitted at onset of a fault at an insulator 146, 148, 150 will cause generation of measurable, corresponding electric signals in the wires. The signals are strongest in the wire closest to the affected insulator. In another embodiment of the invention, the occurrence of a fault in an insulator is effected by detection of that resultant signal. Figure 7 shows diagrammatically a power transmission system 280 having detection apparatus which is effective to determine existence of an insulator fault in this way, and to determine the location of the fault by sensing times of arrival of the detected fault signal at spaced detector locations along the associated power wire.
The apparatus 282 includes voltage detectors Sl, S2, S3, S4, S5, S6, S7, S8 two of which are disposed on each wire 288 of the transmission line, at locations that are predetermined distances, dl, d2 from a known location on each line, for example from the support locations of one set of insulators 286 of the transmission line. The detectors include processing units like the processing units previously described. At occurrence of a partial discharge at an insulator, voltage pulses 290 are generated in the adjacent wire 288 pursuance to incidence thereon of electromagnetic radiation generated by the partial discharge. These pulses propagate in opposite directions from the fault location, along the associated wire 288. These are detected at the detectors 284 associated with thatwire, and are processed in the same way as the in the previous embodiments, and resultant signals indicating occurrence of the discharge are transmitted to a base station 295. Base station 295 may perform the processing steps such a described with reference to Figure 3 or Figure 4. Figure 8 illustrates an exemplary base station 295. It executes data acquisition and signal processing steps 292, 294 similar to the steps to Figure 3. However, hi this case a further comparison step 296 is executed at which the times of detection of fault signals by the detectors 284 are compared and from this the distances dl and d2 are deduced to identify the particular insulator 286 that has developed a fault. This result is provided as the output 298 of the processing steps as illustrated in Figure 10.
The arrangement shown in Figures 7 and 8 may have a series of detectors 284 at relatively far spacing along each wire 288, to enable identification of one of a number of intervening
insulators between each successive pair of detectors. Practically, the maximum effective distance between a pair of detectors is limited by the ability of thethese to reliably detect signal pulses 290, which pulses are naturally progressively attenuated by distance of travel through the wires 288.
While the invention has been described insofar as it is applicable to detection of partial discharge in insulators in power transmission lines and high voltage substations, it may be applied in any context where detection of partial discharge is required, such as for purposes of providing early warning of faults in electrical apparatus is required. For example, as mentioned, it is applicable to detection of faults in surge arresters and power line conductors.
In the described embodiments, output indicative detection of a partial discharge event is passed to the mentioned SCADA system. Alternatively, that output may be passed to any suitable digital or other communications system.
The invention has been described in the context of detection of partial discharge events. It is however applicable to detection of other events giving rise to emanation of radiation, such as surface discharge and fiashover events, and electrical discharge events in general.
The reference to any prior art in this specification is not, and should not be taken as, an acknowledgment or any form of suggestion that that prior art forms part of the common general knowledge in Australia.
Throughout this specification and the claims which follow, unless the context requires otherwise, the word "comprise", and variations such as "comprises" and "comprising", will be understood to imply the inclusion of a stated integer or step or group of integers or steps but not the exclusion of any other integer or step or group of integers or steps.
The invention has been described by way of example only and many modifications and variations may be made thereto without departing from the spirit and scope of the
invention, which includes every novel feature and combination of features herein disclosed.