JP2010133747A - Partial discharge discrimination method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To easily discriminate a signal arising from partial discharge from an electric device to be measured from a continuous noise signal. <P>SOLUTION: A sensor 11 for detecting a signal including a partial discharge signal disposed at the electric device to be measured is provided. The signal detection by the sensor 11 is amplified by an amplifier 12, and distributed into a plurality of k pieces by a distributor 13. The distributed signals are inputted to band-pass filters 14<SB>1</SB>to 14<SB>k</SB>, and signals F<SB>1</SB>to F<SB>k</SB>are sent to the outputs of the band-pass filters 14<SB>1</SB>to 14<SB>k</SB>. The output signals F<SB>1</SB>to F<SB>k</SB>are inputted to pulse converters 15<SB>1</SB>to 15<SB>k</SB>, and pulse signals P<SB>1</SB>to P<SB>k</SB>are outputted to the output. Next, each output signal of the pulse converters 15<SB>1</SB>to 15<SB>k</SB>is sent to a signal extraction part 16 for first time analysis, then supplied to a signal processing part 17 for second time analysis, and sent to a determination part 18 for third time analysis. The signal sent to the determination part 18 is compared with data stored in advance in a diagnosis database 19 to determine whether the signal is the partial discharge signal. <P>COPYRIGHT: (C)2010,JPO&amp;INPIT

Description

  The present invention relates to a partial discharge discriminating method that can satisfactorily discriminate between a partial discharge signal and a noise signal generated from an electric device in operation.

  As a method for discriminating between a partial discharge signal and a noise signal generated from an electric device in operation, there is means for measuring a partial discharge signal by selecting an optimum frequency while changing a cutoff frequency (see Patent Document 1). ).

Also, by creating and subtracting averaged data of partial discharge signals, noise signals with the same intensity are erased at the same timing, but random noise signals etc. are not erased, so it cannot be said that the detection accuracy is good (patent) Reference 2).
JP 2004-101418 A JP 07-260868 A

  Measuring the partial discharge signal generated from the electrical equipment to be measured, and determining whether the signal is a partial discharge signal or a noise signal, distinguishing the noise signal from the partial discharge signal is the biggest problem. It has become.

  The method of Document 1 has a problem that a background noise signal can be discriminated but a pulsed noise signal cannot be discriminated. Moreover, it is necessary to carry out the measurement several times, and there is a problem that the measurement takes time. Furthermore, if there is a strong pulse noise signal, it may be erroneously determined to be partial discharge by measuring it.

  For example, in the measurement at the site where the electric device to be measured is installed, various noise signals are measured in addition to the signal due to the partial discharge, and it is quite difficult to distinguish from the noise signal. In particular, a pulse-like signal that is similar to partial discharge is sometimes measured as a noise signal from a peripheral device. In this case, there is a problem that it is not possible to specify whether the signal is a noise signal or a partial discharge signal.

  The object of the present invention has been made in view of the above circumstances, and makes it possible to easily discriminate between a signal due to partial discharge from a measured electrical device and continuous noise, and a pulse similar to that of partial discharge. It is another object of the present invention to provide a partial discharge discrimination method that can also discriminate between noises.

  In order to achieve the above object, in the invention according to claim 1, a sensor for detecting a signal including a partial discharge signal is disposed in an electric device to be measured, and the signal including the partial discharge signal is detected by the sensor. After that, the detected signal is distributed into a plurality of signals, the distributed signal is filtered for each fixed frequency band, converted into pulse signals, and the converted pulse signals are classified according to the number of pulses. The data was extracted by 1 hour analysis, and the signal generation time of the extracted data was divided into ½ period or n period of the applied voltage and the data was processed by the second time analysis, and then obtained by the signal processing. The data distribution pattern is collated with the distribution pattern of the diagnostic database in which the data distribution pattern and noise signal distribution pattern at the time of partial discharge are stored in advance, And judging a partial discharge signal presence generated from measured electrical equipment by third time analysis.

  According to a second aspect of the present invention, in the first aspect, after the data is signal-processed by the second time analysis, the data distribution pattern obtained by the signal processing is a bar graph display of the number of signal generations with respect to the signal generation time of the data. It is characterized by making it.

  According to the present invention, after amplifying the signal detected by the sensor, the detected signal is distributed to a plurality of signals, filtered for each frequency band, converted to a pulse signal, and a target signal is extracted from the converted signal. After that, by comparing with the data distribution pattern and noise signal distribution pattern at the time of partial discharge occurrence stored in the diagnostic database, the signal from the partial discharge from the electric device to be measured and continued for a certain period of time Thus, there is an advantage that a continuous noise (base noise or the like) signal can be easily discriminated and a pulsed noise signal similar to the partial discharge signal can be discriminated.

  Further, since the data distribution pattern obtained by the signal processing is displayed as a bar graph, the determination of the partial discharge signal distribution and the noise signal distribution becomes clear. Since the configuration can be simplified, there are also advantages that the analysis time is shortened, and that it is compact, lightweight and economically advantageous.

  Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of the present invention. In FIG. 1, reference numeral 11 denotes a signal detection sensor 11 including a partial discharge signal arranged in an electric device to be measured (not shown). The sensor 11 is for measuring a pulse current generated with a partial discharge from the electric device to be measured. For example, a high-frequency CT attached to the ground wire of the electric device to be measured or a ground wire of the electric device to be measured It consists of a magnetic field probe or the like arranged in the vicinity. The sensor 11 may be an antenna that is disposed in the vicinity of the electric device to be measured and detects an electromagnetic wave or the like generated with the partial discharge.

The signal detected by the sensor 11 is amplified to a certain level by the amplifier 12 and then input to the distributor 13 and the output is distributed into a plurality of k pieces. Distributed signal is input from a plurality of k to the band filter 14 ₁ to 14 _k configured, the output of the bandpass filter 14 ₁ to 14 _k, signal F _1, F ₂ ··· F _k is sent Is done. Here, we bandpass filter 14 ₁ to 3 divided by the distributor 13 as k = 3, 14 _2, 14 ₃ of the measurement frequency points _{f 1, f 2, f 3} 50MHz, 100MHz, and 150 MHz.

Of the signals F ₁ , F ₂ , and F ₃ (examples shown in FIG. 2 (a)) sent to the outputs of the band-pass filters 14 ₁ , 14 ₂ , and 14 ₃ , the signals F ₁ and F ₂ have a time t _1. is output to t _2, t _3, the signal F ₃ is output to the time t _1, t _3.

Bandpass filter 14 _1, 14 _2, 14 ₃ signal F ₁ outputted from, F _2, F ₃ are inputted to the pulse converter 15 _1, 15 _2, 15 _3. The pulse converters 15 ₁ , 15 ₂ , and 15 ₃ output the pulse signals P ₁ , P ₂ , and P ₃ shown in FIG. . By this pulse signal conversion, high frequency components are cut, data is simplified and handling becomes easy. Note that the pulse converter, in the above description has dealt with the case of three, the pulse converter is provided a plurality of k equal to the band-pass filter 14 ₁ to 14 _k.

Next, after each output signal of the pulse converters 15 ₁ to 15 _k is input to the signal extraction unit 16, a first time analysis described later is performed, and a signal is output from the signal extraction unit 16, Input to the processing unit 17. In the signal processing unit 17, after a second time analysis described later is performed, a third time analysis described later is performed by collating with a distribution pattern supplied to the determination unit 18 and stored in the diagnosis database (diagnosis DB) 19. Is implemented.

In the above embodiment, the simple time-frequency analysis unit 20 is configured by the distributor 13, the band-pass filters 14 _{1 to} 14 _k , and the pulse converters 15 ₁ to 15 _k .

  Next, a method for measuring the pulse current based on a signal including a partial discharge signal from the electric device to be measured by applying, for example, high frequency CT to the sensor 11 and analyzing the measured pulse current will be described below.

  3 and 4 are characteristic diagrams of the results of time-frequency analysis of data measured by attaching a high-frequency CT to the ground line of the electrical equipment to be measured in actual fields A and B. In both figures, the x-axis represents the frequency. , Y-axis is time, and z-axis is signal output.

3 and 4, A1 and B1 are base noise signals that are always generated regardless of time, A2 and A3 are pulse signals, the signal A2 has a narrow frequency width, and the signal A3 has a frequency width. B2 and B3 are pulse signals, the signal B2 is a signal with a wide frequency width, and the signal B3 is a signal with a narrow frequency width. This signal is sequentially input to the distributor 13, the bandpass filters 14 ₁ , 14 ₂ , 14 ₃ , and the pulse converters 15 ₁ , 15 ₂ , 15 ₃ .

First, the processing in the signal extraction unit 16 in the first time analysis will be described by taking an example in which there are three signals. Pulse output signals P ₁ , P ₂ , P ₃ from the pulse converters 15 ₁ , 15 ₂ , 15 ₃ are sent to the signal extraction unit 16. In this signal extraction unit 16, w3 in the case of only one of the three pulse output signals, w2 in the case of two signals, and w3 in the case of three signals, for each number of pulses (w1, The signals are classified as w2 and w3) and extracted as data. The same effect as that obtained by classifying and extracting for each frequency width can be obtained. (The noise signals A1 and B1 shown in FIGS. 3 and 4 are cut from the extraction result.) In FIG. 2B, the times t ₁ and t ₃ are classified as w3 and the time t ₂ is classified as w2.

  Next, the processing of the signal processing unit 17 in the second time analysis will be described. Paying attention to the signal generation time of the data extracted by the signal extraction unit 16, the data is processed by the second time analysis by dividing the applied voltage every 1/2 cycle or every n cycles (n: integer of 1 or more). Then, as shown in FIG. 5 and FIG. In FIG. 5 and FIG. 6, it was set as 1 period (20 ms since it is 50 Hz) of the applied voltage.

  In the data display method of FIGS. 5 and 6, the origin of the graph was set to “0 ms (start)”, and the generation time of the data extracted by the signal extraction unit 15 of the first time analysis was displayed as follows.

“Displays 0 to 20 ms in the x-axis direction at x = 0, y = 0 to x = 20 ms, y = 0. 20 to 40 ms in the x-axis direction from x = 0, y = 1 to x = 20 ms, y = 1 Display 40-60 ms in the x-axis direction from x = 0, y = 2 to x = 20 ms, y = 2 ”
The above is sequentially repeated to display data as shown in FIGS.

  Next, the process of the determination unit 18 in the third time analysis will be described. The data distribution pattern sent from the signal processing unit 17 of the second time analysis is compared with the distribution pattern stored in the diagnostic database 19, and finally the noise signal is cut to extract a signal due to partial discharge, The third time analysis determines whether or not the signal is a partial discharge signal from the electrical device to be measured.

  For example, in the case of FIG. 5, since w1 is distributed regardless of the applied voltage, it can be determined as a noise signal (noise signal A2 is cut), and w2 is determined as a single noise because of its small number. Then, w3 is determined to be a typical distribution pattern when partial discharge occurs from one phase, and from these determinations, the distribution of w3 is determined to be a partial discharge signal.

  In the case of FIG. 6, since w1 is small, it is determined that it is a single noise, w2 is determined to be a typical distribution pattern when partial discharge occurs from one phase, and w3 is constant. Since it is periodically distributed over time, it is determined that it is a periodic noise signal (noise signal B2 is cut), and from these determinations, the distribution of w2 is determined as a partial discharge signal.

  As described above, it is possible to discriminate the noise signal distribution pattern from the distribution pattern at the signal generation time obtained by classifying the pulse-like signal for each number of pulses, that is, for each frequency width. In the field A shown in FIG. 3, the partial discharge signal A3 has a wider frequency width than the noise signal A2, but in the field B shown in FIG. 4, the noise signal B2 has a frequency width larger than that of the partial discharge signal B3. There was a wide signal. As a result, it is possible to extract even a partial discharge signal whose frequency width is narrower than the noise signal as in the latter field B.

  In addition, it distinguishes continuously generated noise signals such as base noise, noise signals from pulse-like peripheral devices that are similar to partial discharge signals, random noise signals that do not depend on applied voltage, etc. Accurate measurement is possible by measuring.

  FIG. 7 and FIG. 8 are bar graphs showing the results of distribution distribution display by counting the number of occurrences for each time zone by dividing the x-axis shown in FIG. 5 and FIG. 6 into m (m: integer of 2 or more). By displaying the bar graph in this way, it becomes possible to clearly determine the partial discharge signal and the noise signal. 7 and 8 show examples in which 20 ms is divided into 20 parts.

FIG. 1 is a block diagram showing an embodiment of the present invention. (A) is an explanatory diagram showing the output signal of the bandpass filter, (b) is an explanatory diagram showing the output signal of the pulse converter. The characteristic view of the result of having performed the time-frequency analysis of the data measured by attaching high frequency CT to the ground line of the electric equipment in the field A. The characteristic view of the result of having performed the time-frequency analysis of the data measured by attaching high frequency CT to the ground line of the electric equipment in the field B. The time graph with respect to the signal generation time in the field A. The time graph with respect to the signal generation time in the field B. The bar graph of the signal generation time zone in the field A. The bar graph of the signal generation time zone in the field B.

Explanation of symbols

11 ... sensor 12 ... amplifier 13 ... distributor 14 ₁ to 14 _k ... bandpass filter 15 ₁ to 15 _k ... pulse converter 16 ... signal extracting section 17 ... signal processor 18 ... judging unit 19 ... diagnostic database 20 ... Time - Frequency Analysis department

Claims (2)

  A sensor that detects a signal that includes a partial discharge signal is placed in the electrical device to be measured. After detecting a signal that includes a partial discharge signal with this sensor, the detected signal is distributed to multiple units, and the distributed signal is constant. Are converted into pulse signals after being filtered for each frequency band, and the signals are classified as the number of pulses from the converted pulse signals and extracted as data by the first time analysis, and the signal generation time of the extracted data Is divided every 1/2 period or n period of the applied voltage, and the data processing is performed by the second time analysis, and then the data distribution pattern obtained by the signal processing and the data distribution pattern or noise signal at the time of occurrence of partial discharge in advance The distribution pattern is collated with the distribution pattern of the diagnostic database in which the distribution pattern is stored, and from the result of the collation, the presence or absence of the partial discharge signal generated from the electric device to be measured is third. Partial discharge determination method characterized by determining by between analysis.   2. The partial discharge determination according to claim 1, wherein after data processing is performed by the second time analysis, the data distribution pattern obtained by the signal processing displays a bar graph display of the number of signal generations with respect to the signal generation time of the data. Method

Images