CN102707203A - Discriminating and measuring method for partial discharge modes of transformer - Google Patents

Abstract

The invention discloses a measurement method for partial discharge pattern recognition of a transformer. The discharge conditions of different partial discharge models are measured in a measurement device, and the measurement results are collected and analyzed to make a map, and the fingerprint is calculated to form a fingerprint library. For the unknown type of discharge in the transformer, it is only necessary to use the measuring device to collect and analyze it to make a map, calculate the fingerprint and compare it with the fingerprint information in the fingerprint database to identify the partial discharge type of the transformer. The invention establishes a transformer partial discharge fingerprint database, uses the information provided by various discharge spectrograms, can distinguish different types of discharges, provides valuable data for the application of fingerprint diagnosis and pattern recognition technology, and has good engineering practical value , with a confidence level of 99.5%.

Description

Measurement Method for Partial Discharge Pattern Recognition in Transformers

technical field

The invention relates to the field of power transformer fault diagnosis, in particular to a measurement method for transformer partial discharge pattern recognition.

Background technique

The insulation structure of power transformers in operation is complex, and there are many types of internal discharge points and discharge types that may occur, usually as follows:

(1) Oil gap discharge in the middle of the winding oil-baffle insulation;

(2) Oil gap discharge at the end of the winding;

(3) Oil gap discharge in contact with insulated wires and insulating paper (lead insulation, lap insulation);

(4) Partial discharge in oil-paper insulation such as lead wires and lap wires;

(5) Oil gap discharge between coils (longitudinal insulation);

(6) Partial breakdown of inter-turn insulation;

(7) The insulating paper slides and flashes along the surface.

Most of the discharge parts are on certain oil gaps, oil wedges, air gaps, metal bodies with floating potentials, sharp corners of conductors and solid surfaces. Therefore, it can be mainly classified into a variety of typical discharges such as point discharge in oil, oil-barrier discharge, bubble discharge in oil, internal void discharge in paper (cardboard), paper (cardboard) along surface discharge, and suspension potential discharge. form.

The digital measurement provides a powerful method for the research of partial discharge, which makes the partial discharge measurement technology enter a new stage. Through digital measurement, research and inspection personnel can understand the insulation condition of transformers more accurately and easily, so that the identification of partial discharge no longer relies mainly on the test personnel's experimental experience as before. So far, the digital measurement is basically carried out for the pulse current method, which is tested according to the IEC60270 standard, which has many obvious advantages, but there are still some shortcomings: the measurement frequency is low (usually below 1MHz), and a large number of partial discharge characteristics are lost information; when used for online detection, it is susceptible to on-site interference; the existing standard fingerprint (discharge pattern composed of statistical calculations) library provides too few standard discharge patterns, and it is difficult to diagnose comprehensively and accurately.

Contents of the invention

The invention designs a measurement method for identifying the partial discharge pattern of a transformer that can identify the discharge type of the discharge in the transformer.

The method for measuring the partial discharge pattern recognition of a transformer is characterized in that it specifically includes the following steps:

1) Build a partial discharge model:

According to the test results in the laboratory, the discharge types in the transformer are classified into five categories: paper (cardboard) internal void discharge, paper (cardboard) surface discharge, floating potential body discharge, oil bubble discharge, oil-barrier type discharge;

① Construct the internal void discharge model of paper (cardboard)

Use plate-plate electrode and test sample 1. Test sample 1 is two layers of 1.0mm thick oil-impregnated insulating cardboard sandwiching a layer of 0.5-1.5mm thick insulating cardboard. Test sample 1 is placed on the two plates of the plate-plate electrode. The electrodes are immersed in transformer oil; due to several layers of medium, there are a large number of air gaps between the medium and the electrodes. Under the external high voltage, internal discharge occurs in these air gaps, simulating the internal gap discharge of cardboard;

② Construct paper (cardboard) creeping discharge model

Using a plate-plate electrode with a distance of 15mm and test sample 2, test sample 2 is a pretreated 2.5mm thick insulating cardboard, test sample 2 is placed between the two plate electrodes of the plate-plate electrode, immersed in transformer oil, and simulated Surface discharge of insulating cardboard;

③ Constructing a floating potential body discharge model

Column electrode, plate electrode and sample three are used, sample three is a metal conductor, sample three is placed between the plate electrode as the suspension electrode and the column electrode as the high-voltage electrode, and the distance between the column electrode and the plate electrode is selected as 4mm; the whole model is immersed in transformer oil to simulate the floating potential discharge;

④Construct the bubble type discharge model in oil

Plate electrode, column electrode and test sample 4 are used. Test sample 4 is 1.0mm dry insulating cardboard. Test sample 4 is sandwiched between plate electrode and column electrode and immersed in transformer oil. Since test sample 4 has not been immersed in oil beforehand, Therefore, there will be more air bubbles on the surface of the cardboard and inside the cardboard. When the electrode is pressurized, it is conducive to the occurrence of bubble discharge in oil, simulating the discharge of bubbles in oil;

⑤ Constructing an oil-barrier discharge model

Plate electrode, pointed electrode and test sample 5 are used. Test sample 5 is an oil-soaked cardboard with a thickness of 1mm. Barrier discharge;

2) Build the measuring device

The measuring device for transformer partial discharge pattern recognition includes an experimental power supply, which is connected to an auto-transformer, connected to an isolation transformer after output from the auto-transformer, and connected to a non-PD test transformer after output from the isolation transformer. The partial discharge test transformer is connected in series with a low-pass high-resistance impedance and then connected in parallel with an electrostatic voltmeter, then connected in parallel with the coupling capacitance and detection impedance connected in series, and then connected in parallel with the partial discharge model in step 1) respectively; The partial discharge model is placed in parallel in a closed transformer box, the box is filled with transformer oil, and the box shell is grounded to form a shielding structure; UHF sensors are installed on the box wall, and the UHF sensors pass through the cables and frequency spectrum respectively. The analyzer is connected to the signal conditioning unit, and the spectrum analyzer and the signal conditioning unit are connected to the industrial control computer;

3) Set the parameters of the device and measure the discharge parameters of the transformer partial discharge model

① Collect the broadband frequency domain data corresponding to partial discharge model discharge, compare it with the background noise, select the optimal frequency as the center frequency, adjust the center frequency of the spectrum analyzer to the optimal frequency, and set the bandwidth to about 5MHz;

②The high-voltage power supply respectively supplies power to the partial discharge model in step 1) through the high-voltage bushing;

③Electromagnetic waves generated by the discharge of each partial discharge model are respectively received by the UHF sensor installed on the wall of the box, and sent to the spectrum analyzer and signal conditioning unit respectively through the 50Ω measurement cable, and are amplified and filtered by the frequency mixing After the modules are processed, the industrial control computer collects and records the discharge quantity q, voltage u, and discharge time t through the NI5112 acquisition card;

④ Repeat step ③ for more than five times for each partial discharge model;

⑤ The discharge quantity q, voltage u, and discharge time t of each partial discharge model obtained in the above steps are made into spectrograms, and then the fingerprints of these spectrograms are calculated, and the partial discharge model of multiple experiments The discharge fingerprint constitutes the fingerprint library;

4) The measurement results of the partial discharge model are applied to the unknown type of discharge

For an unknown type of discharge, the fingerprint value obtained after collection, spectrum calculation, and fingerprint calculation is compared with the fingerprint of each discharge in the fingerprint database to identify the discharge type.

The method for measuring the partial discharge pattern recognition of the transformer is characterized in that: the transformer oil is pure ^# 25 transformer oil after dehydration and degassing treatment.

The method for measuring the partial discharge pattern recognition of the transformer is characterized in that: the material of the plate electrode, the column electrode and the pointed electrode is brass; the size of the plate electrode is φ100×15mm; the size of the column electrode is φ20×25mm; The electrode size is tip diameter 5mm, tip curvature radius 0.04mm, cone angle 30° and tip length 15mm.

The method for measuring the partial discharge pattern recognition of the transformer is characterized in that: the rated voltage U _N of the partial discharge-free test transformer is 100kV, the rated power S _N is 10kVA, and the discharge capacity at 100kV is less than 3pC.

The method for measuring the partial discharge pattern recognition of a transformer is characterized in that: the low-pass high-resistance impedance is a water resistance with a resistance value of 200-300 kΩ.

The method for measuring the partial discharge pattern recognition of a transformer is characterized in that: the capacitance of the coupling capacitor is 100pF, the withstand power frequency test voltage is 100kV, and the discharge capacity at 100kV is less than 2pC.

Principle of the present invention is:

The invention uses the measuring device to measure the discharge conditions of different partial discharge models, collects and analyzes the measurement results to make an atlas, and calculates the fingerprints to form a fingerprint library. The unknown type of discharge in the transformer only needs to be collected by the measuring device After the analysis, the atlas is made, the fingerprint is calculated and compared with the fingerprint information in the fingerprint database, the partial discharge type of the transformer can be identified.

The beneficial effects of the present invention are:

The invention establishes a transformer partial discharge fingerprint database, uses the information provided by various discharge spectrograms, can distinguish different types of discharges, provides valuable data for the application of fingerprint diagnosis and pattern recognition technology, and has good engineering practical value , with a confidence level of 99.5%.

Description of drawings

Figure 1 shows five typical partial discharge models (where (a) is an internal void type discharge model, (b) is a creeping surface type discharge model, (c) is a floating potential body type discharge model, and (d) is a bubble type discharge in oil model, (e) oil-barrier discharge model).

Fig. 2 is the identification result of a certain type of partial discharge signal using the measurement method of the present invention.

Figure 3 is the experimental wiring diagram of the transformer partial discharge measurement device.

Figure 4 is a general block diagram of the present invention.

Detailed ways

As shown in Figures 1 to 4, the measurement method for transformer partial discharge pattern recognition includes the following steps:

1) Build a partial discharge model:

According to the test results in the laboratory, the discharge types in the transformer are classified into five categories: paper (cardboard) internal void discharge, paper (cardboard) surface discharge, floating potential body discharge, oil bubble discharge, oil-barrier type discharge;

① Construct the internal void discharge model of paper (cardboard)

Use plate-plate electrode 102 and test sample 1. Test sample 1 is two layers of 1.0mm thick oil-impregnated insulating cardboard 101 with a layer of 0.5-1.5mm thick insulating cardboard sandwiched between them. Test sample 1 is placed on the plate-plate electrode 102. The two plate electrodes are immersed in transformer oil; due to several layers of medium, there are a large number of air gaps between the medium and the electrodes. Under the external high voltage, internal discharge occurs in these air gaps, simulating the internal gap discharge of cardboard;

② Construct paper (cardboard) creeping discharge model

Use a plate-plate electrode 102 with a distance of 15mm and test sample 2. Test sample 2 is a pretreated 2.5mm thick insulating cardboard 101. Test sample 2 is placed between the two plate electrodes of the plate-plate electrode 102 and immersed in transformer oil. In , the creeping discharge of insulating cardboard is simulated;

③ Constructing a floating potential body discharge model

A column electrode 103, a plate electrode 102, and a test sample three are adopted. The test sample three is a metal conductor. The test sample three is placed between the plate electrode 102 as a suspension electrode and the column electrode 103 as a high-voltage electrode. The column electrode 103 and the plate electrode The distance between 102 is selected as 4mm; the whole model is immersed in transformer oil to simulate the floating potential discharge;

④Construct the bubble type discharge model in oil

Plate electrode 102, column electrode 103 and test sample 4 are used. Test sample 4 is a 1.0 mm dry insulating cardboard 101. Test sample 4 is sandwiched between plate electrode 102 and column electrode 103 and immersed in transformer oil; It is not immersed in oil, so there will be more bubbles on the surface of the cardboard and inside the cardboard. When the electrode is pressurized, it is conducive to the occurrence of bubble discharge in oil, simulating the discharge of bubbles in oil;

⑤ Constructing an oil-barrier discharge model

Use plate electrode 102, pointed electrode 104 and test sample 5. Test sample 5 is an oil-impregnated insulating cardboard 101 with a thickness of 1 mm. The whole model is immersed in transformer oil to simulate oil-barrier discharge;

2) Build the measuring device

The measuring device for transformer partial discharge pattern recognition includes an experimental power supply, which is connected to the auto-transformer T1, which is output by the auto-transformer T1 and then connected to the isolation transformer T2, which is output by the isolation transformer T2 and then connected to the non-PD The test transformer T3 and the non-PD test transformer T3 are connected in series with a low-pass high-resistance impedance Z and then connected in parallel with an electrostatic voltmeter EVM, and then connected in parallel with the coupling capacitance C _K and the detection impedance Z _m connected in series, and then respectively connected with step 1) The partial discharge model C _x in parallel; put the partial discharge model C _x in step 1) in parallel in a closed transformer box, the box is filled with transformer oil, and the box shell is grounded to form a shielding structure; the box wall A UHF sensor 301 is installed, and the UHF sensor 301 is respectively connected to a spectrum analyzer 302 and a signal conditioning unit 303 through a cable C1, and the spectrum analyzer 302 and the signal conditioning unit 303 are connected to an industrial control computer 304;

3) Set the parameters of the device and measure the discharge parameters of the transformer partial discharge model

① Collect the broadband frequency domain data corresponding to partial discharge model discharge, compare it with the background noise, select the optimal frequency as the center frequency, adjust the center frequency of the spectrum analyzer to the optimal frequency, and set the bandwidth to about 5MHz;

②The high-voltage power supply respectively supplies power to the partial discharge model in step 1) through the high-voltage bushing;

③Electromagnetic waves generated by the discharge of each partial discharge model are respectively received by the UHF sensor installed on the wall of the box, and sent to the spectrum analyzer and signal conditioning unit respectively through the 50Ω measurement cable, and are amplified and filtered by the frequency mixing After the modules are processed, the industrial control computer collects and records the discharge quantity q, voltage u, and discharge time t through the NI5112 acquisition card;

④ Repeat step ③ for more than five times for each partial discharge model;

⑤ The discharge quantity q, voltage u, and discharge time t of each partial discharge model obtained in the above steps are made into spectrograms, and then the fingerprints of these spectrograms are calculated, and the partial discharge model of multiple experiments The discharge fingerprint constitutes the fingerprint library;

4) The measurement results of the partial discharge model are applied to the unknown type of discharge

For an unknown type of discharge, the fingerprint value obtained after collection, spectrum calculation, and fingerprint calculation is compared with the fingerprint of each discharge in the fingerprint database to identify the discharge type.

Transformer oil is pure ^# 25 transformer oil after dehydration and degassing.

The material of plate electrode, column electrode and tip electrode is brass; the size of plate electrode is φ100×15mm; the size of column electrode is φ20×25mm; the size of tip electrode is tip diameter 5mm, tip curvature radius 0.04mm, cone angle 30° tip length 15mm.

The rated voltage U _N of the non-partial discharge test transformer is 100kV, the rated power S _N is 10kVA, and the discharge capacity at 100kV is less than 3pC.

The method for measuring the partial discharge pattern recognition of a transformer is characterized in that: the low-pass high-resistance impedance is a water resistance with a resistance value of 300kΩ.

The method for measuring the partial discharge pattern recognition of a transformer is characterized in that: the capacitance of the coupling capacitor is 100pF, the withstand power frequency test voltage is 100kV, and the discharge capacity at 100kV is less than 2pC.

The isolation transformer T2 can effectively suppress the high-order harmonics entering the grid and improve the quality of the power supply. In order to avoid corona discharge of high-voltage leads, smooth copper rods are used for the leads, and all connectors are specially treated. The low-pass high-resistance impedance Z uses water resistance, which acts as a current limiter to protect the experimental equipment when the partial discharge model C _x breaks down suddenly, and also helps to suppress the interference on the power supply side. The selection of the resistance value of the water resistance must be appropriate. If it is too large, it may affect the voltage on the model. If it is too small, it will not have a protective effect. In the experiment, it is 300kΩ. The role of the coupling capacitor C _K , on the one hand, is to isolate the power frequency high voltage, so that the voltage on the detection impedance Z _m is very low, so as to ensure that the measuring device can work safely; on the other hand, it couples the partial discharge signal of the test product to the detection impedance Z _m up.

The data flow of the entire pattern recognition is shown in Figure 4. First, the data obtained by the measurement method is followed up, and the partial discharge data in the time domain is subjected to software peak processing and simulated hardware peak hold. At the same time, the power frequency voltage cycle is divided into 500 phase windows, and in each phase window, the maximum data exceeding the threshold is held for peak value, and then the obtained results are stored in the format of discharge capacity, voltage, phase, and cycle.

平均放电量相位分布

放电次数相位分布系统提取

H(q)和H(p)共五种二维放电谱图、

三维放电谱图以及放电椭圆图。Then the discharge signals of multiple power frequency cycles are counted to obtain various distribution spectrograms of partial discharge. Then the various partial discharge parameters of multiple power frequency cycles obtained are statistically measured in units of phase windows. The system calculates and displays three important discharge spectra based on phase windowing: maximum discharge phase distribution

Average discharge phase distribution

Phase distribution of discharge times system extraction

There are five kinds of two-dimensional discharge spectra of H(q) and H(p),

Three-dimensional discharge spectrum and discharge ellipsogram.

Then various distributions are statistically analyzed, and quantitative parameters (statistical operators) are used to describe the shape characteristics of a certain distribution. The present invention refers to TE571, in addition to selecting 29 popular statistical operators in the world, it also adds phase median μ for the first time, including a total of 37 statistical operators, including skewness Sk, protruding degree Ku, and number of local peaks Pe, discharge asymmetry Q, phase asymmetry φ, cross-correlation factor cc, phase median μ, correction correlation factor mcc, etc. These characteristic fingerprints can be used to identify discharge types.

In addition, some statistical operators can be selected from the 37 statistical operators according to actual needs, and the characteristic operators that can better reflect the change of discharge characteristics can be studied, so that the subsequent pattern recognition algorithm can be simplified.

Considering that the original data and different discharge types will continue to accumulate in practical applications, the present invention adopts a relatively open discharge fingerprint library management method, and adopts a one-level fingerprint library storage format from large to small to form a complete fingerprint database. Library management and maintenance system. Generally, the structure of "device level-problem level-fingerprint level" is adopted, which increases the flexibility of the system. In practical applications, when users obtain new and valuable data at different levels, they can add data to the corresponding level, which greatly improves the scalability of the system.

Claims (6)

1. the measuring method of a partial discharge of transformer pattern-recognition is characterized in that, specifically may further comprise the steps:

1) make up the shelf depreciation model:

According to the test result in the laboratory, the electric discharge type in the transformer is reduced five types: the discharge of paper (cardboard) internal voids type, paper (cardboard) are along bubble type discharge in the discharge of face type, the discharge of floating potential build, the oil, oil-barrier type discharge;

1. make up paper (cardboard) internal voids type discharging model

Adopt plate-plate electrode and test product one, test product one is the insulating board of impregnated insulation paper plate therebetween one bed thickness 0.5～1.5mm of two bed thickness 1.0mm, and test product one places between two plate electrodes of plate-plate electrode, immerses in the transformer oil; Because all there is a large amount of air gaps in which floor medium between medium and electrode, under plus high-pressure, these air gap generation internal discharges, the internal voids discharge of simulation cardboard;

2. make up paper (cardboard) along face type discharging model

Adopt at a distance of 15mm plate-plate electrode and test product two, test product two is the thick insulating board of the pretreated 2.5mm of process, and test product two places between two plate electrodes of plate-plate electrode, immerses in the transformer oil creeping discharge of analog insulation cardboard;

3. make up floating potential build discharging model

Adopt post electrode, plate electrode and test product three, test product three is a metallic conductor, and test product three places as between the plate electrode of suspension electrode and the post electrode as high-field electrode, and the distance between post electrode and plate electrode is elected 4mm as; The model immerses in the transformer oil, the discharge of simulation floating potential;

4. make up bubble type discharging model in the oil

Adopt plate electrode, post electrode and test product four, test product four is the dry insulating board of 1.0mm, and test product four is clipped between plate electrode and the post electrode, immerses in the transformer oil; Because test product four is immersion oil not in advance, thus paperboard surface and cardboard are inner can be with more bubble, when electrode pressurize, be beneficial to and take place that bubble discharges in the oil, the bubble type discharges in the simulated oil;

5. make up oil-barrier type discharging model

Adopt plate electrode, sharp electrode and test product five, test product five is the oil immersion cardboard of 1mm thickness, oil immersion cardboard close adhesion plate electrode, and sharp electrode and oil immersion cardboard immerse The model in the transformer oil at a distance of 1mm, simulated oil-barrier type discharge;

2) make up measurement mechanism

The measurement mechanism of partial discharge of transformer pattern-recognition includes experimental power supply; Experimental power supply inserts automatic coupling voltage regulator; Insert isolating transformer by automatic coupling voltage regulator output back, insert partial discharge-free test transformer by isolating transformer output back, parallelly connected after the impedance of a low pass high resistant of partial discharge-free test transformer series connection with an electrostatic voltmeter; Again with mutual coupling capacitance of connecting, detect the impedance parallel connection, more respectively with step 1) in the shelf depreciation model parallelly connected; Shelf depreciation model parallel connection in the step 1) is put in the transformer-cabinet of a sealing, is full of transformer oil in the casing, cabinet shell ground connection forms shielding construction; The superfrequency sensor is installed on the box body wall, and the superfrequency sensor is connected with the signal condition unit with spectrum analyzer through cable respectively, and spectrum analyzer is connected industrial control computer with the signal condition unit;

3) discharge parameter of the parameter of setting device and measuring transformer shelf depreciation model

1. gather the corresponding wide band frequency domain data of shelf depreciation model discharge, compare with ground unrest, choose optimal frequency as centre frequency, the centre frequency of spectrum analyzer is adjusted to optimal frequency, bandwidth is set to about 5MHz;

2. high-voltage power supply is given the power supply of the shelf depreciation model in the step 1) respectively through bushing;

3. the electromagnetic wave of each shelf depreciation model discharge generation is respectively by after the superfrequency sensor reception that is installed on the box body wall; Send into spectrum analyzer and signal condition unit respectively through the measurement cable of 50 Ω; After resume module such as mixing amplification filtering wherein, by industrial control computer gather through the NI5112 capture card and the discharge capacity q of record discharge, voltage u, discharge time t;

4. each shelf depreciation model repeating step is 3. more than five times;

The discharge capacity q of each the shelf depreciation model discharge that 5. above-mentioned steps is drawn, voltage u, discharge time, t processed spectrogram; Calculate the fingerprint of being made up of Statistical Operator of these spectrograms then, repeatedly the fingerprint of the discharge of the shelf depreciation model of experiment has just constituted fingerprint base;

4) measurement result of shelf depreciation model is applied in the discharge of UNKNOWN TYPE

For the discharge of UNKNOWN TYPE, compare with the fingerprint of every kind of discharge in the fingerprint base through the fingerprint value of gathering, spectrogram calculates, fingerprint obtains after calculating, identify electric discharge type.

2. the measuring method of partial discharge of transformer pattern-recognition according to claim 1 is characterized in that: described transformer oil pure for after handling through dehydration, the degassing ^#25 transformer oil.

3. the measuring method of partial discharge of transformer pattern-recognition according to claim 1 is characterized in that: the material of described plate electrode, post electrode, sharp electrode is a brass; Plate electrode is of a size of φ 100 * 15mm; The post electrode size is φ 20 * 25mm; The point electrode size is point footpath 5mm, tip curvature radius 0.04mm, 30 ° of long 15mm of point of cone angle.

4. the measuring method of partial discharge of transformer pattern-recognition according to claim 1 is characterized in that: the rated voltage U of described partial discharge-free test transformer _N=100kV, rated power S _N=10kVA, discharge capacity is less than 3pC under the 100kV.

5. the measuring method of partial discharge of transformer pattern-recognition according to claim 1 is characterized in that: the impedance of described low pass high resistant is for selecting water resistance for use, and resistance is 200-300k Ω.

6. the measuring method of partial discharge of transformer pattern-recognition according to claim 1 is characterized in that: the electric capacity of described coupling capacitance is 100pF, and the power frequency experimental voltage that can bear is 100kV, and the discharge capacity under the 100kV is less than 2pC.

Images