Fault diagnosis and preliminary location system and method for transformer core looseness

Abstract

This invention publishes a fault diagnosis and location system for transformer core looseness, consists of vibration sensors, data collection and computer. It is of power transformer fault intelligent diagnosis technology technical field. Fault diagnosis and location method uses three vibration sensors positioned on the top of transformer tank, to obtain vibration signal, uses signal processing to analyze the signal, and obtains fault characteristics of the transformer core looseness. The characteristics respectively are 50 Hz, 150 Hz and 300 Hz frequency components, in which 300 Hz is main feature. When they reach a certain value at one position, it suggests that transformer core looseness near this position. By the comparison of differences among signals of three positions, fault location can be done. This invention manifests fault characteristics accurately and detects core looseness efficiently.

Description

TECHNICAL FIELD

The present invention relates to a technology of online condition monitoring of power transmission and transformation equipment, specificly relates to a fault diagnosis and location system for transformer core looseness. It is an intelligent substation technology.

BACKGROUND

The main reasons cause the transformer tank vibration are vibration from the transformer body and cooling system installations. The fundamental vibration frequency caused by the cooling device is low, and is significantly different to the transformer body vibration; body vibration includes core and winding vibration. When operating, the current in windings produces electromagnetic field in both core and winding; core silicon steel materials in the magnetic field have magnetostriction, namely the size of the atom has a small deformation, causing the core vibration. The solid line in FIG. 1 describes the relationship between the magnetostrictive deformation and core flux density. To simplify, use the quadratic curve instead, namely the dashed line. It can be seen that the magnetostrictive deformation is a linear relationship with the square of the size of the magnetic flux density. The relationship between the load voltage and magnetic flux density is:

$B=\frac{\Phi }{A}=\frac{\sqrt{2}U}{2\pi \mathrm{fNA}}$

Where Φ and B are main flux and magnetic flux density, respectively; A is cross-sectional area of the core; U is load voltage; f is the frequency of load voltage; N is the number of turns of primary winding.

From FIG. 1, core vibration v_core caused by the magnetostriction is approximately proportional to the square of load voltage:

v_core∝U²;

As the duplation frequency of voltage is 100 Hz, the fundamental frequency of magnetostriction is 100 Hz.

Core is made from silicon steel, each piece of silicon steel surface is insulating coated, so there is a gap between segments, resulting in magnetic flux leakage, cause core and tank vibration. But the vibration can be ignored because it is much smaller than the vibration caused by magnetostriction. So the core vibration depends on the magnetostriction of silicon steel, the core vibration fundamental frequency is 100 Hz. Because of the nonlinear magnetostrictive and different magnetic circuit length of core inside and outside the box, the core vibration, in addition to the fundamental frequency, also contains the harmonic components, which are integer multiple of the fundamental frequency component.

The vibration of the winding is caused by electric power. Under the influence of the leakage inductance, current in winding interacts to generate electric force, which is proportional to the square of the current. Winding current is zero on no-load condition, so winding vibration now has no influence on core vibration. The vibration of the no-load transformer depends on the core.

According to above analysis, no-load tank vibration is related to core magnetostriction, namely related to voltage. Transformer vibration transmits to tank via transformer oil and solid structures. Be influenced by various factors, vibration signal changes in amplitude and phase. When it reaches the tank surface, it becomes complex.

After the loosening of the core, the magnetic flux leakage between the silicon steel joints and laminations become larger, resulting in larger electromagnetic attractive force, larger core vibration.

The transformer vibration signal is non-stationary signals. Signal processing methods include Fourier transform, wavelet transform, Hilbert Huang Transform. The Fourier transform is the most classic signal processing methods. It is suitable for stationary signals, to transform time domain of the signals to frequency domain, is widely used in engineering. Wavelet de-noising principle is shown in FIG. 2, the high-frequency signal can be filtered by the signal decomposition and reconstruction, reorganize low-frequency signal to the filtering effect.

CONTENT OF INVENTION

One embodiment of the present invention is a fault diagnosis and preliminary location system and method for transformer core looseness, said system have at least three vibration sensors, a conversion interface, a data collection module and a data analysis module, wherein said vibration sensors collect transformer vibration signal with a set sampling frequency and time, then the vibration signal delivers to said data collection module via conversion interface. The three sensors are fixed on three positions of power transformer tank top surface corresponding with the three-phase winding. Said data collection is used to sample and record vibration signal from sensors, then deliver it to data analysis module. Said data analysis module stores and analyzes data and diagnose fault, finally outputs the result.

The present invention provides a data analysis module including wavelet de-noising unit, Fourier transform unit, data storage unit, calculating unit and output unit.

Wherein said wavelet de-noising unit de-noises the vibration signal from data collection.

Then the said Fourier transform unit do Fourier transform to de-noised signal, to get spectrum.

Said data storage unit stores TH₁, TH₂, CR₁ and CR₂, wherein TH₁ is the threshold at 300 Hz of the spectrum of the vibration signal, TH₂ is the threshold of 50 Hz plus 150 Hz of the spectrum of the vibration signal.

Said calculating unit compares the amplitude of 300 Hz of the spectrum with TH₁. Signal samples are collected at least 3 times at one same condition. When the samplings on the amplitude of 300 Hz are larger than TH1 at least 2 times, then calculate 50 Hz plus 150 Hz at the spectrum, compare it with TH₂; When the samplings on the amplitude of 50 Hz plus 150 Hz is larger than TH₂ at least 2 times, get the conclusion, there is core looseness near the position. Output the result to output unit.

The present also provides test method for said fault diagnosis and preliminary location system and method for transformer core looseness:

• Steps are shown as following on a no-load power transformer at normal condition:
  • (1) According to default sampling frequency and sampling time, use three vibration sensors said in claim 1, to collect transformer vibration, sample at least three times at one same condition;
  • (2) Among signals from all sensors, obtain signals by complete cycles according to sampling frequency, sampling time or sampling number;
  • (3) Use wavelet to de-noise signal of (2), do Fourier transform, then get the spectrum value at 50 Hz, 150 Hz and 300 Hz;
  • (4) Calculate the amplitude of 300 Hz at the spectrum, multiply by a certain magnification, as the value of TH₁; calculate the amplitude of 50 Hz plus 150 Hz, multiply by a certain magnification, as the value of TH₂; wherein said certain magnification generally is from 1.2 to 2.5.

Steps are shown as following, when the transformer is on a no-load stable operating condition, to diagnose the fault:

• • (5) Repeat step (1) to step (3), sample the vibration signal and process the data, to get the amplitude of 50 Hz, 150 Hz, 300 Hz at the spectrum;
  • (6) Take the amplitude of 300 Hz of the spectrum as CR₁. If at least two continuous CR₁ is larger than TH₁, continue to step (7), else, go back to step (5);
  • (7) Calculate 50 Hz plus 150 Hz at the spectrum, take it as CR₂. Compare it with TH₂. If at least 2 times of the samplings that CR₂ is larger than TH₂, get the conclusion, there is core looseness near the position. Output the result to output unit, else, back to step (5).

Compared with present technology, the present invention has these advantages:

Experiments verify that the selected features in the present invention can accurately reflect the fault characteristics of core looseness, effectively detect core looseness of power transformer. Compare signal characteristic of three different positions, orientate the fault location preliminary.

BRIEF DESCRIPTION OF DRAWINGS

FIG. 1 is iron core magnetostrictive changes with magnetic flux density of the invention;

FIG. 2 is schematic diagram of the wavelet de-noising of the invention;

FIG. 3 is the diagnosis system block diagram of the invention;

FIG. 4 is the sensors fixed position of the diagnosis system of the invention;

FIG. 5 is the flow chart of the diagnosis method of the invention;

FIG. 6( a) is the original vibration signal from the data collection of the embodiment of the invention;

FIG. 6( b) is the de-noised signal of the embodiment of the invention;

FIG. 7( a) is the vibration signal spectrum of fault point at normal condition of the embodiment of the invention;

FIG. 7( b) is the vibration signal spectrum of fault point at fault condition of the embodiment of the invention;

FIG. 8( a) is the vibration signal spectrum of non-fault point 1 at normal condition of the embodiment of the invention;

FIG. 8( b) is the vibration signal spectrum of non-fault point 1 at fault condition of the embodiment of the invention;

FIG. 9( a) is the vibration signal spectrum of non-fault point 2 at normal condition of the embodiment of the invention;

FIG. 9( b) is the vibration signal spectrum of non-fault point 2 at fault condition of the embodiment of the invention.

DESCRIPTION OF EMBODIMENTS

The following is further details of the embodiments of the present invention:

As FIG. 3, the present invention provides a fault diagnosis and location system for transformer core looseness, consists of transformer, vibration sensors, conversion interface, data collection and analysis module. Wherein, input sides of the sensors are firmly fixed at positions corresponding with three phases on the surface of transformer tank via magnet. As FIG. 4, A, B, C represent three phases of high voltage side, respectively; a, b, c represent three phases of low voltage side, respectively; 0 represents the zero line of transformer; 1, 2, 3 represent three measuring positions of sensors on the top of transformer tank. The output side of the sensors is linked to the input side of data collection via conversion interface. The output side of data collection is linked to analysis module via network line interface.

FIG. 5 is the flow chart of the fault diagnosis and location system for transformer core looseness method of the invention, including following steps:

• • (1) According to default sampling frequency and sampling time, collect transformer vibration of stable operating condition, sample at least three times at one same condition;
  • (2) Among signals from all sensors, obtain signals by complete cycles according to sampling frequency, sampling time or sampling number;
  • (3) Use wavelet to de-noise signal of (2), then do Fourier transform;
  • (4) Calculate the amplitude of 50 Hz, 150 Hz and 300 Hz at the spectrum;
  • (5) Take the amplitude of 300 Hz at the spectrum, multiply by a certain magnification, as the value of TH₁; take the amplitude of 50 Hz plus 150 Hz, multiply by a certain magnification, as the value of TH₂; wherein said certain magnification generally is from 1.2 to 2.5.

Collect vibration signals on an operating transformer of normal condition, calculate TH₁,TH₂, take them as threshold.

Use sensors to sample the transformer vibration signal, deal with the signal as said in step (3), get the amplitude of 50 Hz, 150 Hz, 300 Hz.

• • (6) Take the amplitude of 300 Hz of the spectrum as CR₁. Sample at least 3 times at one same condition. If at least 2 times of the samplings that CR₁ is larger than TH₁, continue to step (7), else, repeat step (1)˜(4), continue sampling, then go to step (6);
  • (7) Calculate 50 Hz plus 150 Hz at the spectrum, take it as CR₂. Compare CR₂ with TH₂. If at least 2 times of the samplings that CR₂ is larger than TH₂, get the conclusion, there is core looseness near the position. Output the result to output unit, else, repeat step (1)˜(4), continue sampling, then go to step (6);

EXAMPLES

Set core looseness fault on a test power transformer, verify the correctness of the present invention, and follow the steps above to do an experiment. The transformer is made by the Jiangsu Hongyuan Electrical Co., Ltd., its parameters are shown in Tab.1.

TABLE 1
Type	Voltage ratio	Connection Group
S9-M-100/10	10/0.4 kV	Yyn0
High-voltage side IN	Ligh-voltage side IN	Short-circuit impedance
5.77A	144.3A	3.98%

(A) System Connection

As FIG. 3, the type of transformer is S9-M-100/10, the type of vibration sensor is CA-YD-103, use BNC interface as conversion interface, the type of data collection is Nicolet7700, use a computer as the analysis module. The s input sides of the sensors are firmly fixed at the positions corresponding with three phases on the surface of transformer tank via magnet. The output sides of the sensors are linked to the input side of data collection via conversion interface. The output side of data collection is linked to analysis module via network line interface.

(B) The Installation of Sensors

Use CA-YD-103 sensor, its parameters are shown in Tab.2.

TABLE 2
	Maximum
	transverse
Axial sensitivity	sensitivity	Ballistic limit	Frequency response
20 pC/g	<5%	2000 g	0.5~12 kHz

In order to fully measure the vibration of the transformer core, the experiment is done as much as possible on no-load condition; three vibration sensors are installed at three positions on the top. Specific installation location is shown in FIG. 4.

(C) The Setting of Core Looseness

Sling the core with a crane. Use the wrench to loose core fastening screw for about 1 cm. Beat the loosen side with a mallet, and then put a bamboo gently to the gap of the core silicon steel to further loosen the core.

(D) Test Example

Nicolet data collection has charge amplifier inside. Nicolet is used to sample and record vibration signal collected by sensors, and the computer is used to store and process the signal, diagnose the fault and output the results.

In this example, test as the steps described above, obtain vibration signal and de-noise it with wavelet method. FIG. 6(a) and FIG. 6(b) is vibration signals before and after de-noising. The two drawings show the effect of de-noising.

FIG. 7( a) and FIG. 7(b) show the spectrum of position 1#, (namely near fault position). Compare spectrum of normal and fault condition, it can be seen that after looseness, there will produce more 50 Hz harmonic components, the energy of 300 Hz arises.

When the core is loosened from the tank surface vibrations will produce more 50 Hz harmonic components, 300 Hz energy rise.

After calculation, the change of frequency components amplitude is shown in Tab. 3.

TABLE 3
Amplitude of 1#
on no-load condition	normal/normal	looseness/normal
(50 + 150)	Hz	1	14.8
300	Hz	1	3.93

Use Fourier transform, take 300 Hz (CR₁) as the main characteristic. When the value of CR₁ reaches a certain point, it suggests transformer core looseness near this position. When the value of 50 Hz plus 150 Hz (CR₂) reaches a certain point, further determine there is core looseness near the position.

FIG. 8( a) and FIG. 8(b) show the spectrum of position 2#, (namely non-fault position), before and after fault setting, respectively. It can be seen, there is no fault characteristic as at 1#.

FIG. 9( a) and FIG. 9(b) show the spectrum of position 3#, (namely non-fault position), before and after fault setting, respectively. It can be seen, there is no fault characteristic as at 1#.

On the basis of theoretical analysis, with a large number of experiments, it is proved that the above characteristics are with good reproducibility and regularity, and verify that this feature can be used in the transformer core looseness fault diagnosis.

Finally, it should be noted, that the above embodiments are only used to describe the technical solution of the present invention rather than to limit this technique, the present invention can be extended its application to other modifications, changes, applications and embodiments, and therefore all of such modify, change, application, embodiments are included in the spirit and teachings of the present invention.

The invention uses the electricity grid frequency of 50 Hz for example, for 60 Hz power system, above 50 Hz, 150 Hz, 300 Hz characteristic frequencies are 60 Hz, 180 Hz and 360 Hz.

Claims

1. A fault diagnosis and preliminary location system and method for transformer core looseness, said system have at least three vibration sensors, a conversion interface, a data collection module and a data analysis module, wherein said vibration sensors collect transformer vibration signal with a set sampling frequency and time, then the vibration signal delivers to said data collection module via conversion interface. The three sensors are fixed on three positions of power transformer tank top surface corresponding with the three-phase winding. Said data collection is used to sample and record vibration signal from sensors, then deliver it to data analysis module. Said data analysis module stores and analyzes data and diagnose fault, finally outputs the result.

2. The method of claim 1, wherein said analysis module consists of wavelet de-noising processing unit, Fourier transform unit, data storage unit, calculation unit and output unit. Wherein said wavelet de-noising unit de-noises the vibration signal from data collection module. Then the said Fourier transform unit do Fourier transform to de-noised signal, to get spectrum.Said data storage unit stores TH1, TH2, CR1 and CR2, wherein TH1 is the threshold at 300 Hz of the spectrum of the vibration signal, TH2 is the threshold of 50 Hz plus 150 Hz of the spectrum of the vibration signal.Said calculating unit compares the amplitude of 300 Hz of the spectrum with TH1. Signal samples are collected at least 3 times at one same condition. When the samplings on the amplitude of 300 Hz are larger than TH1 at least 2 times, then calculate the amplitude of 50 Hz plus 150 Hz at the spectrum, compare it with TH2; When the amplitude of 50 Hz plus 150 Hz is larger than TH2 at least 2 times, get the conclusion, there is core looseness near the position. Output the result to output unit.

3. Said in claim 1, the present invention also provides test method for fault diagnosis and preliminary location system and method for transformer core looseness: Steps are shown as following on a no-load power transformer at normal condition: (1) According to default sampling frequency and sampling time, use three vibration sensors said in claim 1, to collect transformer vibration, sample at least three times at one same condition;(2) Among signals from all sensors, obtain signals by complete cycles according to sampling frequency, sampling time or sampling number;(3) Use wavelet to de-noise signal of (2), do signal processing, then get the spectrum value at 50 Hz, 150 Hz and 300 Hz;(4) Calculate the amplitude of 300 Hz at the spectrum, multiply by a certain magnification, as the value of TH1; calculate the amplitude of 50 Hz plus 150 Hz, multiply by a certain magnification, as the value of TH2; wherein said certain magnification generally is from 1.2 to 2.5.Steps are shown as following, when the transformer is on a no-load stable operating condition, to diagnose the fault: (5) Repeat step (1) to step (3), sample the vibration signal and process the data, to get the amplitude of 50 Hz, 150 Hz, 300 Hz at the spectrum;(6) Take the amplitude of 300 Hz of the spectrum as CR1. If at least two continuous CR1 is larger than TH1, continue to step (7), else, go back to step (5);(7) Calculate 50 Hz plus 150 Hz at the spectrum, take it as CR2. Compare CR2 with TH2. If at least 2 times of the samplings that CR2 is larger than TH2, get the conclusion, there is core looseness near the position. Output the result to output unit, else, back to step (5).The invention uses the electricity grid frequency of 50 Hz for example, for 60 Hz power system, above 50 Hz, 150 Hz, 300 Hz characteristic frequencies are 60 Hz, 180 Hz and 360 Hz.