CN112946531A - Transformer fault diagnosis device and diagnosis method - Google Patents
Transformer fault diagnosis device and diagnosis method Download PDFInfo
- Publication number
- CN112946531A CN112946531A CN202110149505.4A CN202110149505A CN112946531A CN 112946531 A CN112946531 A CN 112946531A CN 202110149505 A CN202110149505 A CN 202110149505A CN 112946531 A CN112946531 A CN 112946531A
- Authority
- CN
- China
- Prior art keywords
- voiceprint
- transformer
- piezoelectric transducer
- connecting rod
- shaped connecting
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000003745 diagnosis Methods 0.000 title claims abstract description 25
- 238000000034 method Methods 0.000 title claims description 13
- 238000004891 communication Methods 0.000 claims abstract description 8
- 239000013598 vector Substances 0.000 claims description 25
- 238000001228 spectrum Methods 0.000 claims description 24
- 238000003825 pressing Methods 0.000 claims description 11
- 238000000605 extraction Methods 0.000 claims description 10
- 230000007613 environmental effect Effects 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000013527 convolutional neural network Methods 0.000 claims description 4
- 239000002184 metal Substances 0.000 claims description 4
- 229910052751 metal Inorganic materials 0.000 claims description 4
- 238000012549 training Methods 0.000 claims description 3
- 238000013016 damping Methods 0.000 claims description 2
- 230000006835 compression Effects 0.000 claims 2
- 238000007906 compression Methods 0.000 claims 2
- 238000012544 monitoring process Methods 0.000 abstract description 2
- 239000002699 waste material Substances 0.000 abstract description 2
- 238000007689 inspection Methods 0.000 description 8
- 238000012423 maintenance Methods 0.000 description 6
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical group [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 3
- 239000000919 ceramic Substances 0.000 description 3
- 238000004804 winding Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 2
- 230000026683 transduction Effects 0.000 description 2
- 238000010361 transduction Methods 0.000 description 2
- 240000007651 Rubus glaucus Species 0.000 description 1
- 235000011034 Rubus glaucus Nutrition 0.000 description 1
- 235000009122 Rubus idaeus Nutrition 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000003475 lamination Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/50—Testing of electric apparatus, lines, cables or components for short-circuits, continuity, leakage current or incorrect line connections
- G01R31/62—Testing of transformers
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
- G01H17/00—Measuring mechanical vibrations or ultrasonic, sonic or infrasonic waves, not provided for in the preceding groups
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Testing Electric Properties And Detecting Electric Faults (AREA)
Abstract
The invention discloses a transformer fault diagnosis device, which comprises: a controller; the wireless communication module is electrically connected with the controller; the connecting lines at two ends of the U-shaped connecting rod are perpendicular to two arms of the U-shaped connecting rod; the first piezoelectric transducer is fixedly connected to one end of the U-shaped connecting rod, the front face of the first piezoelectric transducer faces the other end of the U-shaped connecting rod, the front face of the first piezoelectric transducer is perpendicular to a connecting line of the two ends of the U-shaped connecting rod, and the first piezoelectric transducer is electrically connected with the controller; the screw hole is formed in one end, opposite to the first piezoelectric transducer, of the U-shaped connecting rod, and the central axis of the screw hole is parallel to the connecting line of the two ends of the U-shaped connecting rod; the screw rod, the screw rod and screw phase-match, screw rod threaded connection is in the screw hole. The problem of prior art to transformer fault monitoring waste manpower and diagnosis untimely is solved.
Description
Technical Field
The invention relates to the field of transformer maintenance, in particular to a transformer fault diagnosis device and a diagnosis method.
Background
The transformer is one of the most commonly used devices in the power industry, and is related to the safety of a power system, so operation and maintenance personnel need to detect the working state of the transformer at any time. The iron core and the winding are main parts of the transformer, the iron core lamination vibrates due to magnetostriction and magnetic leakage, when the winding deforms or the iron core becomes loose, the transformer can vibrate abnormally, if the winding is not deformed in time, the transformer is damaged completely, power failure of the transformer related area is caused, and great economic loss is caused. The prior art adopts the manual inspection mode for monitoring the transformer faults, but the problems of the manual inspection mode are as follows:
1) manpower is wasted, and because the transformers are widely distributed, if operation and maintenance personnel are required to observe each transformer on site, great manpower waste is caused;
2) the diagnosis is not timely, and because manual inspection is needed, each transformer cannot be dispatched to a real-time observation by manual inspection, but the manual inspection is realized by a regular inspection mode, so that the diagnosis result obtained by the inspection mode is not timely, the problem is not found in advance, and the transformer is damaged.
Disclosure of Invention
In order to solve the defects and shortcomings of the prior art, the invention aims to provide a transformer fault diagnosis device and a diagnosis method.
The technical scheme of the invention is as follows: a transformer fault diagnosis apparatus comprising:
a controller;
the wireless communication module is electrically connected with the controller;
the connecting lines at two ends of the U-shaped connecting rod are perpendicular to two arms of the U-shaped connecting rod;
the first piezoelectric transducer is fixedly connected to one end of the U-shaped connecting rod, the front face of the first piezoelectric transducer faces the other end of the U-shaped connecting rod, the front face of the first piezoelectric transducer is perpendicular to a connecting line of the two ends of the U-shaped connecting rod, and the first piezoelectric transducer is electrically connected with the controller;
the screw hole is formed in one end, opposite to the first piezoelectric transducer, of the U-shaped connecting rod, and the central axis of the screw hole is parallel to the connecting line of the two ends of the U-shaped connecting rod;
the screw rod, the screw rod and screw phase-match, screw rod threaded connection is in the screw hole.
Further, still include:
the pressing plate is fixedly connected to the end part, close to the first piezoelectric transducer, of the screw rod, and the pressing plate is perpendicular to the screw rod;
and a rubber layer is arranged on the surface of the pressing plate close to the first piezoelectric transducer.
Further, still include:
the nut, nut fixed connection is in the one end that first piezoelectric transducer was kept away from to the screw rod.
Further, still include:
a second piezoelectric transducer electrically connected with the controller.
Further, the second piezoelectric transducer is connected to the U-shaped connecting rod through a buffer device, the buffer device including:
the inner cavity of the sleeve is polygonal, the sleeve is made of metal, and the lower end of the sleeve is fixedly connected to the U-shaped connecting rod;
the pressure spring is fixedly connected to the bottom of the inner cavity of the sleeve;
the movable rod is a permanent magnet and is matched with the inner cavity of the sleeve, the lower end of the movable rod is fixedly connected to the upper end of the pressure spring, and the movable rod is movably connected into the inner cavity of the sleeve.
A transformer fault diagnosis method comprises the following steps:
step one, transformer voiceprint data are collected and stored in a storage medium, the transformer voiceprint data including good transformer voiceprint data are recorded as S1, namely bad transformer voiceprint data are recorded as P1, and collected background noise is recorded as B1;
step two, extracting the transformer voiceprint data S1, P1 and B1 at time intervals of 10 to 30 milliseconds to obtain voiceprint sequences S2, P2 and B2 in units of frames;
removing the environmental noise B2 of the transformer voiceprint sequences S2 and P2 through a noise reduction algorithm, and only reserving transformer voiceprint data to obtain voiceprint sequences S3 and P3;
step four, extracting the voiceprint characteristics of the voiceprint sequences S3 and P3 through a characteristic extraction algorithm to obtain characteristic vectors S4 and P4;
step five, performing model training on the deep convolutional neural network by adopting the feature vectors S4 and P4 with the number larger than 1 to obtain vector models S5 and P5;
step six, storing the vector models S5 and P5 into a voiceprint database;
collecting voiceprint data of the real-time transformer to be diagnosed as T1, and collecting background noise of the real-time transformer to be diagnosed as D1;
step eight, extracting T1 and D1 at a time interval of 10 to 30 milliseconds to obtain voiceprint sequences T2 and D2 in units of frames;
step nine, removing D2 from the voiceprint sequence T2 through a noise reduction algorithm, and only reserving transformer voiceprint data to obtain a voiceprint sequence T3;
step ten, extracting the voiceprint characteristics of the voiceprint sequence T3 through a characteristic extraction algorithm to obtain T4;
step eleven, comparing the voiceprint characteristics T4 with the voiceprint characteristics S5 and P5 stored in the database to obtain the similarity R1 between S5 and T4 and the similarity R2 between P5 and T4;
step twelve, if R1 is larger than R2, the transformer is judged to be good, and if R1 is smaller than R2, the transformer is judged to be about to be bad.
Further, the noise reduction algorithm in the third step and the ninth step is as follows:
c) carrying out Fourier transform on the background noise voiceprint sequence, and converting the background noise voiceprint sequence from a time domain signal to a frequency domain signal, thereby measuring the frequency spectrum characteristic of the background noise;
d) and performing a reverse compensation operation on the transformer voiceprint sequence according to the frequency spectrum characteristics of the noise to obtain the transformer voiceprint sequence after the noise is reduced.
Further, the feature extraction algorithm in the fourth step and the tenth step is as follows:
g) obtaining a corresponding frequency spectrum by Fourier transform of the voiceprint sequence;
h) the spectrum above is processed by a Mel filter bank to obtain a Mel spectrum;
i) and taking logarithm of the Mel frequency spectrum, then performing DCT discrete cosine transform, and taking the coefficients from 2 nd to 13 th after DCT as the voiceprint characteristic vector.
Further, the similarity R1 between S5 and T4 in the step eleven is calculated by
The calculation method of the similarity R2 between P5 and T4 is as follows:
in the formula (x)21,x22…,x2n) As the feature vector constituting S5, (x)31,x32…,x3n) Is the feature vector that constitutes P5, (x)11,x12…,x1n) The feature vector that constitutes T4.
The invention has the beneficial effects that: compared with the prior art, the method has the advantages that,
1) according to the invention, the thin plate on the shell of the transformer or on the radiator is placed between the first piezoelectric transducer and the end part of the screw rod, the screw rod is rotated in the screw hole by rotating the screw rod, so that the surface of the first piezoelectric transducer is fixed and is tightly attached to the surface of the transformer, the first piezoelectric transducer can detect the voiceprint of the transformer, and the voiceprint of the transformer is sent to a remote operation and maintenance person through the wireless communication module, the transformer fault is predicted by utilizing the difference between the voiceprint characteristic of the transformer before the fault and the voiceprint characteristic of the normal transformer, the fault of the transformer is predicted in advance, the operation and maintenance person does not need to go to the site for inspection, the labor is greatly saved, and meanwhile, the first piezoelectric transducer detects the voiceprint of the transformer in real time and sends the voiceprint of the transformer to the operation and maintenance;
2) according to the invention, the contact surface between the screw and the transformer is larger through the pressing plate, so that the surface of the transformer is prevented from being damaged by the screw;
3) according to the invention, through the rubber layer, the friction force between the pressing plate and the transformer is increased, and the pressing plate is prevented from sliding on the surface of the transformer;
4) according to the invention, the nut is larger in diameter than the screw rod, so that the nut is easier to rotate;
5) according to the invention, the environmental noise is detected through the second piezoelectric transducer, and the controller removes the environmental noise through the detection result of the second piezoelectric transducer, so that the fault of the transformer is more accurately judged;
6) when the vibration of the transformer is connected with the second piezoelectric transduction device through the buffer device, the movable rod of the permanent magnet is connected with the second piezoelectric transduction device, the movable rod and the sleeve move relatively, the movable sleeve and the movable rod are connected through the pressure spring, when the vibration is transmitted to the sleeve, the movable rod and the sleeve can move relatively, the sleeve is metal, the connecting rod is a permanent magnet, the permanent magnet moves in the sleeve to form annular current on the sleeve, the annular current can form a magnetic field for obstructing the movement of the connecting rod, the larger the speed of the relative movement of the sleeve and the connecting rod is, the larger the obstruction is, the buffer effect is achieved on the movement of the movable rod, the vibration of the transformer is weakened, the vibration of the transformer is prevented from being transmitted to the second piezoelectric sensor, and the background noise detected by the second piezoelectric sensor is prevented from being;
7) according to the method, the voiceprint characteristics of the transformer are detected, the similarity judgment is carried out on the currently detected voiceprint characteristics, the normal voiceprint characteristics of the transformer and the voiceprint characteristics of the fault transformer, so that whether the fault of the transformer is about to occur or not is automatically judged, and in addition, only the voiceprint data of the transformer are reserved through the measurement of background noise, so that the fault detection of the transformer is more accurate.
Drawings
FIG. 1 is a perspective view of the present invention;
FIG. 2 is a front view of the present invention;
FIG. 3 is a cross-sectional view taken along line A-A of FIG. 2;
FIG. 4 is a block diagram of the circuit connections of the present invention;
FIG. 5 is a flow chart of the present invention.
Detailed Description
The invention will be further described with reference to the following drawings and specific embodiments:
referring to fig. 1 to 4, a transformer fault diagnosis apparatus includes: a controller 10; the wireless communication module 11, the said wireless communication module 11 is electrically connected with controller 10; the connecting line of two ends of the U-shaped connecting rod 1 is vertical to two arms of the U-shaped connecting rod 1; the first piezoelectric transducer 2 is fixedly connected to one end of the U-shaped connecting rod 1, the front face of the first piezoelectric transducer 2 faces the other end of the U-shaped connecting rod 1, the front face of the first piezoelectric transducer 2 is perpendicular to a connecting line of two ends of the U-shaped connecting rod 1, and the first piezoelectric transducer 2 is electrically connected with the controller 10; the screw hole 3 is formed in one end, opposite to the first piezoelectric transducer 2, of the U-shaped connecting rod 1, and the central axis of the screw hole 3 is parallel to a connecting line of two ends of the U-shaped connecting rod 1; and the screw rod 4 is matched with the screw hole 3, and the screw rod 4 is in threaded connection with the screw hole 3. The controller 10 here may be a control component with peripheral circuitry such as Arduino, PLC or raspberry pi. The needless communication module 11 may be a 4G module or a 5G module.
Furthermore, an elastic pad 2-1 is arranged on the side edge of the front face of the first piezoelectric transducer 2, and the distance between the elastic pad 2-1 and the front face of the first piezoelectric transducer 2 is 1 mm; further comprising: the pressing plate 5 is fixedly connected to the end part, close to the first piezoelectric transducer 2, of the screw rod 4, and the pressing plate 5 is perpendicular to the screw rod 4; the surface of the pressure plate 5 close to the first piezoelectric transducer 2 is provided with a rubber layer 6.
Further, still include: and the nut 7 is fixedly connected at one end, far away from the first piezoelectric transducer 2, of the screw rod 4.
Further, still include: a second piezoelectric transducer 9, the second piezoelectric transducer 9 being electrically connected to a controller 10.
Further, the second piezoelectric transducer 9 is connected to the U-shaped connecting rod 1 through a damping device, which includes: the inner cavity of the sleeve 8-1 is polygonal, the sleeve 8-1 is made of metal, and the lower end of the sleeve 8-1 is fixedly connected to the U-shaped connecting rod 1; the pressure spring 8-2 is fixedly connected to the bottom of the inner cavity of the sleeve 8-1, and the pressure spring 8-2 is fixedly connected to the bottom of the inner cavity of the sleeve 8-1; the movable rod 8-3 is a permanent magnet, the movable rod 8-3 is matched with the inner cavity of the sleeve 8-1, the lower end of the movable rod 8-3 is fixedly connected to the upper end of the pressure spring 8-2, and the movable rod 8-3 is movably connected into the inner cavity of the sleeve 8-1.
Referring to fig. 5, the transformer fault diagnosis method of the present invention includes the steps of,
acquiring transformer voiceprint data through a first piezoelectric transducer 2, and storing the transformer voiceprint data in a storage medium, wherein the transformer voiceprint data comprises good transformer voiceprint data and is recorded as S1, namely recording bad transformer voiceprint data as P1, and acquiring background noise as B1; the sensor used by the acquired voiceprint data is a piezoelectric ceramic sound wave transducer, the piezoelectric ceramic sound wave transducer is attached to a transformer to be tested when the data is acquired, the vibration of the transformer is converted into an electric signal through the piezoelectric ceramic sound wave transducer and the electric signal is transmitted to a controller, and B1 background noise is picked up by arranging a second piezoelectric transducer 5 at a position far away from the transformer, because the second piezoelectric transducer is far away from the transformer and the environments are almost the same, the background noise measured by the second piezoelectric transducer is the background noise voiceprint which eliminates the vibration influence of the transformer;
step two, extracting the transformer voiceprint data S1, P1 and B1 at time intervals of 10 to 30 milliseconds to obtain voiceprint sequences S2, P2 and B2 in units of frames;
step three, removing the environmental noise B2 from the transformer voiceprint sequences S2 and P2 through a noise reduction algorithm, only reserving transformer voiceprint data to obtain voiceprint sequences S3 and P3, wherein the noise reduction algorithm for reducing the noise of S2 and P2 in the step is as follows:
a) carrying out Fourier transform on the voiceprint of the background noise B2, and converting the voiceprint of the background noise B2 from a time domain signal to a frequency domain signal, thereby measuring the frequency spectrum characteristic of the background noise;
b) and performing an inverse compensation operation on the S2 and the PS according to the frequency spectrum of the noise B2 to obtain S3 and P3.
Step four, extracting the voiceprint characteristics of the voiceprint sequences S3 and P3 through a characteristic extraction algorithm to obtain characteristic vectors S4 and P4, wherein the characteristic extraction algorithm comprises the following steps:
d) for S3 and P3, obtaining corresponding frequency spectrums through Fourier transform;
e) the spectrum above is processed by a Mel filter bank to obtain a Mel spectrum;
f) taking logarithm of Mel frequency spectrum, then making DCT discrete cosine transform, taking the 2 nd to 13 th coefficients after DCT as MFCC coefficients, obtaining Mel frequency cepstrum coefficient MFCC, which is the voice print feature vector of S3 or P3.
Step five, performing model training on the deep convolutional neural network by adopting the feature vectors S4 and P4 with the number larger than 1 to obtain vector models S5 and P5, wherein the deep convolutional neural network can be performed by adopting a machine learning system Tensorflow developed by Google company;
step six, storing the vector models S5 and P5 into a voiceprint database;
collecting voiceprint data of the real-time transformer to be diagnosed as T1, and collecting background noise of the real-time transformer to be diagnosed as D1;
step eight, extracting T1 and D1 at a time interval of 10 to 30 milliseconds to obtain voiceprint sequences T2 and D2 in units of frames;
step nine, removing D2 from the voiceprint sequence T2 through a noise reduction algorithm, only reserving transformer voiceprint data, and obtaining a voiceprint sequence T3, wherein the noise reduction algorithm for T2 noise reduction in the step is as follows:
a) carrying out Fourier transform on the voiceprint of the background noise D2, and converting the voiceprint from a time domain signal to a frequency domain signal, thereby measuring the frequency spectrum characteristic of the background noise;
b) and performing an inverse compensation operation on the T2 according to the frequency spectrum of the noise D2 to obtain T3.
Step ten, extracting the voiceprint features from the voiceprint sequence T3 through a feature extraction algorithm to obtain T4, wherein the feature extraction algorithm for the T3 in the step is as follows:
d) for T3, obtaining a corresponding frequency spectrum through Fourier transform;
e) the spectrum above is processed by a Mel filter bank to obtain a Mel spectrum;
f) taking logarithm of Mel frequency spectrum, then realizing by DCT discrete cosine transform, taking DCT 2 nd to 13 th coefficients as MFCC coefficients, and this MFCC is just the voice print feature vector of T3.
Step eleven, comparing the voiceprint characteristics T4 with the voiceprint characteristics S5 and P5 stored in the database to obtain the similarity R1 of S5 and T4, the similarity R2 of P5 and T4 and the voiceprint characteristics S5(x is the same as the similarity of the voiceprint characteristics S5 and P5 in the database21,x22…,x2n)、P5(x31,x32…,x3n) And T4 (x)11,x12…,x1n) The method is a quantity consisting of a plurality of characteristics, and can be marked by an n-dimensional vector, and the similarity of vectors can be judged by the distance between the vectors in Euclidean geometry, and the calculation formula is as follows:
step twelve, if R1 is larger than R2, the transformer is judged to be good, and if R1 is smaller than R2, the transformer is judged to be about to be bad; or adopting another rule, setting an empirical threshold value H, judging that the transformer is about to be deteriorated if R2> H, and judging that the transformer is good if R2< H.
The foregoing is a more detailed description of the invention in connection with specific preferred embodiments and it is not intended that the invention be limited to these specific details. For those skilled in the art to which the invention pertains, several simple deductions or substitutions can be made without departing from the spirit of the invention, and all shall be considered as belonging to the protection scope of the invention.
Claims (9)
1. A transformer fault diagnosis apparatus, characterized by comprising:
a controller (10);
the wireless communication module (11), the wireless communication module (11) is electrically connected with the controller (10);
the connecting line of the two ends of the U-shaped connecting rod (1) is vertical to the two arms of the U-shaped connecting rod (1);
the first piezoelectric transducer (2) is fixedly connected to one end of the U-shaped connecting rod (1), the front face of the first piezoelectric transducer (2) faces the other end of the U-shaped connecting rod (1), the front face of the first piezoelectric transducer (2) is perpendicular to a connecting line of the two ends of the U-shaped connecting rod (1), and the first piezoelectric transducer (2) is electrically connected with the controller (10);
the screw hole (3) is formed in one end, opposite to the first piezoelectric transducer (2), of the U-shaped connecting rod (1), and the central axis of the screw hole (3) is parallel to a connecting line of two ends of the U-shaped connecting rod (1);
the screw rod (4), screw rod (4) and screw (3) phase-match, screw rod (4) threaded connection is in screw (3).
2. The transformer fault diagnosis device according to claim 1, further comprising:
the pressing plate (5) is fixedly connected to the end part, close to the first piezoelectric transducer (2), of the screw rod (4), and the pressing plate (5) is perpendicular to the screw rod (4);
the surface of the pressing plate (5) close to the first piezoelectric transducer (2) is provided with a rubber layer (6).
3. The transformer fault diagnosis device according to claim 1, further comprising:
the nut (7), nut (7) fixed connection is in the one end that first piezoelectric transducer (2) was kept away from in screw rod (4).
4. The transformer fault diagnosis device according to claim 1, further comprising:
a second piezoelectric transducer (9), the second piezoelectric transducer (9) being electrically connected to the controller (10).
5. The transformer fault diagnosis device according to claim 1, characterized in that the second piezoelectric transducer (9) is connected to the U-shaped connection rod (1) by a damping device comprising:
the inner cavity of the sleeve (8-1) is polygonal, the sleeve (8-1) is made of metal, and the lower end of the sleeve (8-1) is fixedly connected to the U-shaped connecting rod (1);
the compression spring (8-2), the compression spring (8-2) is fixedly connected to the bottom of the inner cavity of the sleeve (8-1);
the movable rod (8-3), the movable rod (8-3) is the permanent magnet, the movable rod (8-3) and the inner cavity of the sleeve (8-1) are matched, the lower end of the movable rod (8-3) is fixedly connected to the upper end of the pressure spring (8-2), and the movable rod (8-3) is movably connected in the inner cavity of the sleeve (8-1).
6. A transformer fault diagnosis method is characterized by comprising the following steps:
step one, transformer voiceprint data are collected and stored in a storage medium, the transformer voiceprint data including good transformer voiceprint data are recorded as S1, namely bad transformer voiceprint data are recorded as P1, and collected background noise is recorded as B1;
step two, extracting the transformer voiceprint data S1, P1 and B1 at time intervals of 10 to 30 milliseconds to obtain voiceprint sequences S2, P2 and B2 in units of frames;
removing the environmental noise B2 of the transformer voiceprint sequences S2 and P2 through a noise reduction algorithm, and only reserving transformer voiceprint data to obtain voiceprint sequences S3 and P3;
step four, extracting the voiceprint characteristics of the voiceprint sequences S3 and P3 through a characteristic extraction algorithm to obtain characteristic vectors S4 and P4;
step five, performing model training on the deep convolutional neural network by adopting the feature vectors S4 and P4 with the number larger than 1 to obtain vector models S5 and P5;
step six, storing the vector models S5 and P5 into a voiceprint database;
collecting voiceprint data of the real-time transformer to be diagnosed as T1, and collecting background noise of the real-time transformer to be diagnosed as D1;
step eight, extracting T1 and D1 at a time interval of 10 to 30 milliseconds to obtain voiceprint sequences T2 and D2 in units of frames;
step nine, removing D2 from the voiceprint sequence T2 through a noise reduction algorithm, and only reserving transformer voiceprint data to obtain a voiceprint sequence T3;
step ten, extracting the voiceprint characteristics of the voiceprint sequence T3 through a characteristic extraction algorithm to obtain T4;
step eleven, comparing the voiceprint characteristics T4 with the voiceprint characteristics S5 and P5 stored in the database to obtain the similarity R1 between S5 and T4 and the similarity R2 between P5 and T4;
step twelve, if R1 is larger than R2, the transformer is judged to be good, and if R1 is smaller than R2, the transformer is judged to be about to be bad.
7. The transformer fault diagnosis method according to claim 8, wherein the noise reduction algorithm in the third and ninth steps is as follows:
a) carrying out Fourier transform on the background noise voiceprint sequence, and converting the background noise voiceprint sequence from a time domain signal to a frequency domain signal, thereby measuring the frequency spectrum characteristic of the background noise;
b) and performing a reverse compensation operation on the transformer voiceprint sequence according to the frequency spectrum characteristics of the noise to obtain the transformer voiceprint sequence after the noise is reduced.
8. The transformer fault diagnosis method according to claim 8, wherein the feature extraction algorithm in the fourth and tenth steps is:
d) obtaining a corresponding frequency spectrum by Fourier transform of the voiceprint sequence;
e) the spectrum above is processed by a Mel filter bank to obtain a Mel spectrum;
f) and taking logarithm of the Mel frequency spectrum, then performing DCT discrete cosine transform, and taking the coefficients from 2 nd to 13 th after DCT as the voiceprint characteristic vector.
9. The transformer fault diagnosis method according to claim 8,
the calculation method of the similarity R1 between S5 and T4 in the step eleven comprises the following steps
The calculation method of the similarity R2 between P5 and T4 is as follows:
in the formula (x)21,x22…,x2n) As the feature vector constituting S5, (x)31,x32…,x3n) Is the feature vector that constitutes P5, (x)11,x12…,x1n) The feature vector that constitutes T4.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110149505.4A CN112946531B (en) | 2021-02-03 | 2021-02-03 | Transformer fault diagnosis device and diagnosis method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110149505.4A CN112946531B (en) | 2021-02-03 | 2021-02-03 | Transformer fault diagnosis device and diagnosis method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112946531A true CN112946531A (en) | 2021-06-11 |
CN112946531B CN112946531B (en) | 2024-04-30 |
Family
ID=76242378
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110149505.4A Active CN112946531B (en) | 2021-02-03 | 2021-02-03 | Transformer fault diagnosis device and diagnosis method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112946531B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115588439A (en) * | 2022-12-13 | 2023-01-10 | 杭州兆华电子股份有限公司 | Fault detection method and device of voiceprint acquisition device based on deep learning |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
RU69643U1 (en) * | 2007-04-18 | 2007-12-27 | Федеральное государственное учреждение "13 Государственный научно-исследовательский институт Министерства обороны Российский Федерации" | INSTALLATION MEASURING ULTRASONIC AND MECHANO-ACOUSTIC UNIT FOR IT |
CN105158654A (en) * | 2015-08-24 | 2015-12-16 | 大连世有电力科技有限公司 | Intelligent robot for partial discharge diagnosis of transformer |
CN209014131U (en) * | 2018-09-19 | 2019-06-21 | 安徽大学 | Acoustic velocity tester with variable angle of receiving transducer |
CN110375846A (en) * | 2019-08-14 | 2019-10-25 | 杭州柯林电气股份有限公司 | Perceive the device of vocal print vibration inline diagnosis power transformer interior fault |
CN210609695U (en) * | 2019-10-31 | 2020-05-22 | 精拓丽音科技(北京)有限公司 | Electronic device |
CN112067956A (en) * | 2020-09-21 | 2020-12-11 | 国网河南省电力公司检修公司 | Ultrasonic probe fixing device |
CN214952102U (en) * | 2021-02-03 | 2021-11-30 | 中国电建集团贵州电力设计研究院有限公司 | Transformer fault detector |
-
2021
- 2021-02-03 CN CN202110149505.4A patent/CN112946531B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
RU69643U1 (en) * | 2007-04-18 | 2007-12-27 | Федеральное государственное учреждение "13 Государственный научно-исследовательский институт Министерства обороны Российский Федерации" | INSTALLATION MEASURING ULTRASONIC AND MECHANO-ACOUSTIC UNIT FOR IT |
CN105158654A (en) * | 2015-08-24 | 2015-12-16 | 大连世有电力科技有限公司 | Intelligent robot for partial discharge diagnosis of transformer |
CN209014131U (en) * | 2018-09-19 | 2019-06-21 | 安徽大学 | Acoustic velocity tester with variable angle of receiving transducer |
CN110375846A (en) * | 2019-08-14 | 2019-10-25 | 杭州柯林电气股份有限公司 | Perceive the device of vocal print vibration inline diagnosis power transformer interior fault |
CN210609695U (en) * | 2019-10-31 | 2020-05-22 | 精拓丽音科技(北京)有限公司 | Electronic device |
CN112067956A (en) * | 2020-09-21 | 2020-12-11 | 国网河南省电力公司检修公司 | Ultrasonic probe fixing device |
CN214952102U (en) * | 2021-02-03 | 2021-11-30 | 中国电建集团贵州电力设计研究院有限公司 | Transformer fault detector |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115588439A (en) * | 2022-12-13 | 2023-01-10 | 杭州兆华电子股份有限公司 | Fault detection method and device of voiceprint acquisition device based on deep learning |
Also Published As
Publication number | Publication date |
---|---|
CN112946531B (en) | 2024-04-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Rivas et al. | Condition assessment of power OLTC by vibration analysis using wavelet transform | |
CN110728257A (en) | Transformer winding fault monitoring method based on vibration gray level image | |
CN106338336A (en) | Transformer vibration on-line monitoring system | |
CN103472350A (en) | Transformer diagnosis system and diagnosis method | |
CN214952102U (en) | Transformer fault detector | |
CN112946531A (en) | Transformer fault diagnosis device and diagnosis method | |
CN113551765A (en) | Sound spectrum analysis and diagnosis method for equipment fault | |
CN116230013A (en) | Transformer fault voiceprint detection method based on x-vector | |
CN116432071A (en) | Rolling bearing residual life prediction method | |
CN115618205A (en) | Portable voiceprint fault detection system and method | |
CN113345399A (en) | Method for monitoring sound of machine equipment in strong noise environment | |
KR20210077496A (en) | Equipment failure prediction system having multi-channel sensors for receiving acoustic signals in the ultrasonic band | |
CN110161351A (en) | Transformer winding fault pilot system and diagnostic method under a kind of oscillation wave | |
CN110296679A (en) | A kind of device and monitoring method having the long-term deformation monitoring ability of valve rod | |
CN117292713A (en) | Transformer voiceprint monitoring method and system based on end Bian Yun cooperation | |
CN116884432A (en) | VMD-JS divergence-based power transformer fault voiceprint diagnosis method | |
CN112327218A (en) | Transformer online monitoring and fault diagnosis method | |
CN115828144A (en) | Signal sparse representation and fusion detection method, storage medium and electronic device | |
CN115453158A (en) | Voiceprint monitoring device and adjusting method for adaptively adjusting monitoring position | |
CN115291142A (en) | Transformer acoustic fingerprint uninterrupted power detection and state early warning method | |
CN115754630A (en) | Non-contact partial discharge monitoring method and device based on acoustic array | |
CN204495460U (en) | Transformer mechanical vibration wave spectrum on-line monitoring system | |
CN114966477A (en) | Transformer voiceprint analysis method and system | |
WO2017177190A1 (en) | Systems and methods for determination of health indicators using rank correlation analysis | |
CN110570052A (en) | predictive maintenance method for marine centrifugal pump |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |