A method for identifying abnormal vibration of transformer under short-circuit impact

By building a mechanical and electrical shock testing system, the inherent vibration characteristics of the transformer are extracted and combined with Hilbert transformation and coherence analysis, the abnormal vibration under the short-circuit shock of the transformer is identified, which solves the problem of insufficient anti-interference ability in the existing technology and improves the evaluation of short-circuit resistance. Accuracy and grid safety.

CN115453421BActive Publication Date: 2024-11-05ELECTRIC POWER SCI & RES INST OF STATE GRID TIANJIN ELECTRIC POWER CO +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210985392.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-08-17
Publication Date
2024-11-05
Estimated Expiration
2042-08-17

AI Technical Summary

Technical Problem

The prior art is difficult to accurately identify abnormal vibration signals under the short-circuit impact of transformers, and the anti-interference ability is insufficient and the abnormal signals cannot be effectively separated, which affects the analysis of non-stationary transient signals during the short-circuit process.

Method used

A mechanical shock and electrical shock test system is built, and the inherent vibration characteristics of the transformer are extracted through Hilbert transformation and vibration decomposition, and abnormal signals are identified in combination with coherence analysis, and normal vibration signals are filtered to detect abnormal changes during short-circuit shock.

Benefits of technology

It realizes sensitive identification of abnormal vibrations under short-circuit impact of transformers, eliminates hysteresis stretching and noise interference of silicon steel sheets, improves the accuracy of short-circuit resistance evaluation, and ensures safe operation of the power grid.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115453421B_ABST
    Figure CN115453421B_ABST
Patent Text Reader

Abstract

The present invention relates to a method for identifying abnormal vibration of a transformer under short-circuit impact. By conducting mechanical impact and electrical impact tests before the transformer is put into operation, the inherent vibration characteristics of the transformer are extracted from the acceleration in combination with Hilbert decomposition. By analyzing the short-circuit impact vibration signal, the filtering of the normal vibration signal and the comparison of the inherent vibration characteristics are realized, and the abnormal changes of vibration during short-circuit impact are detected more sensitively. The present invention effectively extracts the inherent vibration characteristics of the transformer, solves the problem of identifying abnormal vibration under short-circuit impact of the transformer, and compared with traditional testing and identification methods, can better eliminate the influence of factors such as hysteresis expansion and background white noise of silicon steel sheets, eliminate mutual interference between similar vibration modes, and is applied to scenarios such as online operation monitoring of transformers. It can timely detect the hidden dangers of the mechanical performance of the transformer, evaluate the state of the transformer after short circuit, help improve the short-circuit resistance level of the transformer, and better ensure the safe operation of the power grid.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of electrical equipment operation detection, and in particular to a method for identifying abnormal vibration of a transformer under short-circuit impact. Background Art

[0002] Transformers are important equipment in the process of electric energy transmission, and their service life is generally more than 20 years. If a three-phase short circuit or other fault occurs on the medium and low voltage side of the transformer during operation, a large instantaneous current and electric force will be generated in the transformer. The short-circuit impact will cause the transformer's operating state to continue to deteriorate, and ultimately endanger the safe operation of the transformer. In order to accurately assess the operating state of the transformer, the operation and maintenance unit needs to monitor and process various physical signals that characterize the short-circuit impact and its impact, including impact current, vibration signals during the impact process, and other measurement values. Vibration signals can directly reflect the impact of instantaneous electric forces on the transformer, and are not affected by various electromagnetic signals during the short-circuit impact process, and have received more and more attention.

[0003] A vibrating transformer can be regarded as a multi-degree-of-freedom mechanical system. The system's internal deformation, stress changes and other characteristics will be reflected in the system's vibration signal. The vibration signal during the short-circuit impact is mainly measured by multiple piezoelectric sensors arranged outside the transformer box, which convert the vibration signal into vibration quantity and output it to the acquisition end. By analyzing the abnormal changes in the vibration signal, hidden dangers such as overall looseness of the transformer and deformation of the winding can be discovered, and the transformer's ability to resist short circuits can be evaluated online.

[0004] When the transformer is operating normally, the vibration signal has obvious power frequency correlation, which mainly comes from the hysteresis expansion and contraction of silicon steel sheets, the relative displacement between silicon steel sheets, and the natural vibration of coils and other structures. In addition, it also includes noise signals related to the operating environment. When the transformer is subjected to a short-circuit shock, the short-circuit current can be regarded as the superposition of multiple frequency components, which can generate electromagnetic forces in different frequency bands, causing obvious changes in the vibration signal in the frequency band close to the natural vibration frequency of the transformer.

[0005] At present, there are several main methods for processing transformer vibration signals: First, Fourier transform the signal to extract the frequency-amplitude characteristics of the time domain signal, but the anti-interference ability is insufficient and the adaptability to non-stationary transient signals during short circuit is poor; second, the frequency components of short circuit are extracted based on vibration characteristics of wavelet transform, but the selection of wavelet basis function will significantly affect the signal analysis results; third, the abnormal signal is extracted by using signal decomposition methods such as variational mode decomposition, but abnormal signals with similar frequencies will produce modal aliasing, and the abnormal signal cannot be accurately separated. Therefore, it is necessary to eliminate the influence of the vibration signal generated by normal operation as much as possible according to the characteristics of non-stationary vibration signals under short circuit impact, accurately extract the inherent vibration signal and detect its changes, and better identify abnormal vibration signals. Summary of the invention

[0006] The purpose of the present invention is to overcome the shortcomings of the prior art and propose a method for identifying abnormal vibration of a transformer under short-circuit impact, which can realize the filtering of normal vibration signals and the comparison of inherent vibration characteristics, and more sensitively detect abnormal changes in vibration during short-circuit impact.

[0007] The present invention solves the technical problem by adopting the following technical solutions:

[0008] A method for identifying abnormal vibration of a transformer under short-circuit impact comprises the following steps:

[0009] Step 1: Build a mechanical shock test system and an electrical shock test system;

[0010] Step 2: extracting the inherent vibration characteristics of the transformer according to the constructed mechanical shock test system and electrical shock test system;

[0011] Step 3: Wait for the short-circuit impact, identify the abnormal signal, and predict the location of the abnormal vibration based on the inherent vibration characteristics.

[0012] Moreover, the mechanical impact test system in step 1 includes a rubber pendulum and a fixing device thereof, a voltage measurement circuit, a vibration measurement circuit, a transmission circuit and a computer, wherein the rubber pendulum is fixed to the side wall of the transformer by its fixing device, and is used to collide with the transformer; the input end of the voltage measurement circuit is connected to the A phase, B phase, C phase and O phase bushing leads on the high-voltage side of the transformer; the input end of the vibration measurement circuit is connected to the four side walls and the top cover surface of the transformer, the output end of the voltage measurement circuit and the output end of the vibration measurement circuit are respectively connected to the input end of the transmission circuit, and the output end of the transmission circuit is connected to the computer;

[0013] The voltage measurement circuit includes a signal amplification unit, a low-pass filter and an A / D conversion unit, wherein the signal amplification unit, the low-pass filter and the A / D conversion unit are connected in series and are used to convert a small voltage signal into a digital quantity; the vibration measurement circuit includes a vibration sensor, a vibration signal amplification unit and an A / D conversion unit, wherein the vibration sensor, the vibration signal amplification unit and the A / D conversion unit are connected in series, and the vibration sensor is respectively installed on the four side walls and the top cover surface of the transformer and is used to convert a small vibration signal into a digital quantity.

[0014] Moreover, the electrical impact test system in step 1 includes a current measurement loop, a vibration measurement loop and a transmission loop, wherein the input end of the current measurement loop is connected to the transformer bushing current transformer for collecting the transformer bushing current, the input end of the vibration measurement loop is connected to the four side walls and the top cover surface of the transformer, the output end of the current measurement loop and the output end of the vibration measurement loop are respectively connected to the input end of the transmission loop, and the output end of the transmission loop is connected to a computer;

[0015] The current measurement circuit includes a current collector; the vibration measurement circuit includes a vibration sensor, a vibration signal amplifying unit and an A / D conversion unit, the vibration sensor, the vibration signal amplifying unit and the A / D conversion unit are connected in series, and the vibration sensors are respectively installed on the four side walls and the top cover surface of the transformer to convert small vibration signals into digital quantities.

[0016] Moreover, extracting the transformer inherent vibration characteristics according to the mechanical shock test system in step 2 includes the following steps:

[0017] Step 2.1.1, disconnect the high, medium and low voltage leads of the transformer, place the rubber pendulum on the four side wall surfaces of the transformer in turn, and hit each surface 3 to 5 times. The transmission circuit intercepts the digital value of the voltage and vibration measurement value during the impact period and uploads it to the computer;

[0018] Step 2.1.2: The computer generates the time domain signal V at each impact iP (t), and the average acceleration of the accelerometer At the same time Mark as invalid data;

[0019] Step 2.1.3, through the time domain signal V iP (t) and its Hilbert transform W iP (t) construct the analytical signal z(t);

[0020] Step 2.1.4: Perform two-component Hilbert vibration decomposition on the analytical signal z(t) to obtain the slowly varying frequency f1 and the fast asymmetric oscillation frequency f2. Based on synchronous detection, the instantaneous amplitude and phase of the first component are obtained to extract the time domain signal V iP (t) maximum amplitude component V1(t);

[0021] Step 2.1.5: Use the slowly varying frequency f1 as the intermediate frequency to construct a bandpass filter. Filter and calculate the sum of the kurtosis of all signals after filtering to obtain the maximum value K1 of the sum of kurtosis and the corresponding acceleration signals and its maximum value A 1m ;

[0022] Step 2.1.6: From the time domain signal V iP (t) minus the maximum amplitude component V1(t), and the subtracted signal is used as the new time domain signal V iP (t), determine whether K d <0.1K1, if K d <0.1K1, the iteration stops, otherwise the new time domain signal V iP (t) is returned to step 2.1.3 as a time domain signal, and the new voltage components Vd (t), slowly changing frequency f d , acceleration signal kurtosis K d , each acceleration signal and its maximum amplitude A dm (d=2,3,...).

[0023] Moreover, the specific implementation method of step 2.1.2 is as follows: when the i-th surface is impacted, i=1, 2, 3, 4 are the four side walls of the transformer, and the vibration average value of the n-th vibration sensor of the k-th surface is recorded as k is the four side walls and top cover of the transformer, n is the number of vibration sensors, and Mark as invalid data, record the voltage between phase A, phase B, phase C and phase O on the high-voltage side of the transformer each time it hits, calculate the average voltage of each phase when it hits the i-th surface, and record it as V iP (t)(P=A,B,C), and converted into a time domain signal V iP (t);

[0024] The specific implementation method of step 2.1.3 is:

[0025]

[0026] The specific implementation method of step 2.1.4 is: based on Hilbert vibration decomposition, the analytical signal z(t) is expressed as two component signals:

[0027]

[0028] Among them, f1 is the slowly varying frequency, f2 is the fast asymmetric oscillation frequency, θ1 is the vector angle of the first component signal, and θ2 is the vector angle of the second component signal. Based on synchronous detection, the instantaneous amplitude and phase of the first component are obtained to obtain the time domain signal V iP The component with the largest amplitude V1(t) in (t):

[0029] V1(t)=u1(t)cos(∫(2πf1t+θ1)dt)

[0030] Where f1 is the mean value of f(t) f1=∫f(t)dt;

[0031] The specific implementation method of step 2.1.5 is: construct M bandpass filters F with f1 as the center frequency m (m=1,...,M), the bandwidth value of the bandpass filter is M uniform discrete points in the interval of 0 to f1 / 2, and each bandpass filter F m Vibration Filter, calculate the kurtosis sum of all signals after filtering, obtain the maximum value K1 of the kurtosis sum and the corresponding vibration signals and its maximum value A 1m .

[0032] Moreover, extracting the inherent vibration characteristics of the transformer according to the electrical impact test system in step 2 includes the following steps:

[0033] Step 2.2.1, the electrical impact test is carried out simultaneously with the no-load closing test before the transformer is put into operation. The medium and low voltage sides of the transformer are no-loaded, and the high voltage side is subjected to 3 to 5 impact closings through the circuit breaker. The transmission circuit intercepts the digital value of the current and vibration measurement value during the impact period and uploads it to the computer;

[0034] Step 2.2.2, the computer records the vibration value and various current values;

[0035] Step 2.2.3, perform FFT transformation on each current, calculate the mean value of the signal amplitude when the frequency is an integer multiple of 50Hz, select the closing time with the smallest mean value, and record the corresponding acceleration value of each sensor as

[0036] Step 2.2.4: Construct a narrow stopband filter with 100N as the center frequency, N = 1, 2, ..., 20, for the acceleration value Filtering is performed to eliminate the influence of grid harmonics and excitation inrush current on the vibration signal. The acceleration after filtering is recorded as

[0037] Step 2.2.5: Using the time domain signal V iP (t) The same processing method is used to perform Hilbert vibration decomposition to obtain new acceleration components, slowly varying frequencies and acceleration signal kurtosis.

[0038] Step 2.2.6: Calculate the acceleration signal The maximum amplitude according to Use interpolation to estimate the magnitude A of the acceleration component when the frequency is f1 r .

[0039] Moreover, the specific implementation method of step 2.2.2 is: the computer records the acceleration value of the nth acceleration sensor on the kth surface during the i-th impact closing as k is the four side walls and top cover of the transformer, n is the number of vibration sensors, and the current value of each phase is I iP (P=A,B,C).

[0040] Moreover, in step 2, the natural vibration characteristics of the transformer are expressed by generating a normalized characteristic vector consisting of characteristic frequency and acceleration components through analysis of mechanical shock and electrical shock test signals:

[0041] C=(ξ i ,α i ) i=1,...,I

[0042] in, α i is the characteristic frequency ξ i is the normalized amplitude of the corresponding acceleration component, and I is the number of characteristic frequencies.

[0043] Furthermore, the step 3 comprises the following steps:

[0044] Step 3.1: After the transformer is running, the arrangement positions of the acceleration sensors on the transformer remain unchanged, waiting for the first short-circuit shock, and recording the acceleration values ​​of the acceleration sensors. right Using the time domain signal V iP (t) The same processing method is used to perform Hilbert vibration decomposition to obtain new acceleration components, slowly varying frequencies and acceleration signal kurtosis.

[0045] Step 3.2: According to The spectral distribution of each component and the eigenvector α i (t) Perform frequency coherence analysis. For i=1,...,I, where I is the number of characteristic frequencies, find Smallest

[0046] Step 3.3, let f, Δf represent the frequency and its resolution, S xy (f) indicates With α i The cross-spectral density of (t), S xx (f),S yy (f) indicates α i (t) and calculate the coherence function And the signal coherence index

[0047] Step 3.4: Starting from the second short-circuit shock, for each accelerometer, repeat steps 3.1 to 3.3 to calculate the signal coherence index Q′ i (i=1,…,I), and count the change in the mean value of the signal coherence index compared with the previous impact. If there is a 95% probability ΔQ i If the normal distribution is not satisfied, the Dixon test method is used to select abnormal data, and the corresponding abnormal signal frequency and sensor position are found; according to the vibration frequency, the abnormal signal frequency is judged to be related to a certain phase V of the transformer. iP (P=A,B,C) Slowly changing frequency f dThe closer the distance is, the greater the probability of abnormal vibration occurring in this phase; judging from the spatial distribution, when the sensor measuring the abnormal vibration has a concentrated distribution feature, the greater the probability of abnormal vibration occurring near the sensor.

[0048] The advantages and positive effects of the present invention are:

[0049] The present invention extracts the inherent vibration characteristics of the transformer from the acceleration by conducting mechanical shock and electrical shock tests before the transformer is put into operation, combining Hilbert decomposition, and realizes the filtering of normal vibration signals and the comparison of inherent vibration characteristics through the analysis method of short-circuit shock vibration signals, so as to more sensitively detect abnormal changes in vibration during short-circuit shock. The present invention effectively extracts the inherent vibration characteristics of the transformer, solves the problem of identifying abnormal vibrations under short-circuit shock of the transformer, and compared with traditional testing and identification methods, can better eliminate the influence of factors such as hysteresis and expansion of silicon steel sheets and background white noise, eliminate mutual interference between similar vibration modes, and is applied to scenarios such as online operation monitoring of transformers, which can timely detect the hidden dangers of mechanical performance of transformers, evaluate the status of transformers after short circuit, help improve the short-circuit resistance level of transformers, and better ensure the safe operation of power grids. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 This is a structural diagram of the mechanical shock testing system of the present invention;

[0051] Figure 2 This is a structural diagram of the electrical shock test system of the present invention;

[0052] Figure 3 This is the mechanical shock test signal processing flow of the present invention. DETAILED DESCRIPTION

[0053] The present invention is further described in detail below with reference to the accompanying drawings.

[0054] A method for identifying abnormal vibration of a transformer under short-circuit impact comprises the following steps:

[0055] Step 1: Build a mechanical shock test system and an electrical shock test system.

[0056] The inherent vibration characteristics of the transformer are extracted through mechanical shock test and electrical shock test. The mechanical shock test is carried out in the power-off state before the transformer is put into operation, and the electrical shock test is carried out when the transformer is impact-closed. Before the test, no less than one acceleration sensor is arranged around the transformer and on each side of the top cover, and the sensor range is 0-50g (where g is the acceleration of gravity).

[0057] like Figure 1As shown, the mechanical impact test system includes a rubber pendulum and its fixing device, a voltage measurement circuit, a vibration measurement circuit, a transmission circuit and a computer, wherein the rubber pendulum is fixed to the side wall of the transformer through its fixing device, and collides with the side of the transformer after being freely swung down from a certain debugging state; the input end of the voltage measurement circuit is connected to the A phase, B phase, C phase and O phase bushing leads of the transformer high voltage side; the input end of the vibration measurement circuit is connected to the four side walls and the top cover surface of the transformer, the output end of the voltage measurement circuit and the output end of the vibration measurement circuit are respectively connected to the input end of the transmission circuit, and the output end of the transmission circuit is connected to the computer. During the test, the leads on each side of the transformer high, medium and low voltage are disconnected, and the A phase, B phase, C phase and O phase bushing leads on the high voltage side are connected to the voltage measurement circuit. The side and top surfaces of the transformer are numbered from 1 to 5, and the pendulum is placed on the four sides of the transformer numbered 1, 2, 3, and 4 in sequence, and each side is intermittently impacted multiple times (preferably 3 to 5 times).

[0058] The voltage measurement circuit includes a signal amplification unit, a low-pass filter and an A / D conversion unit, wherein the signal amplification unit, the low-pass filter and the A / D conversion unit are connected in series and are used to convert a small voltage signal into a digital quantity; the vibration measurement circuit includes a vibration sensor, a vibration signal amplification unit and an A / D conversion unit, wherein the vibration sensor, the vibration signal amplification unit and the A / D conversion unit are connected in series, and the vibration sensor is respectively installed on the four side walls and the top cover surface of the transformer and is used to convert a small vibration signal into a digital quantity.

[0059] like Figure 2 As shown, the electrical impact test system includes a current measurement circuit, a vibration measurement circuit and a transmission circuit, wherein the input end of the current measurement circuit is connected to the transformer bushing current transformer for collecting the transformer bushing current, the input end of the vibration measurement circuit is connected to the four side walls and the top cover surface of the transformer, the output end of the current measurement circuit and the output end of the vibration measurement circuit are respectively connected to the input end of the transmission circuit, and the output end of the transmission circuit is connected to the computer. Before the test, the arrangement position of each acceleration sensor on the transformer remains unchanged, and the electrical impact test is carried out simultaneously with the no-load closing test before the transformer is put into operation. The medium and low voltage sides of the transformer are no-loaded, and the high voltage side is subjected to 3 to 5 impact closings through the circuit breaker.

[0060] The current measurement circuit includes a current collector; the vibration measurement circuit includes a vibration sensor, a vibration signal amplifying unit and an A / D conversion unit, the vibration sensor, the vibration signal amplifying unit and the A / D conversion unit are connected in series, and the vibration sensors are respectively installed on the four side walls and the top cover surface of the transformer to convert small vibration signals into digital quantities.

[0061] Step 2: Extract the inherent vibration characteristics of the transformer based on the constructed mechanical shock test system and electrical shock test system.

[0062] like Figure 3 As shown, extracting the natural vibration characteristics of the transformer according to the mechanical shock test system includes the following steps:

[0063] Step 2.1.1. Disconnect the leads on the high, medium and low voltage sides of the transformer, place the rubber pendulum on the four side wall surfaces of the transformer in turn, and impact each surface 3 to 5 times. The transmission circuit intercepts the digital value of the voltage and vibration measurement value during the impact period and uploads it to the computer.

[0064] Step 2.1.2, when the i-th surface is impacted, the average acceleration of the n-th acceleration sensor on the k-th surface is recorded as Considering that the signal-to-noise ratio is too low when the acceleration sensor and the pendulum are on the same plane, Mark as invalid data. Record the voltage between phases A, B, C and phase O each time a collision occurs, and calculate the average voltage of each phase when the i-th surface is impacted, recorded as V iP (P=A, B, C); The computer generates the time domain signal V at each impact iP (t).

[0065] Step 2.1.3, through the time domain signal V iP (t) and its Hilbert transform W iP (t) Constructing analytical signal

[0066] Step 2.1.4: Perform two-component Hilbert vibration decomposition on the analytical signal z(t) The slowly varying frequency f1 and the fast asymmetric oscillation frequency f2 are obtained. The estimated value of f1 is the mean value of f(t). r =∫f(t)dt, based on synchronous detection, the instantaneous amplitude and phase of the first component are obtained, and the time domain signal V is extracted iP (t) The maximum amplitude component V1(t) = u1(t)cos(∫(2πf1t+θ1)dt).

[0067] Step 2.1.5: Construct a bandpass filter F with the slowly varying frequency f1 as the intermediate frequency m (m=1,...,M) (the bandwidth value is M uniform discrete points in the interval of 0 to f1 / 2), using each F m For acceleration Filter, calculate the kurtosis and kurtosis of all signals after filtering, and obtain the maximum value K1 of the kurtosis and the corresponding acceleration signal and its maximum value A 1m .

[0068] Step 2.1.6: From the time domain signal V iP (t) minus the maximum amplitude component V1(t), and the subtracted signal is used as the new time domain signal V iP(t), determine whether K d <0.1K1, if K d <0.1K1, the iteration stops, otherwise the new time domain signal V iP (t) is returned to step 2.1.3 as a time domain signal, and the new voltage components V d (t), slowly changing frequency f d , acceleration signal kurtosis K d , each acceleration signal and its maximum amplitude A dm (d=2,3,...).

[0069] Extracting the transformer's inherent vibration characteristics based on the electrical impulse test system includes the following steps:

[0070] Step 2.2.1, the electrical impact test is carried out simultaneously with the no-load closing test before the transformer is put into operation. The medium and low voltage sides of the transformer are no-loaded, and the high voltage side is subjected to 3 to 5 impact closings through the circuit breaker. The transmission circuit intercepts the digital value of the current and vibration measurement value during the impact period and uploads it to the computer;

[0071] Step 2.2.2, the computer records the vibration value and various current values. When the i-th impact is closed, the acceleration value of the n-th acceleration sensor on the k-th surface is recorded as The current value of each phase is I iP (P=A,B,C).

[0072] Step 2.2.3, perform FFT transformation on each current, calculate the mean value of the signal amplitude when the frequency is an integer multiple of 50Hz, select the closing time with the smallest mean value, and record the corresponding acceleration value of each sensor as

[0073] Step 2.2.4: Construct a narrow stopband filter (N=1, 2, ..., 20) with 100N as the center frequency, and Filtering is performed to eliminate the influence of grid harmonics and excitation inrush current on the vibration signal. The acceleration after filtering is recorded as

[0074] Step 2.2.5: Using the time domain signal V iP (t) The same processing method is used to perform Hilbert vibration decomposition to obtain new acceleration components, slowly varying frequencies and acceleration signal kurtosis.

[0075] Step 2.2.6: Calculate the acceleration signal The maximum amplitude according to Use interpolation to estimate the magnitude A of the acceleration component when the frequency is f1r .

[0076] The natural vibration characteristics of the transformer are expressed by analyzing the mechanical shock and electrical shock test signals to generate a normalized characteristic vector consisting of characteristic frequency and acceleration components:

[0077] C=(ξ i ,α i ) i=1,…,I

[0078] in, α i is the characteristic frequency ξ i is the normalized amplitude of the corresponding acceleration component, and I is the number of characteristic frequencies.

[0079] Step 3: Wait for the short-circuit impact, identify the abnormal signal, and predict the location of the abnormal vibration based on the inherent vibration characteristics.

[0080] Step 3.1: After the transformer is running, the arrangement positions of the acceleration sensors on the transformer remain unchanged, waiting for the first short-circuit shock, and recording the acceleration values ​​of the acceleration sensors. right Using the time domain signal V iP (t) The same processing method is used to perform Hilbert vibration decomposition to obtain new acceleration components, slowly varying frequencies and acceleration signal kurtosis.

[0081] Step 3.2: According to The spectral distribution of each component and the alpha in the eigenvector i (t) Perform frequency coherence analysis. For i=1,...,I, where I is the number of characteristic frequencies, find Smallest

[0082] Step 3.3, let f, Δf represent the frequency and its resolution, S xy (f) indicates With α i The cross-spectral density of (t), S xx (f),S yy (f) indicates α i (t) and calculate the coherence function And the signal coherence index

[0083] Step 3.4: Starting from the second short-circuit shock, for each accelerometer, repeat steps 3.1 to 3.3 to calculate the signal coherence index Q′ i(i=1,…,I), and count the changes in the mean of the signal coherence index during the previous impact, taking the significance level β as 5% (i.e., there is a 95% probability ΔQ i does not satisfy the normal distribution), use the Dixon test method to select abnormal data, and find the corresponding abnormal signal frequency and sensor position: according to the vibration frequency, the abnormal signal frequency is related to a certain phase V of the transformer. iP (P=A,B,C) Slowly changing frequency f d The closer the distance is, the greater the probability of abnormal vibration occurring in this phase; judging from the spatial distribution, when the sensor measuring the abnormal vibration has a concentrated distribution feature, the greater the probability of abnormal vibration occurring near the sensor.

[0084] It should be emphasized that the embodiments described in the present invention are illustrative rather than restrictive. Therefore, the present invention includes but is not limited to the embodiments described in the specific implementation manner. Any other implementation manners derived by those skilled in the art based on the technical solution of the present invention also fall within the scope of protection of the present invention.

Claims

1. A method for identifying abnormal vibration of a transformer under short-circuit impact, characterized in that: The following steps are involved: Step 1: Build a mechanical shock test system and an electrical shock test system; The mechanical impact test system includes a rubber pendulum and a fixing device thereof, a voltage measurement circuit, a vibration measurement circuit, a transmission circuit and a computer, wherein the rubber pendulum is fixed to the side wall of the transformer through its fixing device to produce an impact with the transformer; the input end of the voltage measurement circuit is connected to the A-phase, B-phase, C-phase and O-phase bushing leads on the high-voltage side of the transformer; the input end of the vibration measurement circuit is connected to the four side walls and the top cover surface of the transformer, the output end of the voltage measurement circuit and the output end of the vibration measurement circuit are respectively connected to the input end of the transmission circuit, and the output end of the transmission circuit is connected to the computer; The voltage measurement circuit includes a signal amplification unit, a low-pass filter and an A / D conversion unit, wherein the signal amplification unit, the low-pass filter and the A / D conversion unit are connected in series and are used to convert a small voltage signal into a digital quantity; the vibration measurement circuit includes a vibration sensor, a vibration signal amplification unit and an A / D conversion unit, wherein the vibration sensor, the vibration signal amplification unit and the A / D conversion unit are connected in series, and the vibration sensor is respectively installed on the four side walls and the top cover surface of the transformer and is used to convert a small vibration signal into a digital quantity; Step 2: extracting the inherent vibration characteristics of the transformer according to the constructed mechanical shock test system and electrical shock test system; Extracting the transformer's inherent vibration characteristics based on the mechanical shock test system includes the following steps: Step 2.1.1, disconnect the high, medium and low voltage leads of the transformer, place the rubber pendulum on the four side wall surfaces of the transformer in turn, and hit each surface 3 to 5 times. The transmission circuit intercepts the digital value of the voltage and vibration measurement value during the impact period and uploads it to the computer; Step 2.1.2: The computer generates the time domain signal V at each impact iP (t), and the average acceleration of the accelerometer At the same time Mark as invalid data; Step 2.1.3, through the time domain signal V iP (t) and its Hilbert transform W iP (t) construct the analytical signal z(t); Step 2.1.4: Perform two-component Hilbert vibration decomposition on the analytical signal z(t) to obtain the slowly varying frequency f1 and the fast asymmetric oscillation frequency f2. Based on synchronous detection, the instantaneous amplitude and phase of the first component are obtained to extract the time domain signal V iP (t) maximum amplitude component V1(t); Step 2.1.5: Use the slowly varying frequency f1 as the intermediate frequency to construct a bandpass filter. Filter and calculate the sum of the kurtosis of all signals after filtering to obtain the maximum value K1 of the sum of kurtosis and the corresponding acceleration signals and its maximum value A 1m ; Step 2.1.6: From the time domain signal V iP (t) minus the maximum amplitude component V1(t), and the subtracted signal is used as the new time domain signal V iP (t), determine whether K d <0.1K1, if K d <0.1K1, then the iteration stops, otherwise the new time domain signal V iP (t) is returned to step 2.1.3 as a time domain signal, and the new voltage components V d (t), slowly changing frequency f d , acceleration signal kurtosis K d , each acceleration signal and its maximum amplitude Extracting the transformer's inherent vibration characteristics based on the electrical impulse test system includes the following steps: Step 2.2.1, the electrical impact test is carried out simultaneously with the no-load closing test before the transformer is put into operation. The medium and low voltage sides of the transformer are no-loaded, and the high voltage side is subjected to 3 to 5 impact closings through the circuit breaker. The transmission circuit intercepts the digital value of the current and vibration measurement value during the impact period and uploads it to the computer; Step 2.2.2, the computer records the vibration value and various current values; Step 2.2.3, perform FFT transformation on each current, calculate the mean value of the signal amplitude when the frequency is an integer multiple of 50Hz, select the closing time with the smallest mean value, and record the corresponding acceleration value of each sensor as Step 2.2.4: Construct a narrow stopband filter with 100N as the center frequency, N = 1, 2, ..., 20, for the acceleration value Filtering is performed to eliminate the influence of grid harmonics and excitation inrush current on the vibration signal. The acceleration after filtering is recorded as Step 2.2.5: Using the time domain signal V iP (t) The same processing method is used to perform Hilbert vibration decomposition to obtain new acceleration components, slowly varying frequencies and acceleration signal kurtosis. Step 2.2.6: Calculate the acceleration signal The maximum amplitude according to Use interpolation to estimate the magnitude A of the acceleration component when the frequency is f1 r ; Step 3: Wait for the short-circuit impact, identify the abnormal signal, and predict the location of the abnormal vibration based on the inherent vibration characteristics; Step 3.1: After the transformer is running, the arrangement positions of the acceleration sensors on the transformer remain unchanged, waiting for the first short-circuit shock, and recording the acceleration values ​​of the acceleration sensors. right Using the time domain signal V iP (t) The same processing method is used to perform Hilbert vibration decomposition to obtain new acceleration components, slowly varying frequencies and acceleration signal kurtosis. Step 3.2: According to The spectral distribution of each component and the eigenvector α i (t) Perform frequency coherence analysis. For i=1,...,I, where I is the number of characteristic frequencies, find Smallest Step 3.3, let f, Δf represent the frequency and its resolution, S xy (f) indicates With α i The cross-spectral density of (t), S xx (f),S yy (f) indicates α i (t) and calculate the coherence function And the signal coherence index Step 3.4: Starting from the second short-circuit shock, for each accelerometer, repeat steps 3.1 to 3.3 to calculate the signal coherence index Q i ′, i = 1, ..., I, and the change in the mean value of the signal coherence index compared with the previous impact is counted. If there is a 95% probability ΔQ i If the normal distribution is not satisfied, the Dixon test method is used to select abnormal data, and the corresponding abnormal signal frequency and sensor position are found; according to the vibration frequency, the abnormal signal frequency is judged to be related to a certain phase V of the transformer. iP , P = A, B, C slow-changing frequency f d The closer the distance is, the greater the probability of abnormal vibration occurring in this phase; judging from the spatial distribution, when the sensor measuring the abnormal vibration has a concentrated distribution feature, the greater the probability of abnormal vibration occurring near the sensor.

2. The method for identifying abnormal vibration of a transformer under short-circuit impact according to claim 1, characterized in that: In the step 1, the electrical impact test system comprises a current measurement loop, a vibration measurement loop and a transmission loop, wherein the input end of the current measurement loop is connected to the transformer bushing current transformer for collecting the transformer bushing current, the input end of the vibration measurement loop is connected to the four side walls and the top cover surface of the transformer, the output end of the current measurement loop and the output end of the vibration measurement loop are respectively connected to the input end of the transmission loop, and the output end of the transmission loop is connected to a computer; The current measurement circuit includes a current collector; the vibration measurement circuit includes a vibration sensor, a vibration signal amplifying unit and an A / D conversion unit, the vibration sensor, the vibration signal amplifying unit and the A / D conversion unit are connected in series, and the vibration sensors are respectively installed on the four side walls and the top cover surface of the transformer to convert small vibration signals into digital quantities.

3. The method for identifying abnormal vibration of a transformer under short-circuit impact according to claim 1, characterized in that: The specific implementation method of step 2.1.2 is: when the i-th surface is impacted, i=1, 2, 3, 4 are the four side walls of the transformer, and the vibration average value of the n-th vibration sensor of the k-th surface is recorded as k is the four side walls and top cover of the transformer, n is the number of vibration sensors, and Mark as invalid data, record the voltage between phase A, phase B, phase C and phase O on the high-voltage side of the transformer each time it hits, calculate the average voltage of each phase when it hits the i-th surface, and record it as V iP (t), P = A, B, C and converted into a time domain signal V iP (t); The specific implementation method of step 2.1.3 is: The specific implementation method of step 2.1.4 is: based on Hilbert vibration decomposition, the analytical signal z(t) is expressed as two component signals: Among them, f1 is the slowly varying frequency, f2 is the fast asymmetric oscillation frequency, θ1 is the vector angle of the first component signal, and θ2 is the vector angle of the second component signal. Based on synchronous detection, the instantaneous amplitude and phase of the first component are obtained to obtain the time domain signal V iP The component with the largest amplitude V1(t) in (t): V1(t)=u1(t)cos(∫(2πf1t+θ1)dt) Where f1 is the mean value of f(t) f1=∫f(t)dt; The specific implementation method of step 2.1.5 is: construct M bandpass filters F with f1 as the center frequency m , m=1,...,M, the bandwidth value of the bandpass filter is M uniform discrete points in the interval of 0 to f1 / 2, using each bandpass filter F m Vibration Filter, calculate the kurtosis sum of all signals after filtering, obtain the maximum value K1 of the kurtosis sum and the corresponding vibration signals and its maximum value A 1m .

4. The method for identifying abnormal vibration of a transformer under short-circuit impact according to claim 1, characterized in that: The specific implementation method of step 2.2.2 is: the computer records the acceleration value of the nth acceleration sensor on the kth surface during the i-th impact closing as k is the four side walls and top cover of the transformer, n is the number of vibration sensors, and the current value of each phase is I iP , P=A,B,C.

5. The method for identifying abnormal vibration of a transformer under short-circuit impact according to claim 1, characterized in that: In step 2, the natural vibration characteristics of the transformer are expressed by generating a normalized characteristic vector consisting of characteristic frequency and acceleration components through analysis of mechanical shock and electrical shock test signals: C=(ξ i ,a i ) i=1,…,I in, α i is the characteristic frequency ξ i is the normalized amplitude of the corresponding acceleration component, and I is the number of characteristic frequencies.