On-load tap changer condition monitoring method based on electromagnetic radiation signals

Abstract

A method for monitoring the status of on-load tap changers based on electromagnetic radiation signals is proposed. The method involves collecting electromagnetic radiation signals during the switching of the on-load tap changer under normal operating conditions, filtering and preprocessing the electromagnetic radiation signals, and extracting peak information of the electromagnetic radiation signals using a dual-threshold peak-finding algorithm. This yields information on the switching sequence of the on-load tap changer, the discharge process between the tap changer contacts, and characteristic quantities during the switching process. Support vector machines are then used to classify these characteristic quantities, and the operating status of the on-load tap changer is determined based on the classification results.

Description

Technical Field

[0001] This invention belongs to the field of electrical equipment fault detection technology, and in particular, it is a method for monitoring the status of on-load tap changers based on electromagnetic radiation signals. Background Technology

[0002] Power systems must not only provide safe and stable electrical energy, but also regulate line voltage and distribute power rationally in a timely manner. Transformers, as a crucial component of the power system, play a vital role by adjusting output voltage in response to demand changes or load variations. The on-load tap changer (OLTC) is the component within a transformer responsible for voltage regulation. It alters the voltage ratio by changing the number of turns in the transformer windings, thereby regulating the line voltage. As the only mechanically active component inside the transformer besides the fan (which serves for ventilation and heat dissipation), the OLTC constantly bears the burden of breaking and closing large currents. Therefore, it is one of the components with the highest failure rate in transformers, accounting for approximately 23.2% of all transformer failures. Thus, condition monitoring of the OLTC is extremely important.

[0003] On-load tap changers regulate voltage by separating and closing the moving and stationary contacts in the switching core. Due to the inductance of the circuit, the current cannot change abruptly at the moment of contact separation, resulting in a follow current arc. The electromagnetic transient process of the arc not only generates high-frequency current in the circuit, causing current distortion, but also radiates electromagnetic signals. Collecting and analyzing voltage, current, and electromagnetic signals can reflect the discharge process of the on-load tap changer, thereby evaluating its electrical performance.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of the present invention, and therefore may contain information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention proposes an on-load tap changer status monitoring method based on electromagnetic radiation signals. This method provides high monitoring accuracy without affecting the normal operation of the tap changer, reduces maintenance costs, and avoids serious accidents.

[0006] The objective of this invention is achieved through the following technical solution: the on-load tap changer status monitoring method based on electromagnetic radiation signals includes the following steps:

[0007] Collect electromagnetic radiation signals during the switching of the on-load tap changer under normal operating conditions.

[0008] The electromagnetic radiation signal is filtered and preprocessed, and the peak information of the electromagnetic radiation signal is extracted using a dual-threshold peak-finding algorithm to obtain the switching timing of the on-load tap changer and the discharge process information between the tap changer contacts. The discharge process information includes characteristic quantities during the switching process.

[0009] The feature quantities are classified using a support vector machine, and the working state of the on-load tap changer is determined based on the classification results.

[0010] The method involves building a test platform for on-load tap changers. The platform measures the voltage between the contacts, the current flowing through them, the high-frequency current signal with a frequency of more than 1 MHz, and the electromagnetic radiation signal radiated outward during discharge, based on the arc generated when the moving and stationary contacts separate.

[0011] In the method described, waveform testing is performed using an on-load tap changer test platform. A measuring circuit equipped with a DC or AC power supply and a matching resistor is connected to the transformer circuit and a current-limiting resistor is connected in series. During the test, the main on / off contact operates, forcing the current to flow through the transition branch of the tap changer. As the transition resistor is connected and disconnected, the current in the circuit will show a regular change. The current is recorded and compared with a standard waveform to determine whether there is any abnormality during the switching process. If there is no abnormality, the electromagnetic radiation signal of the on-load tap changer under normal operating conditions during switching is collected.

[0012] In the method described, the Teager energy operator is used to filter and preprocess the electromagnetic radiation signal. The Teager energy operator for the continuous electromagnetic radiation signal x(t) is calculated as follows:

[0013]

[0014] t is time, w(t) is the Teager energy of x(t), x(t), and The original signal, first-order difference signal, and second-order difference signal of x(t) are respectively, and all of the above signals change with time.

[0015] The signals obtained through data acquisition are all discrete signals f(n). The Teager energy operator is obtained using three consecutive sampling points, and the calculation method is as follows:

[0016] Y[f(n)]=f 2 (n)-f(n+1)f(n-1)

[0017] Where Y[f(n)] is the Teager energy operator of f(n), f(n-1), f(n) and f(n+1) are three consecutive sample values, and n is a variable.

[0018] In the method described, a local maximum-based peak-finding algorithm is used to extract and compare the size of each data point with its neighborhood to determine the local peak value. To prevent misselection of noise peaks and repeated selection when peaks are dense, a dual-threshold method is used for limitation. The amplitude threshold for the search peak is set as 'a' times the noise signal amplitude in the Teager energy operator to reduce misjudgment of background noise peaks. Then, 'b ms' is set as the minimum time interval between the searched peaks to reduce the impact of dense discharge peaks on the overall result. Finally, the electromagnetic radiation signal is divided into time-domain stages based on the electromagnetic radiation signal peaks to obtain the arc initiation time 't' for each arc. i i = 1 to 4, that is, at the moment of action of each contact, t i For i = 2 to 4, subtract them from t1 in turn, and assign the results to T. i i = 2~4, and finally t1 is set to zero, which is set to the time of switching action 0. Therefore, the action time becomes T. i i = 2~4, T1 = 0; the arc extinguishing time L is obtained by subtracting the arc extinguishing time from the arc initiation time. i i = 1 to 4; simultaneously count the number N of electromagnetic signal spikes during the four switching processes. i i = 1 to 4, T i i = 2~4, L i i = 1 to 4, N i , where i = 1 to 4 are the feature quantities used for data classification.

[0019] In the method described, features are extracted for tap changers under normal operating conditions and different fault types to form a feature matrix and a corresponding tag matrix reflecting the state of the tap changer.

[0020] In the method described, SVM is used to classify matrix data to enable the monitoring of tap changer status using feature quantities.

[0021] Compared with existing technologies, this invention has the following advantages: It utilizes an antenna to obtain electromagnetic radiation information and thus determine the tap changer status without affecting the normal operation of the tap changer; the signal processing algorithm is simple, computationally inefficient, and produces good processing results; it can not only determine the state of the tap changer based on the arcing time L... i (i = 1~4) Determine the state of a single contact of the on-load tap changer (excessive arcing time may indicate severe erosion of the contact surface, making it difficult to extinguish the arc), and also based on the contact actuation time T. i (i = 2~4) Infer whether the contact action timing is normal. For the characteristic quantities of on-load tap changers under normal and different fault states, support vector machine (SVM) can achieve a relatively good differentiation effect. Attached Figure Description

[0022] Various other advantages and benefits of the present invention will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiments below. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. It is obvious that the drawings described below are merely some embodiments of the invention, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort. Furthermore, the same reference numerals denote the same parts throughout the drawings.

[0023] In the attached diagram:

[0024] Figure 1(a) and Figure 1(b) are the voltage waveforms and current waveforms between the contacts and the high-frequency current signal and electromagnetic radiation signal in the circuit tested by the on-load tap changer test platform.

[0025] Figure 2 This is a schematic diagram of the current waveform tested on an on-load tap changer test platform;

[0026] Figures 3(a) to 3(d) Figure 3(a) shows the orderly switching of the contacts of the on-load tap changer test platform at four different times. Figure 3(b) shows the action of the moving contact separating from K1, Figure 3(c) shows the action of the moving contact contacting K3, Figure 3(d) shows the action of the moving contact separating from K2, and Figure 3(d) shows the action of the moving contact contacting K4.

[0027] Figure 4 This is a schematic diagram of a test circuit for simulating tap changer faults;

[0028] Figure 5(a) and Figure 5(b) are the signal waveforms collected by each sensor under normal contact conditions. Figure 5(a) is a schematic diagram of the high-frequency current signal in the circuit, and Figure 5(b) is a schematic diagram of the electromagnetic radiation signal.

[0029] Figure 6 This is a schematic diagram of the signal after TEO processing and peak information extraction;

[0030] Figure 7 A schematic diagram of the characteristic quantities under normal conditions during the gear shifting process from gear 1 to gear 19 in a single measurement;

[0031] Figure 8(a) and Figure 8(b) are schematic diagrams of various characteristic quantities under contact erosion and contact wear failure during the gear shifting process from gear 1 to gear 19 in a single measurement. Figure 8(a) is a schematic diagram of the characteristic quantities of the contact wear state, and Figure 8(b) is a schematic diagram of the characteristic quantities of the contact erosion state.

[0032] The present invention will be further explained below with reference to the accompanying drawings and embodiments. Detailed Implementation

[0033] Specific embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While specific embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the invention and to fully convey the scope of the invention to those skilled in the art.

[0034] It should be noted that certain terms are used in the specification and claims to refer to specific components. Those skilled in the art will understand that different terms may be used to refer to the same component. This specification and claims do not distinguish components based on differences in terminology, but rather on differences in function. The terms "comprising" or "including" used throughout the specification and claims are open-ended and should be interpreted as "comprising but not limited to." The following descriptions are preferred embodiments for carrying out the invention; however, these descriptions are for the purpose of understanding the general principles of the specification and are not intended to limit the scope of the invention. The scope of protection of this invention is determined by the appended claims.

[0035] To facilitate understanding of the embodiments of the present invention, further explanations and descriptions will be provided below with reference to the accompanying drawings and specific embodiments. The accompanying drawings do not constitute a limitation on the embodiments of the present invention.

[0036] To better understand, as shown in Figures 1 to 8(b), in one embodiment, the on-load tap changer status monitoring method based on electromagnetic radiation signals includes the following steps:

[0037] Collect electromagnetic radiation signals during the switching of the on-load tap changer under normal operating conditions.

[0038] The electromagnetic radiation signal is filtered and preprocessed, and the peak information of the electromagnetic radiation signal is extracted using a dual-threshold peak-finding algorithm to obtain the switching timing of the on-load tap changer and the discharge process information between the tap changer contacts. The discharge process information includes characteristic quantities during the switching process.

[0039] The feature quantities are classified using a support vector machine, and the working state of the on-load tap changer is determined based on the classification results.

[0040] In a preferred embodiment of the method, a contact separation simulation test platform is built to realize the energized separation of the moving and stationary contacts of an on-load tap changer. When the moving and stationary contacts separate, an electric arc is generated, and high-frequency current signals and electromagnetic radiation signals are generated when the electric arc is generated and extinguished. Sensors are used to measure the voltage between the contacts, the current flowing through them, the high-frequency current signal with a frequency of more than 1 MHz, and the electromagnetic radiation signal radiated outward during discharge.

[0041] In a preferred embodiment of the method, the contact separation simulation test platform performs waveform testing on the tap changer. The measuring circuit, equipped with a DC or AC power supply and a matching resistor, is connected to the transformer circuit and a current-limiting resistor is connected in series. During the test, the main on / off contacts move, forcing the current to flow through the transition branch of the tap changer. As the transition resistor is connected and disconnected, the current in the circuit will show a regular change. The current is recorded and compared with a standard waveform to determine whether there is an abnormality during the switching process. If there is no abnormality, the electromagnetic radiation signal of the on-load tap changer under normal operating conditions during switching is collected.

[0042] In the method described, the Teager energy operator is used to filter and preprocess the electromagnetic radiation signal. The Teager energy operator for the continuous electromagnetic radiation signal x(t) is calculated as follows:

[0043]

[0044] t is time, w(t) is the Teager energy of x(t), x(t), and The original signal, first-order difference signal, and second-order difference signal of x(t) are respectively, and all of the above signals change with time.

[0045] The signals obtained through data acquisition are all discrete signals f(n). The Teager energy operator is obtained using three consecutive sampling points, and the calculation method is as follows:

[0046] Y[f(n)]=f 2 (n)-f(n+1)f(n-1)

[0047] Where Y[f(n)] is the Teager energy operator of f(n), and f(n-1), f(n) and f(n+1) are three consecutive sample values.

[0048] In a preferred embodiment of the method, a local maximum-based peak-finding algorithm is used to extract and compare the size of each data point with its neighborhood to determine local peak values. To prevent misselection of noise peaks and repeated selection when peaks are dense, a dual-threshold method is used for limitation. The amplitude threshold for the searched peak is set as 'a' times the noise signal amplitude in the Teager energy operator to reduce misjudgment of background noise peaks. Then, 'b ms' is set as the minimum time interval between the searched peaks to reduce the impact of dense discharge peaks on the overall result. Finally, the electromagnetic radiation signal is divided into time-domain stages based on the electromagnetic radiation signal peaks to obtain the arc initiation time 't' for each arc. i i = 1 to 4, that is, at the moment of action of each contact, t i For i = 2 to 4, subtract them from t1 in turn, and assign the results to T. i i = 2~4, and finally t1 is set to zero, which is set to the time of switching action 0. Therefore, the action time becomes T. i i = 2~4, T1 = 0; the arc extinguishing time L is obtained by subtracting the arc extinguishing time from the arc initiation time. i i = 1 to 4; simultaneously count the number N of electromagnetic signal spikes during the four switching processes. i i = 1 to 4, T i i = 2~4, L i i = 1 to 4, N i , where i = 1 to 4 are the feature quantities used for data classification.

[0049] Based on the type of tap changer being measured, the values ​​for a and b can be taken as follows:

[0050] Tap switch type a b / ms Combined oil arc extinguishing 3 0.1 Combined vacuum 3.5 0.15

[0051] In a preferred embodiment of the method, features are extracted for tap changers under normal operating conditions and different fault types to form a feature matrix and a corresponding tag matrix reflecting the tap changer status.

[0052] In a preferred embodiment of the method, SVM is used to classify matrix data to enable monitoring of tap changer status using feature quantities.

[0053] In one embodiment, an on-load tap changer test platform is first constructed. Current clamps, differential probes, high-frequency current sensors, and broadband antennas are used to collect various characteristic signals during on-load tap changer switching. Subsequently, the electromagnetic signals are filtered and preprocessed, and a dual-threshold peak-finding algorithm is used to extract the peak information of the electromagnetic signals, thereby obtaining the switching sequence of the on-load tap changer and the discharge process between the tap changer contacts. Relevant features during the switching process are extracted and classified using a support vector machine, yielding good classification results. This on-load tap changer status monitoring method based on electromagnetic radiation signals is of significant importance in reducing operation and maintenance costs and preventing serious accidents.

[0054] In one embodiment, electromagnetic signals radiated during normal operation and switching of the on-load tap changer to be tested are first collected. The Teager energy operator is used for preprocessing and electromagnetic signal spike information is extracted. Based on this information, the switching sequence of the tap changer and the arcing time during each switching are determined and relevant feature quantities are extracted. Finally, the feature quantities are classified using the support vector machine method. The working state of the on-load tap changer is determined by the classification results, thereby realizing the monitoring of the on-load tap changer status.

[0055] In one embodiment, a contact separation simulation test platform was first built in the laboratory to realize the energized separation of moving and stationary contacts. Sensors were used to measure the voltage between the contacts, the current flowing through them, the high-frequency current (frequency above 1MHz), and the electromagnetic signals radiated outward during discharge. The experiment showed that an electric arc was generated when the moving and stationary contacts separated, and significant high-frequency current and electromagnetic radiation signals were generated at the generation and extinguishing of the arc, as shown in Figures 1(a) and 1(b). Figure 1(a) shows the voltage and current waveforms between the contacts, while Figure 1(b) shows the high-frequency current and electromagnetic radiation signals in the circuit. Significant high-frequency current and electromagnetic radiation peaks were observed at the moment of arc initiation and extinguishing. The timing of these peaks was used to estimate the contact separation time and the arc burning time. In Figures 1(a) and 1(b), the contacts separated at time 0, and the arc burning time was 8ms.

[0056] By utilizing the relationship between the electromagnetic signal spikes during discharge and the switching process, it can be used to determine whether there are any abnormalities in the actual tap changer switching process. Taking the M-type tap changer as an example, the method is as follows:

[0057] (1) First, a waveform test should be performed on the tap changer. Waveform testing is a method used in the electrical industry to verify that the on-load tap changer of a transformer has not been damaged during transportation (transformer manufacturer), to ensure the correct installation and wiring of the tap changer (transformer manufacturer), and finally to confirm that no accidental damage occurred during transportation (on-site hoisting). The circuit for this method is as follows: a measuring circuit equipped with a DC or AC power supply and a matching resistor is connected to the transformer circuit, and a current-limiting resistor is connected in series. During the test, the main on / off contact operates, forcing current to flow through the transition branch of the tap changer. As the transition resistor is connected and disconnected, the current in the circuit will show a regular change. Recording the current and comparing it with a standard waveform can determine whether there is an abnormality during the switching process. This method is more effective when performing waveform testing after the transformer is assembled, transported, and repaired. Due to the presence of winding inductance in the test circuit, the waveform will be distorted, making it difficult to obtain accurate information on the switching time of the transition resistor. Therefore, testing without windings can obtain a current waveform such as... Figure 2As shown. The current changes abruptly at four times: t1, t2, t3, and t4, indicating that the contacts switch sequentially at these four times, corresponding to the four actions of the moving contact separating from K1, contacting K3, separating from K2, and contacting K4, respectively. See [link to relevant documentation]. Figures 3(a) to 3(d) The values ​​t1, t2, t3, and t4 are recorded to serve as a reference for subsequent signal division.

[0058] (2) Electromagnetic signal acquisition: An antenna of appropriate size is installed near the tap changer to ensure that the electromagnetic signal during discharge can be acquired. Then, the tap changer is switched. At this time, due to the change of the tap changer position, the moving and stationary contacts inside the switch separate, changing the current path. An electric arc is generated at the moment of contact separation. When the distance between the two contacts is far enough, the electric arc is extinguished. Therefore, there is an electromagnetic signal spike at the beginning and end of the electric arc during one contact separation process. Since one tap changer position change involves the separation and closure of several contacts, the electromagnetic signal acquired by the antenna contains multiple spikes.

[0059] (3) Due to the complex electromagnetic interference in space, if electromagnetic signal spikes are to be extracted, the electromagnetic signal should be preprocessed first. This invention uses the Teager energy operator (TEO) to process the signal, and its principle is as follows:

[0060] In the method described, the Teager energy operator is used to filter and preprocess the electromagnetic radiation signal. The Teager energy operator for the continuous electromagnetic radiation signal x(t) is calculated as follows:

[0061]

[0062] t is time, w(t) is the Teager energy of x(t), x(t), and The original signal, first-order difference signal, and second-order difference signal of x(t) are respectively, and all of the above signals change with time.

[0063] The signals obtained through data acquisition are all discrete signals f(n). The Teager energy operator is obtained using three consecutive sampling points, and the calculation method is as follows:

[0064] Y[f(n)]=f 2 (n)-f(n+1)f(n-1)

[0065] Where Y[f(n)] is the Teager energy operator of f(n), and f(n-1), f(n) and f(n+1) are three consecutive sample values.

[0066] As shown in the above formula, the higher the amplitude and frequency of the signal, the greater its Teager energy. Therefore, TEO operation can enhance high-frequency signals with insufficient amplitude. When applied to the electromagnetic radiation signal of on-load tap changers, it can amplify the electromagnetic signal energy radiated during the switching of moving and stationary contacts and attenuate electromagnetic interference in space.

[0067] (4) For the processed electromagnetic signal, a peak-finding algorithm based on local maxima is used to extract and compare the size of each data point with its neighborhood to determine the local peak value. At the same time, to prevent the misselection of noise peaks and the repeated selection when peaks are dense, a double threshold method is used for restriction:

[0068] First, the amplitude threshold of the search peak is set as 'a' times the amplitude of the noise signal in TEO to reduce misjudgment of background noise peaks. Then, 'b ms' is set as the minimum time interval between the searched peaks to reduce the impact of dense discharge spikes on the overall results.

[0069] (5) Subsequently, the electromagnetic signal is divided into stages in the time domain based on the electromagnetic signal spikes to obtain the start time t of each arc. i (i = 1 to 4) (at the moment of each contact's action), t i (i = 2 ~ 4) Subtract t1 from each i in turn, and assign the result to T. i (i = 2~4), and finally set t1 to zero, setting it to the time of switching action 0, so the action time becomes T. i (i=2~4)(T1=0); The arcing time L can be obtained by subtracting the arc extinction time from the arc initiation time. i (i = 1 to 4); simultaneously count the number N of electromagnetic signal spikes during the four switching processes. i (i = 1 to 4).

[0070] The above T i (i = 2 ~ 4), L i (i = 1 to 4), N i (i = 1 to 4) are the feature quantities used for data classification.

[0071] (6) For tap changers under normal operating conditions and different fault types, perform feature extraction in steps (1)-(5) above to form a feature matrix and a corresponding label matrix reflecting the tap changer status. Use SVM to classify the matrix data to realize the monitoring of the tap changer status using feature signals.

[0072] In one embodiment, in addition to the normal state of the tap changer, a tap changer fault simulation is performed. This involves replacing one phase's stationary contact with a contact that has been eroded by an arc to simulate contact erosion, and replacing part of the stationary contact in another phase with a custom-designed, thinner contact to simulate contact wear. For characteristic signal acquisition, an HFCT is used to measure the high-frequency current of the circuit, a differential voltage probe is used to measure the voltage between the terminal block on the outer wall of the switching oil chamber and the tap selector tap, a current clamp is used to measure the circuit current, and a broadband antenna is fixed at a height of 1.5m above the ground and a horizontal distance of 0.5m from the switching oil chamber to measure the electromagnetic radiation signal released during arcing, such as... Figure 4 As shown.

[0073] According to steps (1)-(5), the signal waveforms collected by each sensor under normal contact conditions are shown in Figure 5(a) and Figure 5(b).

[0074] As shown in the figure, the switching action phase occurs from -1.2ms to 42.7ms.

[0075] At t1 (-1.2ms), the moving contact separates from K1, the transition resistor R1 is connected to the circuit, the recovery voltage is about 20V, the arcing time is about 1.2ms, and the arc is extinguished at 0ms.

[0076] At t2 (19ms), the moving contact quickly approaches K3. When the distance between them is small enough that the electric field strength between them reaches the breakdown field strength, the insulation breaks down and an electric arc is generated. The electric arc is finally extinguished when the moving contact contacts K3. Then R1 and R2 are connected in parallel into the circuit, with a parallel resistance of about 1.28Ω.

[0077] At t3 (24.7ms), the moving contact separates from K2, the transition resistor R1 is cut out of the circuit, the recovery voltage is about -17V, the arcing time is about 0.7ms, and the arc is extinguished at 25.4ms. There is a significant electromagnetic signal peak when the arc is extinguished.

[0078] At t4 (43.7ms), the moving contact quickly approaches K4, and then the insulation breaks down and an arc is generated near the voltage phase 0°. The arc is finally extinguished when the moving contact contacts K4. Then R2 is short-circuited, the loop current increases, and the switching circuit almost no longer divides the voltage.

[0079] The signal after TEO processing and peak information extraction is as follows Figure 6 As shown:

[0080] The red line represents the switching times t1 to t4, inferred from the electromagnetic signal spikes, while the cyan line represents the arc extinguishing time, used to determine the arc burning time. This electromagnetic signal has eight peaks, which can be divided into four switching steps based on the switching principle. The bridging time t1 is approximately -1.5 ms with only one spike; the bridging time t2 is approximately 19.8 ms with two spikes and an arc burning time of approximately 1.3 ms; the bridging time t3 is approximately 24.7 ms with four spikes and an arc burning time of approximately 0.8 ms; and the bridging time t4 is approximately 42.7 ms with one spike. In summary, the characteristic quantities obtainable using the electromagnetic signal include: the start time of the four switching steps, the arc burning time, and the number and amplitude of signal spikes. These can be used to analyze the arcing process of an on-load tap changer and thus determine the contact state.

[0081] Figure 7 The characteristic quantities of the gear shifting process from gear 1 to gear 19 in a single measurement were statistically analyzed. Taking t1 as the timing 0, the switching time t2 to t4, the arcing time, the number of electromagnetic spikes, and other information were recorded. In the figure, each bar represents the switching process from bottom to top. The lower edge of each red block represents the switching time t1 to t4, and the upper edge represents the arc extinguishing time. Therefore, each red block from bottom to top represents the arcing time of that stage. The blue blocks represent the time period without arc during the switching. In the figure below, the arcing time after time t4 is 0 during the switching of each gear, so there is no fourth red block.

[0082] The switching times are distributed as follows: t2 is distributed within 20.9–21.8 ms, t3 within 24.7–26.8 ms, and t4 within 43.3–45.5 ms. Arcing is concentrated around t1, t2, and t3, with an average arcing time of 1.08 ms and a maximum of 2.7 ms. The number of electromagnetic spikes ranges from 7 to 19, with an average of 12.3. In summary, under normal contact conditions, the switching times are relatively concentrated, and there is no significant difference in the switching sequence when the switching position changes. A brief arcing phenomenon occurs during contact switching, and the arcing process is relatively stable.

[0083] Similarly, the characteristic statistics of contact erosion and contact wear faults can be obtained in Figures 8(a) and 8(b). A total of 384 sets of test data were collected, including 128 sets each under the three conditions of normal contact, contact wear, and contact erosion. The above data were preprocessed and 11 characteristics (T) were extracted. i (i = 2 ~ 4), L i (i = 1 to 4), N i (i = 1 to 4) were used to obtain a 384*11 feature matrix, and the corresponding label matrix was also established. Some typical data are shown in Table 1.

[0084]

[0085]

[0086]

[0087]

[0088]

[0089]

[0090]

[0091]

[0092]

[0093]

[0094] For 384 sets of data, a 5-fold cross-validation theory is introduced. In MATLAB, the input dataset is divided into 5 equal parts. One part is used as the validation set in turn, and the remaining 4 parts are used as the training set. The average of the 5 validation accuracies is taken as an estimate of the algorithm's precision. Classification training is performed using different kernel functions of the support vector machine, yielding the following classification results:

[0095]

[0096] Among the training methods, the linear kernel function had the longest training time but the lowest accuracy (85.9%), indicating that it is not suitable for high-dimensional data spaces. Both quadratic and cubic SVMs achieved high accuracy, but the former was time-consuming. For the Gaussian kernel function, the training times for the three methods were similar, but only the intermediate Gaussian SVM achieved high accuracy. In conclusion, both quadratic and intermediate Gaussian SVMs can be used to distinguish between normal and defective contact samples, but the latter achieves higher accuracy with shorter training time.

[0097] Although embodiments of the present invention have been described above in conjunction with the accompanying drawings, the present invention is not limited to the specific embodiments and application fields described above. The specific embodiments described above are merely illustrative and instructive, and not restrictive. Those skilled in the art can make many other forms based on the guidance of this specification and without departing from the scope of protection of the claims of the present invention, and all of these are within the scope of protection of the present invention.

Claims

1. A method for monitoring the status of on-load tap changers based on electromagnetic radiation signals, characterized in that, It includes the following steps, Collect electromagnetic radiation signals during the switching of the on-load tap changer under normal operating conditions. The electromagnetic radiation signal is filtered and preprocessed, and the peak information of the electromagnetic radiation signal is extracted using a dual-threshold peak-finding algorithm to obtain the switching timing of the on-load tap changer, the discharge process between the tap changer contacts, and the characteristic quantities during the switching process. The feature quantities are classified using a support vector machine, and the working status of the on-load tap changer is determined based on the classification results. The electromagnetic radiation signal is preprocessed using the Teager energy operator filter. The Teager energy operator for the continuous electromagnetic radiation signal x(t) is calculated as follows: ； t represents time, and w(t) represents the Teager energy of x(t). , and The original signal, first-order difference signal, and second-order difference signal of x(t) are respectively, and all of the above signals change with time; The signals obtained through data acquisition are all discrete signals f(n). The Teager energy operator is obtained using three consecutive sampling points, and the calculation method is as follows: ； Where Y[f(n)] is the Teager energy operator of f(n), f(n-1), f(n) and f(n+1) are three consecutive sample values, and n is a variable.

2. The method as described in claim 1, characterized in that, When the moving and stationary contacts separate, an electric arc is generated, and high-frequency electromagnetic radiation signals are generated during the generation and extinction of the electric arc. Electromagnetic radiation signals with a frequency of 1MHz or higher are measured using a sensor.

3. The method as described in claim 1, characterized in that, A local maxima-based peak-finding algorithm is used to extract and compare the size of each data point with its neighborhood to determine local peak values. To prevent misselection of noise peaks and repeated selection when peaks are dense, a dual-threshold method is used. The amplitude threshold for the search peak is set as 'a' times the noise signal amplitude in the Teager energy operator to reduce misjudgment of background noise peaks. Then, 'b ms' is set as the minimum time interval between the searched peaks to reduce the impact of dense discharge peaks on the overall result. Finally, the electromagnetic radiation signal is divided into time-domain stages based on the electromagnetic radiation signal peaks to obtain the arc initiation time 't' for each arc. i i=1~4, that is, at the moment of action of each contact, t i For i = 2 ~ 4, subtract them from t1 in turn, and assign the result to T. i i = 2 ~ 4, and finally t1 is set to zero, which is set to the time of switching action 0. Therefore, the action time becomes T. i i = 2 ~ 4, T1 = 0; the arcing time L is obtained by subtracting the arc extinction time from the arc initiation time. i i=1~4; simultaneously count the number N of electromagnetic signal spikes during the four switching processes. i i=1~4, T i i = 2 ~ 4, L i i=1~4, N i , where i=1~4 are the feature quantities used for data classification.

4. The method as described in claim 3, characterized in that, Features are extracted from tap changers under normal operating conditions and different fault types to form a feature matrix and a corresponding tag matrix reflecting the tap changer status.

5. The method as described in claim 4, characterized in that, Using SVM to classify matrix data enables the monitoring of tap changer status using feature quantities.

Images