Handheld high-frequency partial discharge rapid live detection device

Abstract

The invention relates to a handheld high-frequency partial discharge rapid live detection device, which comprises an acquisition module and an analysis module, and is characterized in that the acquisition module comprises a high-frequency current sensor with a handheld area, and the high-frequency current sensor is provided with a semicircular cobalt-based amorphous magnetic core with an opening; the winding is correspondingly wound at the two ends of the cobalt-based amorphous magnetic core, transmits a detection signal to the analysis module and forms detection data; the analysis module is provided with an interactive interface and a clustering algorithm model, a K-Means two-dimensional Manhattan distance algorithm in the clustering algorithm model is utilized to extract the characteristic quantity of the detection data, a clustering cluster is obtained, whether the clustering cluster exceeds a set threshold value or not is judged, and a detection result of the internal insulation defect of the equipment is output; the open semicircular magnetic core does not need to be opened and closed by a buckle, can directly sleeve the outer side of a line, is convenient to operate, is easy to find the internal insulation defect of equipment in a complex pulse environment, and improves the accuracy of judging the damp discharge and insulation discharge conditions.

Description

Technical Field

[0001] The invention belongs to the technical field of partial discharge detection, and in particular relates to a handheld high-frequency partial discharge rapid charging detection device. Background Art

[0002] With the development of society, the reliability requirements of distribution network power supply are becoming more and more stringent. Overhead lines, pole switches, knife switches, fuses, insulators, lightning arresters, cables, transformers and other 10kV power appliances are not only spread across various roads, communities, and units in the city, but also laid in various villages and countryside, which brings a huge workload to power inspection. During the inspection, the commonly used means are visual inspection, infrared temperature measurement, and ultrasonic imaging. However, these technical means cannot completely detect the internal insulation defects of electrical equipment, such as: discharge of lightning arrester valve plates due to moisture, discharge of cable cross-linked polyethylene main insulation, arc discharge of switch contacts, discharge of PT internal winding insulation, and discharge of transformer internal core clamps.

[0003] Technical means also include high-frequency current pulse partial discharge detection method, but the existing high-frequency current pulse partial discharge detection method has the following problems during operation: 1. The high-frequency current sensor HFCT is a circular or square ring structure. On-site testing and installation requires opening and closing the ring structure and fixing it with buckles, which is relatively cumbersome.

[0004] 2. It is necessary to collect the power frequency phase, which is often quite difficult on 10kV lines. For example, most 10kV cables use three-phase core cables, and the power frequency current or voltage signal cannot be obtained on the outer layer; in addition, the grounding flat iron on the lightning arrester pole and the pole switch pole cannot obtain the power frequency signal.

[0005] 3. The traditional phase windowing method such as PRPD is difficult to analyze and judge, and requires rich practical experience or professionals. Even so, there are problems of misjudgment or missed judgment in complex pulse environments.

[0006] 4. When testing in the field, the traditional portable partial discharge live detector needs to install a circular high-frequency current sensor, connect an acquisition card, and turn on the computer, which is cumbersome and time-consuming, especially for older power workers who conduct grassroots inspections.

[0007] Therefore, designing a handheld high-frequency partial discharge rapid charging detection device that is easy to install and operate and has more accurate analysis and judgment has become a technical problem that needs to be solved urgently. Summary of the invention

[0008] In order to solve the above problems, the present invention provides a handheld high-frequency partial discharge rapid charging detection device.

[0009] The technical solution of the present invention is a handheld high-frequency partial discharge rapid charging detection device, comprising an acquisition module and an analysis module, wherein the acquisition module comprises a high-frequency current sensor provided with a handheld area, wherein the high-frequency current sensor is based on the magnetic field gradient expression and the effective magnetic permeability formula of the amorphous magnetic core, and a semicircular cobalt-based amorphous magnetic core with an opening is provided; a winding is wound around the outer side of the cobalt-based amorphous magnetic core, and the winding is correspondingly wound around the two ends of the cobalt-based amorphous magnetic core, one end of the winding is electrically connected to the housing of the high-frequency current sensor, and the other end is connected in series with a Hall element and a BNC female flange seat in sequence, and the Hall element is electrically connected to the seat core of the BNC female flange seat, so as to convert the collected high-frequency current into a voltage signal , forming detection data and sending it to the analysis module, the detection data includes the test voltage amplitude; the analysis module is connected to the handheld area of ​​the high-frequency current sensor, and the analysis module is provided with an interactive interface and a clustering algorithm model, the clustering algorithm model includes a moisture database and an insulation database obtained by collecting historical experience data of high-frequency partial discharge detection, and a K-Means two-dimensional Manhattan distance algorithm, the K-Means two-dimensional Manhattan distance algorithm performs data preprocessing by acquiring the detection data, extracts the feature quantity of the detection data, combines the moisture database and the insulation database in turn, obtains the clustering cluster and determines whether it exceeds the set threshold value, and the result is that when the threshold value is exceeded, the corresponding moisture test amplitude U is output. (受潮) And insulation test amplitude U (绝缘) , and displayed on the interactive interface.

[0010] As a further improvement of the present invention, the three-dimensional magnetic field gradient expression is: Where Δf is the Hamiltonian operator i, j, k are basis vectors in three mutually orthogonal directions in three-dimensional space, and x, y, z correspond to the coordinate system for the propagation of magnetic lines of force; the magnetic lines of force of the semicircular cobalt-based amorphous magnetic core are based on the principle of the shortest distance of the magnetic circuit, and propagate between the two ends of the magnetic core along the end face direction of the magnetic core, excluding changes in the magnetic lines of force in other directions.

[0011] As a further improvement of the present invention, the effective magnetic permeability formula is: Among them, u e is the effective magnetic permeability, u r is the relative magnetic permeability of the amorphous core, u r ≥800, l1 is the magnetic path length of the core, l2 is the air gap length; by substituting the parameters of the semicircular cobalt-based amorphous core, we get l1=πd / 2, l2=d, and then substituting it into the effective permeability formula to get Where V e is the test voltage amplitude, is the conversion voltage amplitude coefficient corresponding to the winding process and material,

[0012] As a further improvement of the present invention, the inner diameter of the semicircular cobalt-based amorphous magnetic core is not greater than 130 mm, the outer diameter is not greater than 180 mm, and the mass is not greater than 0.4 kg.

[0013] As a further improvement of the present invention, the process of the K-Means two-dimensional Manhattan distance algorithm includes: preprocessing the detection data through wavelet denoising to obtain effective pulse data; extracting multidimensional feature data and reducing the dimension to 2 dimensions to obtain frequency domain feature and time domain feature; reading the data of the moisture database, mixing all the data and specifying the central particle of the learning sample; substituting the learning sample into the distance algorithm formula for iteration to obtain the first round of clustering clusters, the formula is d(i, j)=|X i1 -X J1 |+|X i2 -X J2 |+…+|X ip -X Jp |, where d(i, j) represents the distance between two samples, p represents the number of attributes, i and j represent the index values ​​of the two samples, and the value range is {1, 2, 3.....p}, X ip -X Jp Represents the difference of the pth attribute of two samples; judge whether the first round of clustering exceeds the set threshold, and output the corresponding moisture test amplitude U for data exceeding the threshold (受潮) , the data that does not exceed the threshold is output as other test amplitudes U (其它) ; Read the data from the insulation database and compare it with other test amplitudes U (其它) The corresponding data are mixed, and the central mass point of the learning sample is specified again, and the second round of clustering clusters are obtained iteratively; it is determined whether the second round of clustering clusters exceeds the set threshold, and for the data exceeding the threshold, the corresponding moisture test amplitude U is output (绝缘) .

[0014] After adopting the above method, by aligning the opening of the cobalt-based amorphous magnetic core on the high-frequency current sensor with the cable to be detected, moving the high-frequency current sensor so that the cable passes through the opening to the inside of the magnetic core, starting the high-frequency current sensor to collect local discharge data, and transmitting it to the analysis module through the female flange seat; through the analysis module connected to the handheld area of ​​the high-frequency current sensor, the results output by the clustering algorithm model are directly displayed on the interactive interface. During field detection, there is no need to additionally connect the acquisition card and the computer. The operation is simple and saves the time of operating the buckle to open and close the closed-loop magnetic core, which can better meet the daily inspection needs of grassroots personnel.

[0015] By setting a cobalt-based amorphous core for detection, the magnetic core of the high-frequency current sensor has a higher saturation magnetic induction intensity, and has the advantages of high magnetic permeability, low loss, good temperature stability and aging stability, which can reduce the difficulty of collecting the power frequency phase; by calculating the magnetic field gradient expression of the cobalt-based amorphous core, the cobalt-based amorphous core with an opening is adjusted to obtain a semicircular core structure with the lowest impact on the magnetic circuit; by substituting the parameters of the semicircular cobalt-based amorphous core into the effective magnetic permeability formula, combined with the conversion coefficient of the winding, V is obtained. e ≈2.57u r The magnetic permeability of the semicircular cobalt-based amorphous core is 2.57 times that of the original closed-loop core, making the high-frequency current sensor more sensitive in detecting parameter amplitude.

[0016] By extracting feature quantities through the K-Means two-dimensional Manhattan distance algorithm, it is possible to target the internal defects of 10kV zinc oxide arresters, eliminate the similar central frequency factors in moist electric shock protection and leakage magnetic discharge, and learn and iterate the time domain waveforms with large differences. The iterative process is simple and the analysis difficulty is low. The analysis module does not require the support of large computer equipment and the judgment of experienced professionals. It can be directly fixed on the high-frequency current sensor for use, which is beneficial for grassroots personnel to easily discover internal insulation defects of equipment in complex pulse environments during daily inspections, thereby improving the accuracy of judging moisture discharge and insulation discharge. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 Shown is a schematic diagram of the amorphous core structure.

[0018] Figure 2 The above is a schematic diagram of the magnetic circuit of an amorphous magnetic core.

[0019] Figure 3 The above is a schematic diagram of an equivalent circuit diagram of an amorphous magnetic core.

[0020] Figure 4 The above is a schematic diagram of the clustering algorithm model flow. DETAILED DESCRIPTION

[0021] like Figure 1-Figure 4A handheld high-frequency partial discharge rapid charging detection device is shown, comprising an acquisition module and an analysis module, wherein the acquisition module comprises a high-frequency current sensor provided with a handheld area, wherein the high-frequency current sensor is provided with a semicircular cobalt-based amorphous magnetic core with an opening based on the magnetic field gradient expression and the effective magnetic permeability formula of the amorphous magnetic core; a winding is wound around the outer side of the cobalt-based amorphous magnetic core, wherein the winding is correspondingly wound around the two ends of the cobalt-based amorphous magnetic core, wherein one end of the winding is electrically connected to the housing of the high-frequency current sensor, and the other end is connected in series with a Hall element and a BNC female flange seat in sequence, wherein the Hall element is electrically connected to the seat core of the BNC female flange seat, and the collected high-frequency current is converted into a voltage signal to form a detection signal. The detection data is sent to the analysis module, and the detection data includes the test voltage amplitude; the analysis module is connected to the handheld area of ​​the high-frequency current sensor, and the analysis module is provided with an interactive interface and a clustering algorithm model, and the clustering algorithm model includes a moisture database and an insulation database obtained by collecting historical experience data of high-frequency partial discharge detection, and a K-Means two-dimensional Manhattan distance algorithm. The K-Means two-dimensional Manhattan distance algorithm performs data preprocessing by acquiring the detection data, extracts the feature quantity of the detection data, and sequentially combines the moisture database and the insulation database to obtain a cluster cluster and judge whether it exceeds the set threshold value, and the result is that when the threshold value is exceeded, the corresponding moisture test amplitude U is output. (受潮) And insulation test amplitude U (绝缘) , and displayed on the interactive interface.

[0022] The three-dimensional magnetic field gradient expression is: Where Δf is the Hamiltonian operator i, j, k are basis vectors in three mutually orthogonal directions in three-dimensional space, and x, y, z correspond to the coordinate system for the propagation of magnetic lines of force; the magnetic lines of force of the semicircular cobalt-based amorphous magnetic core are based on the principle of the shortest distance of the magnetic circuit, and propagate between the two ends of the magnetic core along the end face direction of the magnetic core, excluding changes in the magnetic lines of force in other directions.

[0023] The effective magnetic permeability formula is: Among them, u e is the effective magnetic permeability, u r is the relative magnetic permeability of the amorphous core, u r ≥800, l1 is the magnetic path length of the core, l2 is the air gap length; by substituting the parameters of the semicircular cobalt-based amorphous core, we get l1=πd / 2, l2=d, and then substituting it into the effective permeability formula to get Where V e is the test voltage amplitude, is the conversion voltage amplitude coefficient corresponding to the winding process and material,

[0024] The inner diameter of the semicircular cobalt-based amorphous magnetic core is no more than 130 mm, the outer diameter is no more than 180 mm, and the mass is no more than 0.4 kg.

[0025] The process of the K-Means two-dimensional Manhattan distance algorithm includes: preprocessing the detection data through wavelet denoising to obtain effective pulse data; extracting multi-dimensional feature data and reducing the dimension to 2 dimensions to obtain frequency domain feature and time domain feature; reading the data of the moisture database, mixing all the data and specifying the central particle of the learning sample; substituting the learning sample into the distance algorithm formula for iteration to obtain the first round of clustering clusters, the formula is d(i, j)=|X i1 -X J1 |+|X i2 -X J2 |+…+|X ip -X Jp |, where d(i, j) represents the distance between two samples, p represents the number of attributes, i and j represent the index values ​​of the two samples, and the value range is {1, 2, 3.....p}, X ip -X Jp Represents the difference of the pth attribute of two samples; judge whether the first round of clustering exceeds the set threshold, and output the corresponding moisture test amplitude U for data exceeding the threshold (受潮) , the data that does not exceed the threshold is output as other test amplitudes U (其它) ; Read the data from the insulation database and compare it with other test amplitudes U (其它) The corresponding data are mixed, and the central mass point of the learning sample is specified again, and the second round of clustering clusters are obtained iteratively; it is determined whether the second round of clustering clusters exceeds the set threshold, and for the data exceeding the threshold, the corresponding moisture test amplitude U is output (绝缘) .

[0026] By aligning the opening of the cobalt-based amorphous magnetic core on the high-frequency current sensor with the cable to be tested, moving the high-frequency current sensor so that the cable passes through the opening to the inside of the magnetic core, starting the high-frequency current sensor to collect local discharge data, and transmitting it to the analysis module through the female flange seat; through the analysis module connected to the handheld area of ​​the high-frequency current sensor, the results output by the clustering algorithm model are directly displayed on the interactive interface. During field detection, there is no need to additionally connect the acquisition card and the computer. The operation is simple and saves the time of operating the buckle to open and close the closed-loop magnetic core, which can better meet the daily inspection needs of grassroots personnel.

[0027] By setting a cobalt-based amorphous core for detection, the magnetic core of the high-frequency current sensor has a higher saturation magnetic induction intensity, and has the advantages of high magnetic permeability, low loss, good temperature stability and aging stability, which can reduce the difficulty of collecting the power frequency phase; by calculating the magnetic field gradient expression of the cobalt-based amorphous core, the cobalt-based amorphous core with an opening is adjusted to obtain a semicircular core structure with the lowest impact on the magnetic circuit; by substituting the parameters of the semicircular cobalt-based amorphous core into the effective magnetic permeability formula, combined with the conversion coefficient of the winding, V is obtained. e ≈2.57u r The magnetic permeability of the semicircular cobalt-based amorphous core is 2.57 times that of the original closed-loop core, making the high-frequency current sensor more sensitive in detecting parameter amplitude.

[0028] By extracting feature quantities through the K-Means two-dimensional Manhattan distance algorithm, it is possible to target the internal defects of 10kV zinc oxide arresters, eliminate the similar central frequency factors in moist electric shock protection and leakage magnetic discharge, and learn and iterate the time domain waveforms with large differences. The iterative process is simple and the analysis difficulty is low. The analysis module does not require the support of large computer equipment and the judgment of experienced professionals. It can be directly fixed on the high-frequency current sensor for use, which is beneficial for grassroots personnel to easily discover internal insulation defects of equipment in complex pulse environments during daily inspections, thereby improving the accuracy of judging moisture discharge and insulation discharge.

Claims

1. A handheld high-frequency partial discharge rapid charging detection device, comprising a collection module and an analysis module, characterized in that: The acquisition module includes a high-frequency current sensor with a handheld area, and the high-frequency current sensor is provided with a semicircular cobalt-based amorphous magnetic core with an opening based on the magnetic field gradient expression and the effective magnetic permeability formula of the amorphous magnetic core; a winding is wound around the outer side of the cobalt-based amorphous magnetic core, and the winding is correspondingly wound around the two ends of the cobalt-based amorphous magnetic core, one end of the winding is electrically connected to the housing of the high-frequency current sensor, and the other end is connected in series with a Hall element and a BNC female flange seat in sequence, and the Hall element is electrically connected to the seat core of the BNC female flange seat, and the collected high-frequency current is converted into a voltage signal to form detection data and send it to the analysis module, and the detection data includes a test voltage amplitude; The analysis module is connected to the handheld area of ​​the high-frequency current sensor. The analysis module is provided with an interactive interface and a clustering algorithm model. The clustering algorithm model includes a moisture database and an insulation database obtained by collecting historical experience data of high-frequency partial discharge detection, and a K-Means two-dimensional Manhattan distance algorithm. The K-Means two-dimensional Manhattan distance algorithm performs data preprocessing by acquiring detection data, extracts feature quantities of the detection data, and sequentially combines the moisture database and the insulation database to obtain cluster clusters and judge whether they exceed a set threshold. The result is that when the threshold is exceeded, the corresponding moisture test amplitude U is output (受潮) And insulation test amplitude U (绝缘) , and displayed on the interactive interface.

2. A handheld high-frequency partial discharge rapid charging detection device according to claim 1, characterized in that: The three-dimensional magnetic field gradient expression is: Where Δf is the Hamiltonian operator i, j, k are basis vectors in three mutually orthogonal directions in three-dimensional space, and x, y, z correspond to the coordinate system for the propagation of magnetic lines of force; the magnetic lines of force of the semicircular cobalt-based amorphous magnetic core are based on the principle of the shortest distance of the magnetic circuit, and propagate between the two ends of the magnetic core along the end face direction of the magnetic core, excluding changes in the magnetic lines of force in other directions.

3. According to claim 1, a handheld high-frequency partial discharge rapid charging detection device is characterized in that: The effective magnetic permeability formula is: Among them, u e is the effective magnetic permeability, u r is the relative magnetic permeability of the amorphous core, u r ≥800, l1 is the magnetic path length of the core, l2 is the air gap length; by substituting the parameters of the semicircular cobalt-based amorphous core, we get l1=πd / 2, l2=d, and then substituting it into the effective permeability formula to get Where V e is the test voltage amplitude, is the conversion voltage amplitude coefficient corresponding to the winding process and material, 4. According to claim 1, a handheld high-frequency partial discharge rapid charging detection device is characterized in that: The inner diameter of the semicircular cobalt-based amorphous magnetic core is no greater than 130 mm, the outer diameter is no greater than 180 mm, and the mass is no greater than 0.4 kg.

5. According to claim 1, a handheld high-frequency partial discharge rapid charging detection device is characterized in that: The process of the K-Means two-dimensional Manhattan distance algorithm includes: preprocessing the detection data by wavelet denoising to obtain effective pulse data; Extract multi-dimensional feature data and reduce it to 2 dimensions to obtain frequency domain feature and time domain feature; read the data of the moisture database, mix all the data and specify the central mass point of the learning sample; Substitute the learning sample into the distance algorithm formula for iteration to obtain the first round of clustering clusters, the formula is d(i, j) = |X i1 -X J1 |+|X i2 -X J2 |+…+|X ip -X Jp |, where d(i, j) represents the distance between two samples, p represents the number of attributes, i and j represent the index values ​​of the two samples, and the value range is {1, 2, 3.....p}, X jp -X Jp Represents the difference of the pth attribute of two samples; judge whether the first round of clustering exceeds the set threshold, and output the corresponding moisture test amplitude U for data exceeding the threshold (受潮) , the data that does not exceed the threshold is output as other test amplitudes U (其它) ; Read the data from the insulation database and compare it with other test amplitudes U (其它) The corresponding data are mixed, and the central mass point of the learning sample is specified again, and the second round of clustering clusters are obtained iteratively; it is determined whether the second round of clustering clusters exceeds the set threshold, and for the data exceeding the threshold, the corresponding moisture test amplitude U is output (绝缘) .

Images