Converter transformer partial discharge monitoring and positioning system based on quaternary antenna array

Abstract

The invention discloses a converter transformer partial discharge monitoring and positioning system based on a quaternary antenna array, and the system comprises a control host which is connected with a plurality of data collection units through a calculation processing unit, and is also connected with an upper computer through a wireless communication module. Each data acquisition unit is connected with the corresponding broadband antenna through the respective analog signal processing unit; each analog signal processing unit is used for denoising the electromagnetic wave signal received by the respective broadband antenna; each data acquisition unit is used for performing analog-to-digital conversion on the electromagnetic wave signal processed by the respective analog signal processing unit; the calculation processing unit is used for performing data analysis on the converted electromagnetic wave signal to obtain index data required by monitoring; and the control host is used for receiving a monitoring task instruction of the upper computer, running a monitoring control software program, controlling the analog signal processing unit and the data acquisition unit to execute tasks, controlling the calculation processing unit to perform data analysis, and storing an analysis result to the solid state disk or transmitting the analysis result to the upper computer in real time, or calling historical partial discharge data from the solid state disk.

Description

Technical Field

[0001] The present invention relates to the technical field of power system fault detection, and in particular to a commutation transformer partial discharge monitoring and positioning system based on a four-element antenna array. Background Art

[0002] In the context of building a new power system, UHV DC transmission projects are facing unprecedented development opportunities. As the core equipment in the converter station, the converter transformer is located at the intersection of the AC and DC power grids, and undertakes the vital electrical connection and voltage conversion tasks. Its working status is directly related to the safety and stability of the entire system.

[0003] According to statistics, the insulation condition of the converter transformer is one of the key factors affecting its performance. However, for the converter transformers that have been put into use, in addition to regular power outage inspections, the insulation status monitoring methods during operation mainly include oil chromatography analysis, ultrasonic partial discharge detection and high-frequency current partial discharge detection. These methods are often greatly reduced in effectiveness due to the influence of unfavorable factors such as complex electromagnetic interference on site and the vibration and noise of the equipment itself. In particular, the oil chromatography analysis is slow and difficult to capture sudden faults, while ultrasonic and high-frequency current partial discharge detection are difficult to accurately evaluate due to interference problems. Summary of the invention

[0004] The present invention provides a satellite remote sensing real-time monitoring method based on spatiotemporal feature fusion, which solves the problems that the threshold of the existing threshold method for detecting fire points is fixed and cannot be changed according to actual conditions, and the semantic segmentation model usually requires the input image to have a high spatial resolution and is difficult to apply to geosynchronous satellite data.

[0005] In order to solve the above technical problems, the present invention provides a partial discharge monitoring and positioning system for a commutator transformer based on a four-element antenna array. This technology has the characteristics of non-contact signal collection, will not cause physical damage to the main power equipment or change its grounding structure, and at the same time has the advantages of being safe, easy to deploy, and able to quickly achieve device-level positioning and distinguish interference. It is very suitable for live partial discharge detection of electrical equipment.

[0006] The present invention can be achieved through the following technical solutions:

[0007] A quaternary antenna array-based partial discharge monitoring and positioning system for commutation transformers includes a control host, the control host is connected to a plurality of data acquisition units via a computing processing unit, and is also connected to a host computer via a wireless communication module, each of the data acquisition units is connected to a corresponding broadband antenna via a respective analog signal processing unit,

[0008] Each of the analog signal processing units is used to perform denoising processing on the electromagnetic wave signal received by the respective broadband antenna;

[0009] Each of the data acquisition units is used to perform analog-to-digital conversion on the electromagnetic wave signal processed by the corresponding analog signal processing unit;

[0010] The computing and processing unit is used to perform data analysis on the converted electromagnetic wave signal to obtain the index data required for monitoring;

[0011] The control host is used to receive monitoring task instructions from the host computer, run the monitoring control software program, control the analog signal processing unit and the data acquisition unit to perform tasks, and control the computing processing unit to perform data analysis, store the analysis results in a solid-state hard disk or transmit them to the host computer in real time, or retrieve historical partial discharge data from the solid-state hard disk.

[0012] Further, the computing processing unit is connected to the control host via a serial port line;

[0013] The control host is connected to the solid state drive via a data cable, and its host kernel is set to RK3556-CPU based on ARM architecture, and communicates with the computing processing unit via an RS-232 interface.

[0014] Furthermore, the broadband antenna, the analog signal processing unit, and the data acquisition unit are each provided in four, and they are provided in one-to-one correspondence.

[0015] Further, the monitoring and control software program includes the following steps:

[0016] Step 1: Supply high-frequency signals from four broadband antennas to four analog signal processing units respectively in real time;

[0017] Step 2: The control host receives multiple monitoring tasks issued by the host computer in real time. The multiple monitoring tasks include indicator measurement tasks, indicator storage tasks, spectrum scanning tasks, positioning tasks, and historical query tasks;

[0018] Step 3: Control four analog signal processing units and data acquisition units to perform signal processing and sampling tasks;

[0019] Step 4: The computing and processing unit receives the output data from the four data acquisition units and completes multiple monitoring tasks according to the priority of each monitoring task and the preset logic;

[0020] Step 5: The control host stores the completion results of each task in the solid state drive or transmits them to the host computer in real time, or retrieves historical partial discharge data from the solid state drive to execute historical query tasks.

[0021] Furthermore, when performing monitoring tasks, clustering algorithms are used to perform classification and denoising to screen out electromagnetic wave signals.

[0022] Further, the clustering algorithm comprises the following steps:.

[0023] Step a: Assume that there are M electromagnetic wave pulse signals {x1, x2, …, x M}, initially each electromagnetic wave pulse signal is regarded as an individual class, and the similarity distance d between each two classes is calculated using the following formula i,j ;

[0024] d xy =b-dist(x,y)

[0025]

[0026]

[0027] Where x, y represent two electromagnetic wave pulse signals, b represents a constant, Disp represents the cross-correlation information, Disp∈[0,1], TCD∈[0,2], α represents the weight coefficient between TCD and Disp, α∈[0,1], m represents the total number of time domain points of the electromagnetic wave pulse signal, δ represents the moving factor of the electromagnetic wave pulse signal x, and its range is (0,m), xi and yi represent the time domain sample points of the two electromagnetic wave pulse signals, x i+δ Represents the moving sample point of xi, and the subscript represents the time;

[0028] Step b: Set these similarity distances d i,j Arrange them in order from large to small, select the two categories with the smallest similarity distance and merge them into one category. At this time, the number of categories is M-1;

[0029] Step c, recalculate the similarity distance between the newly generated class and the remaining classes;

[0030] Step d, repeatedly iterate the merging process, that is, repeat steps bc until all clusters are merged into two clusters, calculate the average value of the peak values ​​of the pulse signals in the two clusters, and select the cluster with the larger average value as the partial discharge signal cluster.

[0031] Furthermore, when performing a positioning task, a positioning algorithm is used to confirm the specific location point of the partial discharge.

[0032] Further, the positioning algorithm comprises the following steps:

[0033] Step A, dividing the monitoring area of ​​the converter transformer subject into a plurality of small grid areas, a total of n;

[0034] Step B: Set the position coordinates Sj(xj,yj) of the four broadband antennas, j=1…4, and for each grid node Ni(xi,yi), i=1…n, calculate the distance Lij from each grid node Ni to the four broadband antennas Sj, and then obtain the propagation time t from the grid node Ni to the antenna Sj according to the wave speed c and Lij of the electromagnetic wave. ij ;

[0035] Step C: Construct a theoretical arrival time difference vector, that is, for each grid node Ni, calculate its arrival time for the four broadband antennas and obtain the theoretical arrival time difference vector of each grid node Ni:

[0036] Step D: The time difference vector of the actual partial discharge signal arriving at the four broadband antennas obtained after actual monitoring cluster analysis is the actual arrival time difference vector t of the array. P , record the time points at which the same real PD signal reaches the four broadband antennas respectively, and calculate the actual arrival time difference vector t according to step C P ,

[0037] The spatial grid optimization algorithm is used to calculate the theoretical arrival time difference vector of each grid point using the following formula: The actual arrival time difference vector t P The shortest distance to the grid node N k The coordinates are the location coordinates of the partial discharge signal.

[0038]

[0039] Further, a two-dimensional coordinate system is established with the lower left corner of the converter as the origin of the coordinate system, and the specific position coordinates of the four broadband antennas are set to broadband antenna 1 (0m, Wm), broadband antenna 2 (0m, 0m), broadband antenna 3 (Cm, 0m), broadband antenna 4 (Cm, W+2m),

[0040] Wherein, W is the width of the converter transformer, and C is the length of the converter transformer.

[0041] The beneficial technical effects of the present invention are:

[0042] 1) In view of the current lack of timely, effective and safe means for detecting partial discharge of converter transformers, the present invention proposes a system and method for monitoring and locating partial discharge of converter transformers based on broadband radio frequency. The system has non-contact signal acquisition capabilities, does not require physical intervention in the main power equipment or change its grounding structure, can quickly complete the location and identification of partial discharge and interference signals at the equipment level, has the advantages of high safety and easy deployment, and effectively makes up for the shortcomings of existing partial discharge detection technology.

[0043] 2) The present invention uses a high-speed sampling rate of 2.3GHz to capture the original time domain waveform of the electromagnetic pulse signal, and uses a hierarchical clustering algorithm based on the cross-correlation of the original waveform to classify the signal. This method shows excellent signal classification performance. At the same time, it uses the arrival time difference method to accurately locate the device level, and uses a spatial grid search algorithm to optimize the solution. This method is not only computationally efficient, but also particularly suitable for engineering practice. Combining cluster analysis and positioning technology, the present invention can effectively distinguish partial discharge signals from interference signals, showing excellent interference recognition capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a schematic diagram of the system framework of the present invention;

[0045] Figure 2 A schematic diagram of the flow of the monitoring and control application program of the present invention;

[0046] Figure 3 It is a schematic diagram of the flow of the clustering algorithm of the present invention;

[0047] Figure 4 It is a flowchart of the positioning algorithm of the present invention;

[0048] Figure 5 It is a schematic diagram of the time-frequency waveform of the partial discharge and interference signal of the commutation transformer of the present invention;

[0049] Figure 6 It is a schematic diagram of the cross-correlation analysis result between the partial discharge signal and the interference signal of the present invention;

[0050] Figure 7 A schematic diagram of positioning results using the positioning algorithm of the present invention;

[0051] Figure 8 The figure is a schematic diagram of error distribution of the positioning algorithm of the present invention and the existing positioning algorithm. DETAILED DESCRIPTION

[0052] The specific implementation of the present invention is described in detail below with reference to the accompanying drawings and preferred embodiments.

[0053] like Figure 1As shown, the present invention provides a broadband radio frequency converter transformer partial discharge monitoring and positioning system, which samples the original time domain waveform of the electromagnetic wave pulse signal at a high rate of 2.3GHz, and classifies the signal using a hierarchical clustering algorithm based on the cross-correlation of the original waveform. At the same time, by accurately judging the arrival time of the pulse signal, the arrival time difference method is used to achieve equipment-level fault location, and the monitoring area grid search algorithm is used to solve the specific positioning problem of partial discharge faults. In this way, using cluster analysis and positioning technology, the system can effectively distinguish partial discharge signals from interference signals, showing a strong interference recognition ability.

[0054] The details are as follows:

[0055] First, as Figure 1 As shown, the partial discharge monitoring and positioning system of the commutation transformer based on the four-element antenna array of the present invention mainly includes a host computer, a control host, a solid state hard disk, four broadband antennas, four analog signal processing units, four data acquisition units, a calculation processing unit and a module power supply.

[0056] The control host is connected to a plurality of data acquisition units through a computing processing unit, and is also connected to a host computer through a wireless communication module, and each of the data acquisition units is connected to a corresponding broadband antenna through a respective analog signal processing unit; here, four broadband antennas, four analog signal processing units, and four data acquisition units are provided, and they are provided one by one, and the four broadband antennas are respectively connected to the four analog signal processing units through high-frequency lines, and the four analog signal processing units are respectively connected to the four data acquisition units through serial port lines, and the four data acquisition units are respectively connected to the computing processing unit through serial port lines, and the computing processing unit is connected to the control host through a serial port line;

[0057] Each analog signal processing unit is used to perform noise reduction on the electromagnetic wave signal received by its corresponding antenna. It can reduce noise through low-noise amplification, gain control and optional bandpass filtering. At this stage, periodic narrowband interference, such as radio broadcast interference and mobile phone communication interference, can be filtered out;

[0058] Each data acquisition unit is used to perform analog-to-digital conversion on the electromagnetic wave signal processed by the corresponding analog signal processing unit;

[0059] The calculation and processing unit is used to perform data analysis on the converted electromagnetic wave signal to obtain the index data required for monitoring; including clustering algorithm, positioning algorithm, partial discharge spectrum phase diagram, etc. After calculation and analysis, the original time domain waveform of the partial discharge signal, positioning results and other technical indicators are sent to the control host through the serial port line;

[0060] The control host is used to receive the monitoring task instructions from the host computer, run the monitoring control software program, control the analog signal processing unit and the data acquisition unit to perform multiple monitoring tasks, and control the calculation processing unit to perform data analysis, store the analysis results in the solid-state hard disk or transmit them to the host computer in real time, or retrieve historical partial discharge data from the solid-state hard disk to perform historical query tasks. Here, the control host is connected to the solid-state hard disk via a data cable and connected to the host computer via a network cable;

[0061] Among them, multiple monitoring tasks include indicator collection and measurement tasks, indicator storage tasks, spectrum scanning tasks, positioning tasks, historical query tasks, etc. The results after data analysis by the computing and processing unit can be uploaded to the host computer for display, such as the corresponding partial discharge spectrum phase diagram, which highlights the abnormal pulse signal and alarms. Then, according to the command control of the host computer, further positioning analysis is carried out to obtain the accurate partial discharge location point, etc.

[0062] The module power supply is connected with the control host, the calculation processing unit, the data acquisition unit and the analog signal processing unit through a power line.

[0063] The solid state drive is used to store system programs, monitoring and control application software programs, various instructions and indicator data, databases, etc.

[0064] The specific position coordinates of these four broadband antennas need to be measured and calibrated. A two-dimensional coordinate system is established with the lower left corner of the converter transformer as the origin of the coordinate system: broadband antenna 1 (0m, Wm), broadband antenna 2 (0m, 0m), broadband antenna 3 (Cm, 0m), and broadband antenna 4 (Cm, W+2m), where W is the width of the converter transformer and C is the length of the converter transformer.

[0065] like Figure 2 As shown, after receiving the task instruction sent by the host computer, the control host runs the monitoring and control software program to perform the task according to the following steps:

[0066] Step 1: Supply high-frequency signals from four broadband antennas to four analog signal processing units respectively in real time;

[0067] Step 2: The control host receives multiple monitoring tasks issued by the host computer in real time, including indicator measurement tasks, indicator storage tasks, spectrum scanning tasks, positioning tasks, historical query tasks, etc.

[0068] Step 3: The four analog signal processing units and the data acquisition unit perform signal processing and sampling tasks according to the instructions issued by the control host;

[0069] Step 4: The computing and processing unit receives the output data results from the four data acquisition units and completes multiple monitoring tasks according to the priority of each monitoring task according to the preset logic. For example, it can first monitor the abnormal pulse signal, and once found, it will immediately alarm, and then perform positioning analysis, etc.;

[0070] like Figure 3 As shown, the clustering algorithm includes: using an improved cosine similarity function to describe the similarity between multi-source partial discharge signals and interference pulse signals, and the function is defined as:

[0071]

[0072] TCD is the traditional cosine similarity, x and y are two time domain signals, such as electromagnetic wave pulse signals, Disp is the cross-correlation information, Disp∈[0,1], TCD∈[0,2], α is the weight coefficient of TCD and Disp, α∈[0,1].

[0073]

[0074] Among them, xi and yi represent two time domain sample points of electromagnetic wave pulse signal, x i+δ It represents the moving sample point of xi, the subscript represents the time, m is the total number of time domain points of the signal, and δ is the moving factor of the signal x, which ranges from (0, m).

[0075] To this end, the difference between TCD and the constant b is defined as the similarity distance between the two time domain signals, that is,

[0076] d xy =b-dist(x,y)

[0077] And d xy The smaller it is, the higher the similarity of the time domain waveforms is.

[0078] The specific steps are as follows:

[0079] Step a. Assume that there are M pulse signals {x1, x2, …, x M}, the four broadband antennas perform cyclic detection according to a preset time, such as 10s, and then aggregate the collected electromagnetic wave pulse signals into a data group. Initially, each pulse signal is regarded as an individual class separately, and the above formula d xy = b-TCD(x,y) calculates the similarity distance d between each two categories i,j ;

[0080] Step b. Set these similarity distances d i,j Arrange them in order from large to small, select the two categories with the smallest similarity distance and merge them into one category. At this time, the number of categories is M-1;

[0081] Step c. Recalculate the similarity distance between the newly generated class and the remaining classes;

[0082] Step d. Repeat the iterative merging process, that is, repeat steps bc until all clusters are merged into two clusters, calculate the average value of the peak values ​​of the pulse signals in the two clusters, and select the cluster with the larger average value as the partial discharge signal cluster.

[0083] The time domain waveform and frequency domain waveform of the partial discharge signal and the interference signal are shown in Figure 5 , the left side shows the time domain waveform and frequency domain waveform of the partial discharge signal, and the right side shows the time domain waveform and frequency domain waveform of the interference signal. Although both signals show steep pulse characteristics, the difference between them can be clearly seen from the waveforms. The peak value of the partial discharge (PD) signal is about 10 mV. In the time domain waveform, the selected blue part does not contain the steep pulse signal, while the white part contains these signals. These two parts correspond to the blue line and the white line in the frequency domain waveform, respectively. By comparison, it can be seen that the spectral components of the partial discharge signal are mainly concentrated below 600 MHz. On the other hand, the peak value of the interference signal is about 8 mV. From the comparison of the frequency domain waveforms, it can be seen that its spectral components are mainly distributed below 400 MHz. Figure 6 The results of the cross-correlation analysis between the PD signal and the interference signal are shown. The maximum value of the cross-correlation function is 0.175, which indicates that the waveform similarity of the two signals is very low.

[0084] After that, the partial discharge signal after clustering algorithm analysis is spatially located. Usually, a sensor array composed of four broadband antennas is used to detect the electromagnetic wave signal of partial discharge. By reading the starting time of each signal and making a difference, six relative propagation time values ​​(i.e., propagation time difference or arrival time difference) are obtained. Then, based on these propagation time differences, a propagation time-distance difference equation group is established, also known as a time difference equation group. For example, if the propagation time from the discharge radiation source P to the sensor S1 is t1, then the distance r1 from point P to S1 satisfies r1=c*t1, where c is the wave velocity of the electromagnetic wave. For other sensors Si, the corresponding distance relationship equation can be established according to the relative time difference 1i=t-ti (i=2,3,4) between the signal reaching the reference sensor S1 and the sensor Si. Finally, by solving the above time difference equation group, the position coordinates of the discharge source are obtained.

[0085] In this implementation case, we use a spatial grid optimization algorithm to solve this set of equations, namely the positioning algorithm. This method is based on the principle of the shortest optical path of electromagnetic wave propagation and has the advantages of high computational efficiency and suitability for engineering applications. Figure 4 As shown, the specific steps are:

[0086] Step A, dividing the monitoring area of ​​the converter transformer into a plurality of small grid areas, n in total, each grid representing a potential partial discharge source location;

[0087] Step B: Set the grid spacing size to 2 meters and determine the position coordinates Sj(xj,yj) of the four broadband antennas (j=1…4). For each grid node Ni(xi,yi), i=1…n, calculate its distance to the four broadband antennas as Lij. Then, according to the electromagnetic wave speed c and Lij, the propagation time t from the grid node Ni to the antenna Sj is obtained. ij ;

[0088] Step C: Construct the theoretical arrival time difference vector, that is, for each grid node Ni, calculate its arrival time for the four broadband antennas and obtain the theoretical arrival time difference vector of each grid node

[0089] Step D: The time difference vector of the actual partial discharge signal arriving at the four broadband antennas obtained after cluster analysis in actual monitoring is the actual arrival time difference vector t of the array P , we can record the time points at which the same real PD signal reaches the four broadband antennas respectively, and calculate the actual arrival time difference vector t according to step C. P ,

[0090] The spatial grid optimization algorithm is used to calculate the theoretical arrival time difference vector of each grid point using the following formula: The actual arrival time difference vector t P The shortest distance to the grid node N k The coordinates are the location coordinates of the partial discharge signal.

[0091]

[0092] Step 5: The control host stores the completion result of the monitoring task in the solid state drive or transmits it to the host computer in real time, or retrieves historical partial discharge data such as partial discharge signal waveform data and historical indicator data from the solid state drive to execute the historical query task.

[0093] At the same time, it can also monitor each hardware device and software program in real time to determine whether a partial discharge signal occurs. If so, an alarm signal is immediately sent to the host computer, and a fault clearance signal is sent after the fault is cleared. The alarm information is then stored in the local database according to a predetermined data frame structure for future reference.

[0094] In order to verify the feasibility of the quaternary antenna array based commutation transformer partial discharge monitoring and positioning system of the present invention, we conducted the following experiments:

[0095] After the detected pulse signal was located at the equipment level in a substation in Shanghai, the positioning results are as follows: Figure 7 In order to more intuitively show the relative position relationship between the converter transformer, broadband antenna and positioning results, the top view of the converter transformer is added to the figure. Figure 7 It can be seen that the signal marked by the green dot (suspected to be a partial discharge signal) is precisely located in the vicinity of the bushing on the grid side of the converter transformer. Based on this location information, it can be preliminarily inferred that the signal is likely to come from the gap on the right side of the bushing or the converter transformer body, and these signals have leaked into the surrounding space. The subsequent actual disassembly work verified this point, that is, the actual partial discharge defect is indeed located in the upper part of the grid coil of the converter transformer column 1; on the other hand, the red dots represent interference signals, which are located on the right side of the monitoring area. Combined with the actual situation on site, there is a running oil filter near this location, which may be the main cause of these interference signals. In this way, not only can the partial discharge signal be effectively distinguished from the external interference source, but it can also provide an important reference for subsequent fault diagnosis and maintenance.

[0096] In order to verify the effect of our positioning algorithm, typical TDOA algorithms were selected for comparison, including the two-step weighted least squares method (TS-WLS), which improves robustness in noisy environments through step-by-step optimization, but is more complex to implement and its performance depends on the quality of the position estimate obtained in the first step. The Chan-Ho algorithm uses linearization to simplify the calculation and is suitable for small-scale positioning. When the target is far away, the positioning accuracy will decrease due to the increase in error caused by linear approximation. The Taylor series expansion method is suitable for dealing with nonlinear problems and provides high accuracy but relies on good initial values; the closed-form solution (CFS) provides a fast and direct analytical solution, which is usually only applicable under specific conditions, such as when the distance between receivers is relatively small. In addition, these methods may be sensitive to measurement errors. The particle filter (PF) performs well in dealing with dynamic and nonlinear scenarios, but has a large computational overhead because a large number of particles need to be maintained to represent the posterior probability distribution, and there is a problem of particle degradation, which requires regular resampling to maintain diversity. Maximum likelihood (GCC) can provide the optimal solution under ideal noise conditions, but the computational complexity is high, especially when searching for the global optimal solution in a multidimensional search space, a lot of computing resources are required; in addition, this method is very sensitive to the choice of initial values, and inaccurate initial guesses may lead to convergence to local extreme values ​​rather than the global optimal solution; the PSO algorithm is easy to implement, explores the solution space by simulating the social behavior of bird flocks or fish schools, has good global search capabilities, and is suitable for solving nonlinear and non-convex optimization problems; however, PSO also has some disadvantages, including the possibility of falling into local optimal solutions, critical parameter settings (such as inertia weights, acceleration coefficients, etc.), and slow convergence when dealing with large-scale problems.

[0097] The above algorithm is used to solve 10 times, and the error with the actual partial discharge source coordinates is calculated, such as Figure 8 The 10 solutions of other TDOA algorithms are unstable and have large errors, while the error results of the algorithm proposed in this paper have small fluctuations and are relatively small and stable.

[0098] Although specific embodiments of the present invention are described above, those skilled in the art should understand that these are merely examples and that various changes or modifications may be made to these embodiments without departing from the principles and essence of the present invention. Therefore, the scope of protection of the present invention is limited by the appended claims.

Claims

1. A partial discharge monitoring and positioning system for commutation transformer based on a four-element antenna array, characterized in that: The control host comprises a control host, which is connected to a plurality of data acquisition units through a computing processing unit and is also connected to a host computer through a wireless communication module. Each of the data acquisition units is connected to a corresponding broadband antenna through a respective analog signal processing unit. Each of the analog signal processing units is used to perform denoising processing on the electromagnetic wave signal received by the respective broadband antenna; Each of the data acquisition units is used to perform analog-to-digital conversion on the electromagnetic wave signal processed by the corresponding analog signal processing unit; The computing and processing unit is used to perform data analysis on the converted electromagnetic wave signal to obtain the index data required for monitoring; The control host is used to receive monitoring task instructions from the host computer, run the monitoring control software program, control the analog signal processing unit and the data acquisition unit to perform tasks, and control the computing processing unit to perform data analysis, store the analysis results in a solid-state hard disk or transmit them to the host computer in real time, or retrieve historical partial discharge data from the solid-state hard disk.

2. The partial discharge monitoring and positioning system for commutation transformer based on four-element antenna array according to claim 1 is characterized in that: The computing processing unit is connected to the control host via a serial port line; The control host is connected to the solid state drive via a data cable, and its host kernel is set to RK3556-CPU based on ARM architecture, and communicates with the computing processing unit via an RS-232 interface.

3. The quaternary antenna array-based partial discharge monitoring and positioning system for commutation transformers according to claim 2 is characterized in that: The broadband antenna, the analog signal processing unit and the data acquisition unit are each provided with four, and they are provided in one-to-one correspondence.

4. The quaternary antenna array-based partial discharge monitoring and positioning system for commutation transformers according to claim 1 is characterized in that The monitoring and control software program includes the following steps: Step 1: Supply high-frequency signals from four broadband antennas to four analog signal processing units respectively in real time; Step 2: The control host receives multiple monitoring tasks issued by the host computer in real time. The multiple monitoring tasks include indicator measurement tasks, indicator storage tasks, spectrum scanning tasks, positioning tasks, and historical query tasks; Step 3: Control four analog signal processing units and data acquisition units to perform signal processing and sampling tasks; Step 4: The computing and processing unit receives the output data from the four data acquisition units and completes multiple monitoring tasks according to the priority of each monitoring task and the preset logic; Step 5: The control host stores the completion results of each task in the solid state drive or transmits them to the host computer in real time, or retrieves historical partial discharge data from the solid state drive to execute historical query tasks.

5. The quaternary antenna array-based partial discharge monitoring and positioning system for commutation transformers according to claim 4 is characterized in that: When performing monitoring tasks, clustering algorithms are used to classify and denoise to screen out electromagnetic wave signals.

6. The quaternary antenna array-based partial discharge monitoring and positioning system for commutation transformers according to claim 5 is characterized in that The clustering algorithm comprises the following steps: Step a: Assume that there are M electromagnetic wave pulse signals {x1, x2, …, x M }, initially each electromagnetic wave pulse signal is regarded as an individual class, and the similarity distance d between each two classes is calculated using the following formula i,j ; d xy =b-dist(x,y) Where x, y represent two electromagnetic wave pulse signals, b represents a constant, Disp represents the cross-correlation information, Disp∈[0,1], TCD∈[0,2], α represents the weight coefficient between TCD and Disp, α∈[0,1], m represents the total number of time domain points of the electromagnetic wave pulse signal, δ represents the moving factor of the electromagnetic wave pulse signal x, and its range is (0,m), xi and yi represent the time domain sample points of the two electromagnetic wave pulse signals, x i+δ Represents the moving sample point of xi, and the subscript represents the time; Step b: Set these similarity distances d i,j Arrange them in order from large to small, select the two categories with the smallest similarity distance and merge them into one category. At this time, the number of categories is M-1; Step c, recalculate the similarity distance between the newly generated class and the remaining classes; Step d, repeatedly iterate the merging process, that is, repeat steps bc until all clusters are merged into two clusters, calculate the average value of the peak values ​​of the pulse signals in the two clusters, and select the cluster with the larger average value as the partial discharge signal cluster.

7. The quaternary antenna array-based partial discharge monitoring and positioning system for commutation transformers according to claim 4 is characterized in that: When performing positioning tasks, a positioning algorithm is used to confirm the specific location of the local discharge.

8. The quaternary antenna array-based partial discharge monitoring and positioning system for commutation transformers according to claim 7 is characterized in that The positioning algorithm comprises the following steps: 。 Step A, dividing the monitoring area of ​​the converter transformer subject into a plurality of small grid areas, a total of n; Step B: Set the position coordinates Sj(xj,yj) of the four broadband antennas, j=1…4, and for each grid node Ni(xi,yi), i=1…n, calculate the distance Lij from each grid node Ni to the four broadband antennas Sj, and then obtain the propagation time t from the grid node Ni to the antenna Sj according to the wave speed c and Lij of the electromagnetic wave. ij ; Step C: Construct a theoretical arrival time difference vector, that is, for each grid node Ni, calculate its arrival time for the four broadband antennas and obtain the theoretical arrival time difference vector of each grid node Ni: Step D: The time difference vector of the actual partial discharge signal arriving at the four broadband antennas obtained after actual monitoring cluster analysis is the actual arrival time difference vector t of the array. P , record the time points at which the same real PD signal reaches the four broadband antennas respectively, and calculate the actual arrival time difference vector t according to step C P , The spatial grid optimization algorithm is used to calculate the theoretical arrival time difference vector of each grid point using the following formula: The actual arrival time difference vector t P The shortest distance to the grid node N k The coordinates are the location coordinates of the partial discharge signal.

9. The quaternary antenna array-based partial discharge monitoring and positioning system for commutation transformers according to claim 1 is characterized in that: A two-dimensional coordinate system is established with the lower left corner of the converter as the origin of the coordinate system. The specific position coordinates of the four broadband antennas are set to broadband antenna 1 (0m, Wm), broadband antenna 2 (0m, 0m), broadband antenna 3 (Cm, 0m), and broadband antenna 4 (Cm, W+2m). Wherein, W is the width of the converter transformer, and C is the length of the converter transformer.

Images