Method and device for recognizing metal foreign matter in GIS based on photon-phase technology

Abstract

The application relates to a GIS metal foreign matter identification method and device based on a photon-phase technology, the method utilizes a photon counting detection system, the photon counting detection system comprises photon counting platforms, partial discharge sensors, partial discharge-free alternating current excitation sources and data acquisition devices connected with the photon counting platforms and the partial discharge sensors in sequence, and data processing devices connected with the data acquisition devices, the method comprises the following steps: acquiring photon pulse data of a current metal foreign matter by using the photon counting detection system; performing phase analysis on the photon pulse data of the current metal foreign matter to obtain a photon counting phase analysis spectrum diagram of the current metal foreign matter; extracting characteristic parameters of the current metal foreign matter based on the photon counting phase analysis spectrum diagram, inputting the characteristic parameters of the current metal foreign matter into a pre-trained metal foreign matter identification model, and obtaining the type of the current metal foreign matter. Thus, the problems that the related art is greatly influenced by the environment are solved, and accurate identification of unknown metal foreign matter is realized.

Description

Technical Field

[0001] This application relates to the field of high-voltage AC transmission equipment operation and maintenance technology, and in particular to a method and device for identifying metallic foreign objects in GIS (Gas Insulated Switchgear) based on photon-phase technology. Background Technology

[0002] Gas-insulated switchgear (GIS) has been widely used in power systems due to its advantages such as small footprint, environmental friendliness, and high safety, and is a key piece of equipment in modern power transmission and distribution networks. However, during the production, transportation, installation, and operation of GIS, mechanical vibration, collisions, or thermal expansion and contraction friction inevitably generate metallic foreign objects. These free metallic foreign objects, once charged, move freely within the cavity under the influence of electric field forces and gravity. In particular, when metallic foreign objects adhere to the surface of the basin insulator, they can easily induce surface flashover of the insulator, thus seriously threatening the stable operation of the GIS.

[0003] Existing methods for detecting metallic foreign objects, such as pulsed current methods, ultra-high frequency methods, and ultrasonic methods, mostly detect them by detecting the current, electromagnetic waves, or ultrasonic waves generated by partial discharge induced by the metallic foreign object. However, these methods are significantly affected by environmental noise and are difficult to detect stationary metallic foreign objects attached to the surface of an insulating basin. Furthermore, optical detection methods, as novel methods for detecting and analyzing optical signals induced by defects, have advantages such as strong resistance to electromagnetic interference, strong insulation, and high sensitivity; however, research on their application in detecting metallic foreign objects remains limited. In addition, existing research largely focuses on defect characteristic analysis and rarely addresses condition assessment.

[0004] Therefore, there is an urgent need to study an optical method for detecting metal foreign objects and to combine it with artificial intelligence to establish a diagnostic and identification model, so as to provide new ideas for future research on practical metal foreign object diagnostic technology. Summary of the Invention

[0005] This application provides a GIS metal foreign object identification method and device based on photon-phase technology to solve the problem that related technologies are greatly affected by the environment, and realizes the accurate identification of unknown metal foreign objects.

[0006] The first aspect of this application provides a GIS-based method for identifying metallic foreign objects based on photon-phase technology. The method utilizes a photon counting detection system, which includes a photon counting platform, a partial discharge sensor, a partial discharge-free AC excitation source, a data acquisition unit connected to the photon counting platform and the partial discharge sensor, and a data processing unit connected to the data acquisition unit, all connected in parallel. The method includes:

[0007] The photon counting and detection system is used to acquire the photon pulse data of the current metallic foreign object;

[0008] Phase analysis is performed on the photon pulse data of the current metallic foreign object to obtain the photon count phase analysis spectrum of the current metallic foreign object;

[0009] Based on the photon count phase analysis spectrum of the current metallic foreign object, the feature parameters of the current metallic foreign object are extracted, and the feature parameters of the current metallic foreign object are input into a pre-trained metallic foreign object recognition model to obtain the type of the current metallic foreign object.

[0010] According to one embodiment of this application, before inputting the feature parameters into the pre-trained metal foreign object recognition model, the method further includes:

[0011] The photon counting detection system was used to acquire photon pulse data of multiple metallic foreign object samples under multiple voltages;

[0012] Phase analysis is performed on the photon pulse data of the multiple metal foreign matter samples under multiple voltages to obtain the photon count phase analysis spectrum of each metal foreign matter sample. Based on the photon count phase analysis spectrum of each metal foreign matter sample, the feature parameters of each metal foreign matter sample are obtained. The feature parameters of each metal foreign matter sample are fitted to obtain a feature parameter database.

[0013] The Extreme Learning Machine (ELM) network is optimized based on a pre-defined improved particle swarm optimization algorithm. The optimized ELM network is then trained using the feature parameter database to obtain the pre-trained metal foreign object recognition model.

[0014] According to one embodiment of this application, training the optimized Extreme Learning Machine network using the feature parameter database to obtain the pre-trained metal foreign object recognition model includes:

[0015] The feature parameter data in the feature parameter database is input into the optimized extreme learning machine network for iterative training. When the optimized extreme learning machine network has been iterated for a preset number of times, the global optimal solution is determined, and the pre-trained metal foreign object recognition model is constructed based on the global optimal solution.

[0016] According to one embodiment of this application, the step of performing phase analysis on the photon pulse data of the current metallic foreign object to obtain the photon count phase analysis spectrum of the current metallic foreign object includes:

[0017] The phase of the power frequency cycle is divided into multiple phase windows on average;

[0018] Calculate the number of photon pulses within each phase window and sum the number of photon pulses within each phase window to obtain the photon number-phase sequence;

[0019] The photon number-phase sequence is visualized to obtain the photon count phase resolution spectrum.

[0020] According to one embodiment of this application, the characteristic parameters include at least one of the following: total number of photons, mean phase asymmetry, skewness, steepness, and Pearson correlation coefficient.

[0021] According to the GIS metal foreign object identification method based on photon-phase technology according to embodiments of this application, a photon counting detection system is used to acquire photon pulse data of the current metal foreign object; phase analysis is performed on the photon pulse data of the current metal foreign object to obtain the photon count phase analysis spectrum of the current metal foreign object; feature parameters of the current metal foreign object are extracted based on the photon count phase analysis spectrum, and the feature parameters of the current metal foreign object are input into a pre-trained metal foreign object identification model to obtain the type of the current metal foreign object. This solves the problem that related technologies are greatly affected by the environment, and achieves accurate identification of unknown metal foreign objects.

[0022] A second aspect of this application provides a GIS metal foreign object identification device based on photon-phase technology. The device utilizes a photon counting detection system, which includes a photon counting platform, a partial discharge sensor, a partial discharge-free AC excitation source connected in parallel, a data acquisition unit connected to the photon counting platform and the partial discharge sensor, and a data processing unit connected to the data acquisition unit. The device further includes:

[0023] The acquisition module is used to acquire the photon pulse data of the current metallic foreign object using the photon counting and detection system;

[0024] The phase analysis module is used to perform phase analysis on the photon pulse data of the current metallic foreign object to obtain the photon count phase analysis spectrum of the current metallic foreign object;

[0025] The identification module is used to extract the feature parameters of the current metal foreign object based on the photon counting phase analysis spectrum of the current metal foreign object, and input the feature parameters of the current metal foreign object into the pre-trained metal foreign object identification model to obtain the type of the current metal foreign object.

[0026] According to one embodiment of this application, before inputting the feature parameters into the pre-trained metallic foreign object recognition model, the recognition module is further configured to:

[0027] The photon counting detection system was used to acquire photon pulse data of multiple metallic foreign object samples under multiple voltages;

[0028] Phase analysis is performed on the photon pulse data of the multiple metal foreign matter samples under multiple voltages to obtain the photon count phase analysis spectrum of each metal foreign matter sample. Based on the photon count phase analysis spectrum of each metal foreign matter sample, the feature parameters of each metal foreign matter sample are obtained. The feature parameters of each metal foreign matter sample are fitted to obtain a feature parameter database.

[0029] The Extreme Learning Machine (ELM) network is optimized based on a pre-defined improved particle swarm optimization algorithm. The optimized ELM network is then trained using the feature parameter database to obtain the pre-trained metal foreign object recognition model.

[0030] According to one embodiment of this application, the identification module is used for:

[0031] The feature parameter data in the feature parameter database is input into the optimized extreme learning machine network for iterative training. When the optimized extreme learning machine network has been iterated for a preset number of times, the global optimal solution is determined, and the pre-trained metal foreign object recognition model is constructed based on the global optimal solution.

[0032] According to one embodiment of this application, the phase analysis module is used for:

[0033] The phase of the power frequency cycle is divided into multiple phase windows on average;

[0034] Calculate the number of photon pulses within each phase window and sum the number of photon pulses within each phase window to obtain the photon number-phase sequence;

[0035] The photon number-phase sequence is visualized to obtain the photon count phase resolution spectrum.

[0036] According to one embodiment of this application, the characteristic parameters include at least one of the following: total number of photons, mean phase asymmetry, skewness, steepness, and Pearson correlation coefficient.

[0037] According to the embodiments of this application, the GIS metal foreign object identification device based on photon-phase technology acquires photon pulse data of the current metal foreign object using a photon counting detection system; performs phase analysis on the photon pulse data of the current metal foreign object to obtain a photon count phase analysis spectrum of the current metal foreign object; extracts feature parameters of the current metal foreign object based on the photon count phase analysis spectrum, and inputs the feature parameters of the current metal foreign object into a pre-trained metal foreign object identification model to obtain the type of the current metal foreign object. This solves the problem that related technologies are greatly affected by the environment, and achieves accurate identification of unknown metal foreign objects.

[0038] A third aspect of this application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the program to implement the GIS metal foreign object identification method based on photon-phase technology as described in the above embodiments.

[0039] A fourth aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement the GIS metal foreign object identification method based on photon-phase technology as described in the above embodiments.

[0040] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0041] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0042] Figure 1 This is a schematic diagram of the structure of a photon counting and detection system according to an embodiment of this application;

[0043] Figure 2 This is a flowchart of a GIS metal foreign object identification method based on photon-phase technology according to an embodiment of this application;

[0044] Figure 3 This is a schematic diagram of the structure of a GIS linear particle defect model according to an embodiment of this application;

[0045] Figure 4 This is a schematic diagram of the structure of a traditional ELM network;

[0046] Figure 5 This is a schematic diagram illustrating the construction process of a metal foreign object recognition model based on an improved PSO (Particle Swarm Optimization)-ELM (Extreme Learning Machine) according to an embodiment of this application.

[0047] Figure 6 This is a schematic diagram of the PRPC pattern of aluminum wires of different lengths at 10kV according to an embodiment of this application;

[0048] Figure 7 This is a schematic diagram of the feature parameter fitting curve according to an embodiment of this application;

[0049] Figure 8 This is a schematic diagram of the classification confusion matrix of an improved PSO-ELM model according to an embodiment of this application;

[0050] Figure 9 This is a block diagram of a GIS metal foreign object identification device based on photon-phase technology according to an embodiment of this application;

[0051] Figure 10 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0052] The embodiments of this application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0053] The following describes a GIS metal foreign object identification method and apparatus based on photon-phase technology according to embodiments of this application, with reference to the accompanying drawings.

[0054] Before introducing the GIS metal foreign object identification method based on photon-phase technology in the embodiments of this application, we will first introduce the photon counting detection system used in the GIS metal foreign object identification method based on photon-phase technology in this application.

[0055] Specifically, the photon counting detection system of this application includes a photon counting platform, a partial discharge sensor, a partial discharge-free AC excitation source, a data acquisition unit connected to the photon counting platform and the partial discharge sensor, and a data processing unit connected to the data acquisition unit, all connected in parallel.

[0056] The photon counting platform includes a photomultiplier tube (PMT), a GIS model, and an airtight device (test chamber). The data acquisition component can be a data acquisition card, and the data processing component can be a computer.

[0057] For example, such as Figure 1As shown, the photon counting detection system can consist of a personal computer, a partial discharge-free AC excitation source, a PMT, a GIS model, an airtight device (test chamber), a data acquisition card, a partial discharge sensor, and an anechoic chamber. In this embodiment, a partial discharge-free AC voltage source is used as the test power source to pressurize a defect sample placed in a sealed chamber. To simulate the normal operating environment of the GIS, considering the influence of air pressure on photon release and the pressure resistance of the test chamber, the chamber is filled with 0.3 MPa SF6 gas. To obtain reliable experimental data, an MPD 800 partial discharge sensor is used to detect the partial discharge initiation voltage and monitor the occurrence of PD (Partial Discharge) during the pressurization process. During the experiment, the PMT measures photon pulse data for a certain metallic foreign object, a certain voltage, and a certain time length.

[0058] The following describes the GIS metal foreign object identification method based on photon-phase technology that uses the above-mentioned photon counting detection system proposed in this application.

[0059] Specifically, Figure 2 This is a flowchart illustrating a GIS metal foreign object identification method based on photon-phase technology, provided in an embodiment of this application.

[0060] like Figure 2 As shown, the GIS metal foreign object identification method based on photon-phase technology includes the following steps:

[0061] In step S201, the photon pulse data of the current metallic foreign object is acquired using a photon counting detection system.

[0062] Optionally, embodiments of this application can use aluminum wires of varying degrees as the current metallic foreign object, forming defect samples such as... Figure 3 As shown, the aluminum wire used has a cross-sectional diameter D of 0.4 mm and a length H of 2.0 mm, 4.0 mm, 6.0 mm, 8.0 mm, or 10.0 mm.

[0063] Specifically, in this embodiment of the application, a defect sample containing the current metallic foreign object can be placed on the photon counting platform of the photon counting detection system, the defect sample can be pressurized by the photon counting detection system, and the photon pulse data of the current metallic foreign object can be obtained by the PMT in the photon counting platform.

[0064] In step S202, phase analysis is performed on the photon pulse data of the current metallic foreign object to obtain the photon count phase analysis spectrum of the current metallic foreign object.

[0065] Furthermore, in some embodiments, phase analysis is performed on the photon pulse data of the current metallic foreign object to obtain a photon count phase analysis spectrum of the current metallic foreign object, including: dividing the phase of the power frequency cycle into multiple phase windows on average; calculating the number of photon pulses in each phase window and accumulating the number of photon pulses in each phase window to obtain a photon count-phase sequence; and visualizing the photon count-phase sequence to obtain a photon count phase analysis spectrum.

[0066] Specifically, in this embodiment, the phase of one power frequency cycle (360°) can be divided into N phase windows. The more phase windows there are, the higher the phase-time accuracy and the more accurate the calculated statistical characteristic parameters. However, due to the limited data storage and processing capabilities of the hardware system, the number of phase windows cannot be too large. In this embodiment, N can be 360, meaning each 1° is a phase window. Since a single photon pulse signal is a fixed square wave, photon counting can be performed. This involves accumulating the photon pulse data acquired using a photon counting detection system across the corresponding phase windows to obtain a photon count-phase sequence. This photon count-phase sequence can then be visualized to obtain a Phase Resolved Photon Counting (PRPC) spectrum.

[0067] In step S203, the feature parameters of the current metal foreign object are extracted based on the photon counting phase analysis spectrum of the current metal foreign object, and the feature parameters of the current metal foreign object are input into the pre-trained metal foreign object recognition model to obtain the type of the current metal foreign object.

[0068] Understandably, to establish the mapping relationship between photon pulse data and defect parameters, it is necessary to quantify this variation pattern. Considering the differences in phase distribution, photon count, and positive / negative half-cycle differences in PRPC spectra, this application embodiment extracts seven statistical parameters of the photon count-phase sequence, as follows:

[0069] 1) The total number of photons, Q, is the sum of the photon number sequence, which reflects the overall characteristics of the PRPC spectrum.

[0070] 2) The mean phase asymmetry ψ reflects the difference in the peak phases of the positive and negative half-cycles. The calculation formula is:

[0071] ψ=μ + / μ - (1)

[0072] Where, μ + μ - These represent the mean phase of the PRPC spectrum during the positive and negative half-cycles, respectively. The general calculation formula is:

[0073]

[0074] Where ω is the number of phase windows; Let p be the phase of the i-th phase window, and let p be the width of the phase window. i Let be the number of photons in the i-th phase window of the PRPC spectrum.

[0075] 3) Skewness S k Skewness S k This is used to describe the skewness direction and degree of the PRPC map distribution shape. A skewness of 0 indicates a symmetrical distribution; a skewness greater than or less than 0 corresponds to a leftward or rightward skewness, respectively, and the absolute value of the skewness is positively correlated with the degree of skewness. The specific calculation formula is as follows:

[0076]

[0077] Where σ is the standard deviation, calculated using the formula:

[0078]

[0079] 4) Steepness K u Kurtosis reflects the degree of prominence in the shape of the PRPC map. When the kurtosis is 0, the PRPC map outline conforms to the normal distribution curve; when the kurtosis is greater than 0, it indicates that the PRPC map shape distribution is sharper and steeper than the normal distribution; and when the kurtosis is less than 0, it indicates that the PRPC map shape is flatter than the normal distribution. The kurtosis K is... u The definition is:

[0080]

[0081] 5) Pearson correlation coefficient ρ. The Pearson correlation coefficient measures the similarity between the positive and negative half-circles of the PRPC map. The Pearson correlation coefficient value ranges between 0 and 1. The closer the Pearson correlation coefficient is to 0, the greater the difference between the positive and negative half-circles of the PRPC map; the closer the Pearson correlation coefficient is to 1, the more similar the positive and negative half-circles of the PRPC map. The Pearson correlation coefficient ρ is defined as follows:

[0082]

[0083] Furthermore, the feature parameters of the current metal foreign object extracted above are input into the pre-trained metal foreign object recognition model. The pre-trained metal foreign object recognition model performs calculations and inferences based on the input feature parameters and outputs the type of the current metal foreign object.

[0084] Furthermore, in some embodiments, before inputting the feature parameters into the pre-trained metal foreign object recognition model, the method further includes: acquiring photon pulse data of multiple metal foreign object samples under multiple voltages using a photon counting detection system; performing phase analysis on the photon pulse data of multiple metal foreign object samples under multiple voltages to obtain a photon count phase analysis spectrum for each metal foreign object sample, and obtaining feature parameters for each metal foreign object sample based on the photon count phase analysis spectrum for each metal foreign object sample, and fitting the feature parameters of each metal foreign object sample to obtain a feature parameter database; optimizing the Extreme Learning Machine network based on a preset improved particle swarm optimization algorithm, and training the optimized Extreme Learning Machine network using the feature parameter database to obtain a pre-trained metal foreign object recognition model.

[0085] Among them, multiple metal foreign object samples can be aluminum wires of different degrees. For example, multiple metal foreign object samples can be aluminum wires with lengths of 2.0mm, 4.0mm, 6.0mm, 8.0mm and 10.0mm respectively.

[0086] Specifically, in this embodiment of the application, before inputting the feature parameters into the pre-trained metal foreign object recognition model, it is necessary to construct the metal foreign object recognition model. First, similar to the step of obtaining the photon pulse data of the current metal foreign object using the photon counting detection system described above, multiple metal foreign object samples are tested at different voltages using the photon counting detection system, thereby obtaining the photon pulse data of multiple metal foreign object samples at multiple voltages.

[0087] Furthermore, similar to the steps described above of performing phase analysis on the photon pulse data of the current metallic foreign object to obtain the photon count phase analysis spectrum of the current metallic foreign object, phase analysis is performed on the photon pulse data of multiple metallic foreign object samples under multiple voltages to obtain the photon count phase analysis spectrum of each metallic foreign object sample. Based on the photon count phase analysis spectrum of each metallic foreign object sample, the characteristic parameters of each metallic foreign object sample are obtained, namely, the total number of photons, mean phase asymmetry, skewness, steepness, and Pearson correlation coefficient.

[0088] Understandably, effective metal foreign object diagnosis requires training on a massive database. If characteristic parameter data for different types and sizes of metal foreign objects are obtained solely through experimentation, the workload associated with database development would be enormous. Therefore, for quantities with functional relationships, such as the relationship between metal foreign object size and PRPC characteristic parameters, a curve fitting can be performed on the relationship between the metal foreign object size and PRPC characteristic parameters after obtaining a finite number of data points experimentally. This yields the corresponding functional relationship, theoretically generating an infinitely large characteristic parameter database.

[0089] Furthermore, embodiments of this application can classify metallic foreign objects based on the obtained photon information, for example, classifying the size of metallic foreign objects based on the total number of photons. Furthermore, the intelligent algorithm used in embodiments of this application is an improved PSO-ELM, while traditional ELM networks, such as... Figure 4 As shown, the ELM model consists of an input layer, hidden layers, and an output layer. A unique optimal solution can be obtained simply by setting the number of hidden layer nodes *l* and the activation function *g(x)*. However, manually selecting the parameters *l* and the function *g(x)* often fails to guarantee optimal performance. Therefore, this application introduces an improved Particle Swarm Optimization (PSO) algorithm to optimize the ELM parameters. In PSO, particles represent possible solutions to the optimization problem, and their velocity is used as the metric. Location and fitness Let represent the state of the i-th particle in the solution space after the k-th iteration, and update it for the k+1th iteration using equations (8) and (9) until the stopping condition is met and the global optimal solution is reached.

[0090]

[0091] in, Let ω be the velocity of the i-th particle in the solution space after the (k+1)-th iteration, ω be the inertia weight, c1 and c2 be non-negative acceleration factors, and r1 and r2 be random numbers between 0 and 1. This represents the globally optimal value of particle fitness.

[0092] Furthermore, to balance the algorithm's convergence ability and optimization efficiency, while avoiding getting trapped in local optima during iteration, this application's embodiments make two improvements to the conventional PSO algorithm:

[0093] 1) The fixed inertia weight ω is changed to a linearly decreasing value. That is:

[0094] w k =w start -(w start -w end )*k / k max (5)

[0095] Among them, w k Let w be the inertia weight after the k-th iteration. start w end Let k be the initial inertia weight and the final inertia weight during the iteration process, respectively. max This represents the maximum number of iterations. Therefore, a larger inertia weight in the early stages maintains the algorithm's global search capability, while a smaller inertia weight in the later stages helps improve optimization accuracy.

[0096] 2) Introduce particle mutation. During each iteration, the particle state is reinitialized with a certain probability, expanding the spatial range of the particle swarm and thus increasing the probability of finding a better value. Specifically, for particle i after the k-th iteration, a random number q between [0,1] is generated. If q is not greater than the mutation probability p, then:

[0097]

[0098] in, X is the inertia weight value. random The random positions in the solution space are represented by the mutation operation. Therefore, the shrinking particle population can be expanded through mutation, which helps the algorithm achieve global optimum.

[0099] Furthermore, after finding the optimal solution through the improved particle swarm optimization algorithm, the number of hidden layer nodes and activation function of the ELM network are optimized, and the optimized ELM network is trained through a feature parameter database to construct a pre-trained metal foreign object recognition model.

[0100] Furthermore, in some embodiments, the optimized Extreme Learning Machine Network is trained using a feature parameter database to obtain a pre-trained metal foreign object recognition model. This includes: inputting feature parameter data from the feature parameter database into the optimized Extreme Learning Machine Network for iterative training, and determining the global optimal solution when the optimized Extreme Learning Machine Network has undergone a preset number of iterations, so as to construct the pre-trained metal foreign object recognition model based on the global optimal solution.

[0101] Specifically, in this embodiment, the parameters of the Extreme Learning Machine network are optimized based on the constructed feature parameter database and the preset improved particle swarm optimization algorithm. After a preset number of iterations, a solution close to the global optimum is obtained, namely the optimal number of hidden layer nodes and the optimal activation function. Based on the optimized parameters, a pre-trained metal foreign object recognition model is constructed.

[0102] The construction process of the metal foreign object recognition model based on the improved PSO-ELM is as follows: Figure 5 As shown, firstly, the defect sample is measured by photon counting detection system, and PRPC spectrum is plotted based on photon pulse data. Feature parameters are extracted based on PRPC spectrum, and the feature parameters are fitted to construct a database. Then, an effective metal foreign object identification model is constructed using an improved PSO algorithm-optimized ELM network, thereby classifying and identifying metal foreign objects in GIS.

[0103] Therefore, to achieve high-sensitivity detection and effective identification of metallic foreign objects in GIS, this application first constructs a photon counting detection system. Then, photon signals from different metallic foreign objects (e.g., scaled-down basin insulators with aluminum wires of varying lengths attached to their surfaces) are acquired under different voltages. Based on Phase-resolved Photon Counting (PRPC) spectra, feature parameters such as total photon count, skewness, steepness, mean, and phase asymmetry are extracted. A large dataset of feature parameters is obtained by fitting curves to sample points. Finally, an identification model is constructed based on an improved particle swarm optimization extreme learning machine algorithm and the training dataset to intelligently identify metallic foreign objects. This demonstrates that the PRPC-based feature parameters effectively reflect the defect-induced photon response characteristics, and the proposed identification model achieves a detection accuracy of over 90.00% for linear metallic foreign objects.

[0104] To enable those skilled in the art to more clearly and intuitively understand the technical effects of the GIS metal foreign object identification method based on photon-phase technology proposed in this application, a detailed description is provided below with reference to specific embodiments.

[0105] Specifically, the embodiments of this application are first based on Figure 3 The photon test data of aluminum wire samples of different lengths at 10kV are shown below, and the resulting PRPC spectra are as follows. Figure 6 As shown, where, Figure 6 (a) is the PRPC spectrum of an aluminum wire sample with a length of 0 mm at 10 kV. Figure 6 (b) is the PRPC spectrum of an aluminum wire sample with a length of 2.0 mm at 10 kV. Figure 6 (c) is the PRPC spectrum of an aluminum wire sample with a length of 4.0 mm at 10 kV. Figure 6 (d) shows the PRPC spectrum of an aluminum wire sample with a length of 6.0 mm at 10 kV. Figure 6 (e) shows the PRPC spectrum of an 8.0 mm long aluminum wire sample at 10 kV. Figure 6 (f) shows the PRPC spectrum of an aluminum wire sample with a length of 10.0 mm at 10 kV.

[0106] Depend on Figure 6 It can be seen that the PRPC spectrum has a polarity effect similar to that of the PRPD spectrum. Based on the in-depth mining of the PRPC spectrum, characteristic parameters for characterizing the defect-induced photon response can be extracted.

[0107] Furthermore, feature parameters are extracted from the obtained PRPC map, namely the total number of photons Q and the skewness S. k Steepness K u The mean phase asymmetry ψ and Pearson correlation coefficient ρ were calculated, and a smooth spline fit was performed, such as... Figure 6 As shown, a database of characteristic parameters related to the length of aluminum wire was thus constructed.

[0108] Furthermore, the total number of photons can indirectly reflect the severity of the damage to insulation caused by aluminum wire particles. In order to facilitate defect classification, this application embodiment roughly divides the severity of aluminum wire particle defects into 4 levels based on the photon counting results. The correspondence between these levels and the length of the aluminum wire is shown in Table 1.

[0109] Table 1

[0110] level Length range Label 1 L < 2.0 mm S1 2 2.0mm≤L<4.0mm S2 3 4.0mm≤L<8.0mm S3 4 L≥8.0mm S4

[0111] Furthermore, by utilizing the improved PSO algorithm and Figure 7 A feature parameter database was constructed, and the ELM network parameters were optimized. After 100 iterations, a near-global optimum was obtained, with an optimal number of hidden layer nodes of 72 and an optimal activation function of the sine function. The ELM classification model built based on the optimized parameters yielded the following classification results on the test set: Figure 8 As shown, the results indicate that the characteristic parameters based on PRPC can effectively reflect the defect-induced photon response characteristics, and the proposed identification model achieves a detection accuracy of 95% for linear metallic foreign objects.

[0112] Those skilled in the art will understand that when a metallic foreign object is present in an insulation system within a GIS (Gas Insulation System), since the metallic foreign object is a conductor, its presence will inevitably cause electric field distortion and induce ionization and luminescence of the insulating medium. This application's embodiments detect the optical signal of the metallic foreign object and, combined with photon-phase technology, plot a photon-counted phase resolution spectrum. By extracting its characteristic parameters reflecting the metallic foreign object's features, and training it using an optimized Extreme Learning Machine network based on an improved particle swarm optimization algorithm and the collected feature dataset, a metallic foreign object diagnostic model is obtained to identify unknown metallic foreign objects.

[0113] According to the GIS metal foreign object identification method based on photon-phase technology according to embodiments of this application, a photon counting detection system is used to acquire photon pulse data of the current metal foreign object; phase analysis is performed on the photon pulse data of the current metal foreign object to obtain the photon count phase analysis spectrum of the current metal foreign object; feature parameters of the current metal foreign object are extracted based on the photon count phase analysis spectrum, and the feature parameters of the current metal foreign object are input into a pre-trained metal foreign object identification model to obtain the type of the current metal foreign object. This solves the problem that related technologies are greatly affected by the environment, and achieves accurate identification of unknown metal foreign objects.

[0114] Next, referring to the accompanying drawings, a GIS metal foreign object identification device based on photon-phase technology according to an embodiment of this application is described.

[0115] Figure 9 This is a block diagram of a GIS metal foreign object identification device based on photon-phase technology according to an embodiment of this application.

[0116] In this embodiment, the device utilizes a photon counting and detection system, which includes a photon counting platform, a partial discharge sensor, a partial discharge-free AC excitation source connected in parallel, a data acquisition unit connected to the photon counting platform and the partial discharge sensor, and a data processing unit connected to the data acquisition unit.

[0117] like Figure 9 As shown, the GIS metal foreign object identification device 10 based on photon-phase technology includes: an acquisition module 100, a phase analysis module 200, and an identification module 300.

[0118] The acquisition module 100 is used to acquire photon pulse data of the current metallic foreign object using a photon counting detection system; the phase analysis module 200 is used to perform phase analysis on the photon pulse data of the current metallic foreign object to obtain the photon count phase analysis spectrum of the current metallic foreign object; and the identification module 300 is used to extract the feature parameters of the current metallic foreign object based on the photon count phase analysis spectrum of the current metallic foreign object, and input the feature parameters of the current metallic foreign object into a pre-trained metallic foreign object identification model to obtain the type of the current metallic foreign object.

[0119] Furthermore, in some embodiments, before inputting the feature parameters into the pre-trained metal foreign object recognition model, the recognition module 300 is further configured to: acquire photon pulse data of multiple metal foreign object samples under multiple voltages using a photon counting detection system; perform phase analysis on the photon pulse data of multiple metal foreign object samples under multiple voltages respectively to obtain a photon count phase analysis spectrum of each metal foreign object sample, and obtain the feature parameters of each metal foreign object sample based on the photon count phase analysis spectrum of each metal foreign object sample, and fit the feature parameters of each metal foreign object sample to obtain a feature parameter database; optimize the extreme learning machine network based on a preset improved particle swarm optimization algorithm, and train the optimized extreme learning machine network using the feature parameter database to obtain a pre-trained metal foreign object recognition model.

[0120] Furthermore, in some embodiments, the identification module 300 is used to: input feature parameter data from the feature parameter database into the optimized extreme learning machine network for iterative training, and determine the global optimal solution when the number of iterations of the optimized extreme learning machine network reaches a preset number of iterations, so as to construct a pre-trained metal foreign object identification model based on the global optimal solution.

[0121] Furthermore, in some embodiments, the phase analysis module 200 is used to: divide the phase of the power frequency cycle into multiple phase windows on average; calculate the number of photon pulses in each phase window and accumulate the number of photon pulses in each phase window to obtain a photon count-phase sequence; and visualize the photon count-phase sequence to obtain a photon count phase analysis spectrum.

[0122] Furthermore, in some embodiments, the characteristic parameters include at least one of the following: total number of photons, mean phase asymmetry, skewness, steepness, and Pearson correlation coefficient.

[0123] It should be noted that the foregoing explanation of the GIS metal foreign object identification method based on photon-phase technology also applies to the GIS metal foreign object identification device based on photon-phase technology in this embodiment, and will not be repeated here.

[0124] According to the embodiments of this application, the GIS metal foreign object identification device based on photon-phase technology acquires photon pulse data of the current metal foreign object using a photon counting detection system; performs phase analysis on the photon pulse data of the current metal foreign object to obtain a photon count phase analysis spectrum of the current metal foreign object; extracts feature parameters of the current metal foreign object based on the photon count phase analysis spectrum, and inputs the feature parameters of the current metal foreign object into a pre-trained metal foreign object identification model to obtain the type of the current metal foreign object. This solves the problem that related technologies are greatly affected by the environment, and achieves accurate identification of unknown metal foreign objects.

[0125] Figure 10 A schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include:

[0126] The memory 901, the processor 902, and the computer program stored on the memory 901 and capable of running on the processor 902.

[0127] When the processor 902 executes the program, it implements the GIS metal foreign object identification method based on photon-phase technology provided in the above embodiments.

[0128] Furthermore, electronic devices also include:

[0129] Communication interface 903 is used for communication between memory 901 and processor 902.

[0130] The memory 901 is used to store computer programs that can run on the processor 902.

[0131] The memory 901 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0132] If the memory 901, processor 902, and communication interface 903 are implemented independently, then the communication interface 903, memory 901, and processor 902 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 9 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0133] Optionally, in a specific implementation, if the memory 901, processor 902, and communication interface 903 are integrated on a single chip, then the memory 901, processor 902, and communication interface 903 can communicate with each other through an internal interface.

[0134] The processor 902 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0135] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described GIS metal foreign object identification method based on photon-phase technology.

[0136] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0137] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0138] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A GIS-based method for identifying metallic foreign objects using photon-phase technology, characterized in that, The method utilizes a photon counting detection system, which includes a photon counting platform, a partial discharge sensor, a partial discharge-free AC excitation source connected in parallel, a data acquisition unit connected to the photon counting platform and the partial discharge sensor, and a data processing unit connected to the data acquisition unit. The method includes: The photon counting and detection system is used to acquire the photon pulse data of the current metallic foreign object; Phase analysis is performed on the photon pulse data of the current metallic foreign object to obtain the photon count phase analysis spectrum of the current metallic foreign object; Based on the photon counting phase analysis spectrum of the current metallic foreign object, the feature parameters of the current metallic foreign object are extracted, and the feature parameters of the current metallic foreign object are input into a pre-trained metallic foreign object recognition model to obtain the type of the current metallic foreign object; The method further includes, before inputting the feature parameters into the pre-trained metal foreign object recognition model: acquiring photon pulse data of multiple metal foreign object samples under multiple voltages using the photon counting detection system; performing phase analysis on the photon pulse data of the multiple metal foreign object samples under multiple voltages to obtain a photon count phase analysis spectrum for each metal foreign object sample; obtaining the feature parameters of each metal foreign object sample based on the photon count phase analysis spectrum; fitting the feature parameters of each metal foreign object sample to obtain a feature parameter database; optimizing the Extreme Learning Machine network based on a preset improved particle swarm optimization algorithm; training the optimized Extreme Learning Machine network using the feature parameter database to obtain the pre-trained metal foreign object recognition model. The preset improved particle swarm optimization algorithm will use fixed inertial weights. Change to linear decreasing, that is: ； in, The inertia weights after the k-th iteration are... These are the initial and final inertia weights during the iteration process, respectively. This represents the maximum number of iterations. The preset improved particle swarm optimization algorithm introduces particle mutation, for the first... k Particles after the next iteration i Generate random numbers between [0,1] q ,like q Not greater than the probability of mutation p Then we have: ； in, This is the inertia weight value. For random positions in the solution space.

2. The method according to claim 1, characterized in that, The step of training the optimized Extreme Learning Machine network using the feature parameter database to obtain the pre-trained metal foreign object recognition model includes: The feature parameter data in the feature parameter database is input into the optimized extreme learning machine network for iterative training. When the optimized extreme learning machine network has been iterated for a preset number of times, the global optimal solution is determined, and the pre-trained metal foreign object recognition model is constructed based on the global optimal solution.

3. The method according to claim 1, characterized in that, The step of performing phase analysis on the photon pulse data of the current metallic foreign object to obtain the photon count phase analysis spectrum of the current metallic foreign object includes: The phase of the power frequency cycle is divided into multiple phase windows on average; Calculate the number of photon pulses within each phase window and sum the number of photon pulses within each phase window to obtain the photon number-phase sequence; The photon number-phase sequence is visualized to obtain the photon count phase resolution spectrum.

4. The method according to any one of claims 1-3, characterized in that, The characteristic parameters include at least one of the following: total number of photons, mean phase asymmetry, skewness, steepness, and Pearson correlation coefficient.

5. A GIS metal foreign object identification device based on photon-phase technology, characterized in that, The device utilizes a photon counting and detection system, which includes a photon counting platform, a partial discharge sensor, a partial discharge-free AC excitation source connected in parallel, a data acquisition unit connected to the photon counting platform and the partial discharge sensor, and a data processing unit connected to the data acquisition unit. The device further includes: The acquisition module is used to acquire the photon pulse data of the current metallic foreign object using the photon counting and detection system; The phase analysis module is used to perform phase analysis on the photon pulse data of the current metallic foreign object to obtain the photon count phase analysis spectrum of the current metallic foreign object; The identification module is used to extract the feature parameters of the current metal foreign object based on the photon counting phase analysis spectrum of the current metal foreign object, and input the feature parameters of the current metal foreign object into the pre-trained metal foreign object identification model to obtain the type of the current metal foreign object; Before inputting the feature parameters into the pre-trained metal foreign object recognition model, the recognition module is further configured to: acquire photon pulse data of multiple metal foreign object samples under multiple voltages using the photon counting detection system; perform phase analysis on the photon pulse data of the multiple metal foreign object samples under multiple voltages to obtain a photon count phase analysis spectrum of each metal foreign object sample, and obtain the feature parameters of each metal foreign object sample based on the photon count phase analysis spectrum of each metal foreign object sample, and fit the feature parameters of each metal foreign object sample to obtain a feature parameter database; optimize the extreme learning machine network based on a preset improved particle swarm optimization algorithm, and train the optimized extreme learning machine network using the feature parameter database to obtain the pre-trained metal foreign object recognition model; The preset improved particle swarm optimization algorithm will use fixed inertial weights. Change to linear decreasing, that is: ； in, The inertia weights after the k-th iteration are... These are the initial and final inertia weights during the iteration process, respectively. This represents the maximum number of iterations. The preset improved particle swarm optimization algorithm introduces particle mutation, for the first... k Particles after the next iteration i Generate random numbers between [0,1] q ,like q Not greater than the probability of mutation p Then we have: ； in, This is the inertia weight value. For random positions in the solution space.

6. The apparatus according to claim 5, characterized in that, The identification module is used for: The feature parameter data in the feature parameter database is input into the optimized extreme learning machine network for iterative training. When the optimized extreme learning machine network has been iterated for a preset number of times, the global optimal solution is determined, and the pre-trained metal foreign object recognition model is constructed based on the global optimal solution.

7. An electronic device, characterized in that, include: The memory, the processor, and the computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the GIS metal foreign object identification method based on photon-phase technology as described in any one of claims 1-4.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the GIS metal foreign object identification method based on photon-phase technology as described in any one of claims 1-4.

Images