An Optimization Method for Deploying Optical Sensors for Partial Discharge Detection in a 10kV Metal-Enclosed Switchgear

Abstract

This invention belongs to the field of sensor deployment technology inside switchgear, and relates to an optimized deployment method for optical sensors used in partial discharge detection inside a 10kV metal-enclosed switchgear. First, a discharge source characteristic testing platform is established to measure the energy distribution and power of typical partial discharge spectra, calibrating the detection limit of the optical sensing device. Then, the optical signal propagation characteristics and luminous flux characteristics of the switchgear are analyzed using an optical path simulation model, clarifying the reasonable arrangement range of the light-collecting units. Based on the chi-square test, a light-collecting unit layout strategy that accurately reflects the fault location is selected. The optimized sensor arrangement method is verified through a switchgear discharge optical simulation experimental platform, providing a reference for the performance requirements and optimized arrangement of optical sensors for partial discharge in switchgear.

Description

Technical Field

[0001] This invention belongs to the field of sensor deployment technology inside switchgear, and particularly relates to an optimized method for deploying optical sensors for partial discharge detection inside a 10kV metal-enclosed switchgear. Background Technology

[0002] Deterioration or defects in the internal insulation of switchgear can trigger partial discharges or even explosions, threatening the safety of the power grid and personnel. Therefore, effective detection of partial discharges in switchgear is crucial. Optical detection methods for partial discharges offer good anti-interference capabilities and high sensitivity, showing promising application prospects. However, the reflection, absorption, and loss of optical signals due to the internal structure of the switchgear pose challenges to optical detection.

[0003] Currently, in the partial discharge detection of switchgear, due to the limited internal space of the switchgear enclosed in a metal shell, there are certain limitations when the optical sensors are not properly configured. The sensors have low detection efficiency, inaccurate positioning, and are susceptible to environmental interference, which makes it impossible to comprehensively, objectively, and truthfully reflect the fault status of the switchgear. Summary of the Invention

[0004] The purpose of this invention is to provide an optimized deployment method for optical sensors used in partial discharge detection in 10kV metal-enclosed switchgear, which solves the problems of low sensor detection efficiency, inaccurate positioning, and susceptibility to environmental interference in traditional deployment methods.

[0005] This invention is achieved through the following technical solution:

[0006] An optimized method for deploying optical sensors for partial discharge detection in a 10kV metal-enclosed switchgear includes the following steps:

[0007] S1. Conduct actual measurements on the light source characteristics of partial discharge in typical defects and obtain calibration results;

[0008] The calibration results include discharge light power, discharge light intensity range, and the detection limit of the optical sensing device;

[0009] S2. Based on the calibration results obtained in S1, the optical path simulation model is used to analyze the optical signal propagation characteristics and optical flux characteristics of the switch cabinet, and to determine the reasonable layout range of the light-collecting unit.

[0010] S3. Based on the reasonable layout range of the lighting units, a lighting unit layout strategy that can accurately reflect the fault location is selected using the chi-square test.

[0011] Furthermore, in S1, the light source characteristics of partial discharge of typical defects are measured using the constructed experimental platform;

[0012] The experimental platform uses a total internal reflection optical integrating sphere to construct a partial discharge source calibration system.

[0013] The partial discharge source calibration system includes an integrating sphere, a discharge power supply, and a high-voltage lead. The discharge power supply is located inside the integrating sphere, and the high-voltage lead extends into the integrating sphere and connects to the discharge power supply.

[0014] The integrating sphere is externally connected to a fiber optic probe and a photomultiplier tube; the fiber optic probe is connected to a fiber optic spectrometer via fiber optic cable, and the photomultiplier tube is sequentially connected to a power meter, an oscilloscope, and a high-frequency current sensor.

[0015] An insulating sleeve is provided at the top of the integrating sphere, and the high-voltage lead passes through the insulating sleeve;

[0016] The inner wall of the integrating sphere is coated with a total internal reflection MgO coating.

[0017] Furthermore, the light source characteristics of partial discharge in typical defects were measured using the constructed experimental platform. The specific test methods are as follows:

[0018] First, record the background light noise level and the spectral intensity of the background light noise inside the integrating sphere to eliminate the interference of the basic light.

[0019] Then, a reference spectrum is recorded using a standard light source and a fiber optic spectrometer to provide a benchmark for subsequent discharge spectrum determination;

[0020] The discharge range was determined by recording the initial discharge voltage and breakdown voltage using a step-by-step voltage application method.

[0021] Subsequently, starting from a predetermined initial discharge voltage, the voltage was gradually increased to the breakdown voltage, and the discharge light intensity at each voltage level was recorded.

[0022] Finally, the sensor was connected to the fiber optic probe, replacing the fiber optic spectrometer, to determine the range of discharge light intensity of the sensor.

[0023] Furthermore, the discharge optical power is calculated using the standard light source comparison method, and the expression is:

[0024]

[0025] Among them, t d S is the effective integration time of the spectrum. l S represents the discharge light intensity of a standard light source during the integration time. d These represent the discharge light intensity of the discharge source during the integration time.

[0026] Furthermore, in S1, the process for determining the detection lower limit of the optical sensing device is as follows:

[0027] A silicon photomultiplier tube (SMT) is used to determine the lower limit of the received optical power. The optical power received by the SMT sensor is Ps, calculated based on the area of ​​the integrating sphere and the effective aperture of the optical fiber.

[0028]

[0029] In the formula: P d r is the discharge optical power. f r is the fiber aperture. s Let be the inner radius of the integrating sphere.

[0030] Furthermore, in S2, the specific process of building the optical path simulation model is as follows:

[0031] Based on the actual structure, dimensions, and material properties of the switchgear under test, an optical path simulation model based on TracePro was built. The simulation uses a bidirectional reflection distribution function to describe the reflection distribution characteristics of the material, and its calculation formula is as follows:

[0032]

[0033] Wherein, BRDF represents the reflection distribution characteristics; A is the scale constant, representing the total intensity of scattering; B and g are the reflection shape constants, representing the scattering shape; and r is the radius of the receiving cylinder in the reflection direction.

[0034] Furthermore, in S3, the chi-square test algorithm is used to analyze the irradiance discrimination at different phase positions of various discharges. The inner surface is divided into several sensing points, and the area of ​​each sensing point is set to X mm. 2 The irradiance data for each pixel is solved to obtain the irradiance distribution matrix on the inner surface of the switch cabinet.

[0035] Furthermore, in S3, three layout strategies are designed. One strategy is selected based on the deployment situation. The three layout strategies are as follows:

[0036] Strategy 1: Layout strategy based on detection effectiveness;

[0037] Strategy 2: Location-based full-detection layout strategy;

[0038] Strategy 3: Location-based non-full detection layout strategy.

[0039] Compared with the prior art, the present invention has the following beneficial technical effects:

[0040] This invention proposes an optimized arrangement method for partial discharge optical sensors in a typical 10kV switchgear structure to improve discharge detection rate and sensor deployment efficiency. First, a discharge source characteristic testing platform is established to measure the spectral energy distribution and power of typical partial discharges, calibrating the detection lower limit of the optical sensing device. Then, an optical path simulation model is used to analyze the optical signal propagation characteristics and luminous flux features of the switchgear, clarifying the reasonable arrangement range of the light-collecting units. Based on the chi-square test, a light-collecting unit layout strategy that accurately reflects the fault location is selected. The optimized sensor arrangement method is verified through a switchgear discharge optical simulation experimental platform, providing a reference for the performance requirements and optimized arrangement of partial discharge optical sensors in switchgear. This invention improves the detection sensitivity and accuracy of partial discharge signals by optimizing the sensor layout and quantity, while reducing the false alarm rate and enhancing the stability and reliability of the system. Attached Figure Description

[0041] Figure 1 Here is a structural diagram of the discharge source integration test system;

[0042] In the diagram: 1. High-voltage lead; 2. Insulating sleeve; 3. Fiber optic probe; 4. Fiber optic cable; 5. Fiber optic spectrometer; 6. Discharge power supply; 7. MgO total internal reflection coating; 8. Test base; 9. High-frequency current sensor; 10. Oscilloscope; 11. Power meter; 12. Photomultiplier tube; 13. Integrating sphere; 14. Collimating probe.

[0043] Figure 2 Measured waveforms of the light and electrical pulses during discharge;

[0044] Figure 3 A schematic diagram of the discharge integral spectrum distribution measured by the integrating sphere at different voltages;

[0045] Figure 4 The graph shows the variation of discharge optical power with discharge intensity for three typical discharges.

[0046] Figure 5 This is a graph showing the SiPM response as a function of optical power.

[0047] Figure 6 A simulation model of the optical path for partial discharge in a switchgear and a diagram illustrating the fault setting method.

[0048] Figure 7 A schematic diagram illustrating the light propagation process of various faults within the switchgear.

[0049] Figure 8 The relationship between optical power and the upper limit of reflection count;

[0050] Figure 9 An irradiance distribution diagram on the inner wall of the switchgear for a phase A fault in the main busbar through-wall bushing.

[0051] Figure 10 Irradiance diagrams of various phase faults on the right inner wall of the switchgear;

[0052] Figure 11 A similarity diagram of irradiance on the inner wall of the switchgear for various faults;

[0053] Figure 12 This is a layout diagram of the lighting unit that enables the detection of all types of faults;

[0054] Figure 13 This is a diagram illustrating the model's universality under various layout strategies. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of the present invention clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention; that is, the described embodiments are only a part of the embodiments of the present invention, and not all of them.

[0056] The components described and illustrated in the accompanying drawings and embodiments of this invention can be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of the invention provided in the following drawings is not intended to limit the scope of the claimed invention, but merely to illustrate one selected embodiment of the invention. All other embodiments obtained by those skilled in the art based on the accompanying drawings and embodiments of this invention without inventive effort are within the scope of protection of this invention.

[0057] It should be noted that the terms “comprising,” “including,” or any other variations are intended to cover non-exclusive inclusion, such that a process, element, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to the process, element, method, article, or apparatus.

[0058] The features and performance of the present invention will be further described in detail below with reference to embodiments.

[0059] The proposed method for optimizing the deployment of optical sensors for partial discharge detection in a 10kV metal-enclosed switchgear includes the following steps:

[0060] S1. Experimental analysis of the characteristics of the partial discharge light source in the switchgear, and the calibration results obtained.

[0061] First, an experimental platform was built to conduct actual measurements on the light source characteristics of partial discharge in typical defects, so as to obtain the discharge light power, discharge light intensity range, and detection limit of the optical sensing device.

[0062] 1.1 Setting up the experimental platform

[0063] The radiation band and power characteristics of the discharge source itself determine the sensor performance selection and detection limit. To accurately measure the optical radiation characteristics generated by partial discharge inside the switchgear, a partial discharge source calibration system was constructed using a total internal reflection optical integrating sphere. The partial discharge source calibration system includes a test base 8, an integrating sphere 13, a discharge power supply 6, and a high-voltage lead 1. The discharge power supply is located inside the integrating sphere 13, and the high-voltage lead 1 extends into the integrating sphere 13 and connects to the discharge power supply.

[0064] An optical fiber probe 3 and a photomultiplier tube 12 are externally connected to an integrating sphere 13; the optical fiber probe 3 is connected to an optical fiber spectrometer 5 via an optical fiber 4; the photomultiplier tube 12 is sequentially connected to a power meter 11, an oscilloscope 10, and a high-frequency current sensor 9; the photomultiplier tube 12 is connected to the integrating sphere 13 via a collimating probe 14.

[0065] An insulating sleeve 2 is provided at the top of the integrating sphere 13, and the high-voltage lead 1 passes through the insulating sleeve 2;

[0066] The inner wall of the integrating sphere 13 is coated with an MgO total reflection coating 7;

[0067] The inner wall of the integrating sphere is coated with MgO for total internal reflection to ensure that light radiation forms a uniform light field inside the sphere. The sphere has a diameter of 20 cm and is equipped with a detachable insulating sleeve 2 for introducing a high-voltage lead 1. Typical defect models can be set inside the sphere to generate surface discharge, corona discharge, and suspension discharge, and serve as the discharge source for testing, such as... Figure 1 As shown.

[0068] The light-gathering probe is a collimated multi-core fiber optic probe. On one hand, a fiber optic spectrometer is used to integrate and measure the energy distribution of the discharge spectrum, with an integration time of 80 power frequency cycles. On the other hand, a photomultiplier tube (PMT) is used to calibrate the optical power. The spectrometer is a QEPRO (detection range 185–1100 nm, resolution 0.14–7.7 nm, integration time 8 ms–60 mins). The PMT driver and readout unit use a low-noise regulated power supply (~1000V) and a transimpedance amplifier (25 dB), respectively, ensuring a high signal-to-noise ratio for weak light signals. A high-speed data acquisition card is used to simultaneously acquire the discharge and optical signals, ensuring data accuracy and synchronization.

[0069] The testing process is as follows:

[0070] First, record the background light noise level and the spectral intensity of the background light noise inside the integrating sphere to eliminate the interference of the basic light.

[0071] Then, reference spectra were recorded using a standard light source and a fiber optic spectrometer to provide a benchmark for subsequent discharge spectral measurements; the discharge range was determined by recording the initial discharge voltage (PDIV) and breakdown voltage (BDV) using a stepwise pressurization method.

[0072] Subsequently, starting from 1.1 times PDIV, the voltage was gradually increased to BDV, and the discharge light intensity S at each voltage level was recorded. d ;

[0073] Finally, a silicon photoelectric sensor was connected to the fiber optic probe, replacing the fiber optic spectrometer, to determine the sensor's detection energy range.

[0074] Figure 2 The figure shows a typical waveform of corona discharge during the positive half-cycle under an applied 50Hz AC voltage of 11.5kV.

[0075] 1.2 Analysis of Discharge Source Characteristics

[0076] Depend on Figure 3 It is known that the spectrum generated by partial discharge is mainly a band spectrum, which is related to the excitation and ionization of N2 molecules during the discharge process. The emission spectrum in the early stage of the discharge is mainly composed of the band spectrum of the first and second positive band systems of N2. As the discharge energy increases, the particle groups participating in energy level transitions in the air become more diverse, and the discharge process transforms into energy level transitions in a multi-atom system. The spectral characteristics of corona discharge and surface discharge are similar, but there are differences in intensity and distribution, especially the emission peaks (378.9 nm, 403.05 nm) formed by the transition of the second positive band system of N2 and the decomposition of O3 molecules into O+ ions. The average distribution ratio obtained by changing the applied voltage and tracking and measuring the discharge quantity and optical power yielded the main spectral energy distribution of various discharges. During the experiment, the voltage was gradually adjusted upward from PDIV, and the apparent discharge quantity and discharge optical power values ​​were tracked and recorded. The discharge quantity ranged from 33 pC (discharge onset) to 3.2 nC (near breakdown), and the optical power ranged from 1 μW to 52 μW.

[0077] In general, 81.5% of the main optical emission energies of various discharges are distributed in the ranges of 350nm–490nm and 550nm–900nm. Therefore, the selection of sensor wavelengths should include these main wavelengths. The discharge optical power P is calculated using the standard light source comparison method. d :

[0078]

[0079] Among them, t d S is the effective integration time of the spectrum. l S represents the discharge light intensity of a standard light source during the integration time. d These represent the discharge light intensity of the discharge source during the integration time.

[0080] The results of discharge optical power under different discharge conditions are shown (e.g.) Figure 4As shown in the figure, during the continuous increase of the cumulative discharge within the cycle, the discharge light power of a single discharge also increases. Among them, the discharge light power of corona discharge is in the range of 5 to 30 μW, the light power of surface discharge is below 1 to 11 μW, and the light power of suspended spark discharge is in the range of 3 to 52 μW. Therefore, in the simulation of the optical signal of the switch cabinet, the light source emission power should be set to 1 to 2 μW for reference.

[0081] 1.3 Determination of the detection limit of optical sensing devices

[0082] Related studies have shown that when using optical sensors to detect weak discharge light, the output signal has an approximately linear relationship with the discharge intensity. However, when using a spectrometer for lower limit calibration, there is often a low signal-to-noise ratio. Therefore, in this invention, a silicon photomultiplier tube (SiPM) is used to determine the lower limit of the received optical power.

[0083] The experiment used three typical discharges as light sources, adjusted the applied voltage to PDIV, and used a silicon photomultiplier tube (SiPM) sensor to receive the optical power P. S It can be calculated based on the area of ​​the integrating sphere and the effective aperture of the optical fiber:

[0084]

[0085] In the formula: P d —Discharge optical power / W,r f —Effective aperture of optical fiber / mm, r s —Integrating sphere inner radius / mm.

[0086] The optical pulse response intensity is represented by the average optical pulse amplitude, and its variation curve with the incident light power of the device is shown in the figure below. Figure 5 As shown, the silicon photomultiplier tube (SiPM) can effectively respond to incident light at the pW level. In measurements of three typical discharges, its minimum response optical power is 8.2 pW (corresponding to a 33 pC surface discharge). It should be noted that in actual sensor measurements, the effective light flux radius r should be considered. f Make corrections.

[0087] S2. Based on the calibration results obtained in S1, the optical path simulation model is used to analyze the optical signal propagation characteristics and luminous flux characteristics of the switchgear, and to determine the reasonable layout range of the light-collecting units.

[0088] 2.1 Construction of Optical Path Simulation Model

[0089] Switchgear equipment generally consists of four parts: circuit breaker compartment, busbar compartment, cable compartment, and relay instrument compartment. Among them, the busbar compartment has a compact structure, numerous busbars, and crowded components, making it most prone to partial discharge faults.

[0090] This invention focuses on the widely used KYN28 type 10kV switchgear. Based on its actual structure, dimensions, and material properties, a simulation model based on TracePro was built. Since the surface optical properties of the cabinet's inner surface and the materials of each component have a significant impact on the optical path simulation results, the simulation uses a bidirectional reflection distribution function (BRDF) to describe the material's reflection distribution characteristics. The calculation formula is as follows:

[0091]

[0092] Where A is the scale constant, representing the total intensity of scattering; B and g are reflection shape constants, representing the scattering shape; and r is the radius of the receiving cylinder in the reflection direction. The main components constituting the optical environment of the switchgear are the busbar copper bus, epoxy insulation components, and the inner wall of the painted aluminum shell. Their optical parameters are set as shown in Table 1.

[0093] Table 1 Surface Material Settings

[0094] Table 1Surface Material Settings

[0095]

[0096] The optical path simulation model of the switchgear and the location distribution of fault light sources are as follows: Figure 6 As shown, fault light sources are set inside the wall bushing of the main busbar of the three phases ABC in the model, on the surface of the post insulator, and inside the stationary contact box to simulate partial discharge faults caused by condensation and other reasons. Each type is equipped with four sets of simulation light sources (upper, lower, left, and right) to simulate partial discharge of the same type at different locations. When the fault changes to other locations in the spatial field of view, these four situations can be compared.

[0097] To investigate the influence of the shape and size of the discharge source on the luminous flux characteristics inside the switchgear, three sets of discharge shapes—cube, uniform sphere, and irregular shape—were set at each location. Each set of shapes was set with three sizes: large, medium, and small. The geometric parameters of the light source are shown in Table 2. The wavelength of the light source was set with reference to the spectral distribution of surface discharge, and the optical power was set with reference to the minimum optical power of surface discharge (1.12 μW).

[0098] Table 2 Geometric parameters of the fault source

[0099] Table 2Geometric Parameters of Fault Light Source

[0100]

[0101] The simulation employs a ray tracing algorithm, specifically: the discharge source is decomposed into discrete rays. Each ray strictly follows the laws of absorption, reflection, refraction, diffraction, and scattering on the material surface. When it encounters a medium interface, it is decomposed into several rays that continue to propagate forward, and its energy is changed according to the material properties. When the ray is decomposed a certain number of times, its energy will be less than a critical value, at which point ray tracing stops. The propagation path and energy change of a single ray are calculated, and the energy distribution under a specific light field is obtained by superimposing a sufficient number of ray samples.

[0102] The present invention sets the total number of rays from the discharge light source to 2.5 × 10⁻⁶. 4 strip. Figure 7 The results of ray tracing for the array of light sources are shown, with 25 tracks displayed for ease of presentation. Red rays represent light beams whose energy has attenuated to 66%–100% of their initial energy, green rays to 33%–66%, and blue rays to below 33%.

[0103] The results show that when a discharge fault occurs in phase A of the main busbar bushing, the total irradiance on all surfaces inside the equipment is 0.496 μW, accounting for 43.3% of the total energy of the light source. In this type of fault, some of the light loses a lot of energy after refraction and reflection inside the bushing, some shines directly onto the front wall of the switchgear, forming a relatively stable irradiance on that surface, and the rest encounters other components such as post insulators and main busbars during propagation, and is then reflected and propagated to the inner walls. At this time, the irradiance is significantly lower than that on other surfaces.

[0104] When a discharge fault occurs on phase A of the post insulator, the total irradiance on all surfaces inside the equipment is 0.558 μW, accounting for 49.8% of the total energy of the light source. In this type of fault, half of the light shines on the insulator and spreads outward after reflection and attenuation, while the other half is directly radiated outward. The light has few refractions and reflections, and it is easy to measure on all surfaces.

[0105] When phase A of the stationary contact box discharges, the total irradiance on all surfaces inside the equipment is 0.147 μW, which accounts for only 13.1% of the total energy of the light source. In this type of fault, most of the light rays will undergo multiple refractions and reflections within the small space of the contact box, resulting in severe attenuation of the emitted light energy and weak beam energy, making detection difficult.

[0106] To quantitatively describe the impact of reflection count on irradiance, a simulation was performed to investigate the relationship between the total irradiance of the inner walls of each surface and the upper limit of the reflection count. The results are as follows: Figure 8 As shown, for faults in the main busbar and post insulators, light rays with fewer than 5 reflections account for more than 50% of the total light energy, and the total irradiance tends to stabilize after 10 reflections. However, for faults in the stationary contact box, light rays with fewer than 5 reflections account for 30.2% of the light energy, proving that under this type of fault, the light rays undergo more reflections within the contact box space, resulting in severe attenuation of the emitted light energy.

[0107] 2.2 Determine the appropriate layout range of the light-collecting units, i.e., the optical sensors.

[0108] Simulation analysis of the luminous flux distribution inside the switch cabinet shows that the discharge type and the position of the light source have the greatest impact on the luminous flux characteristics. Based on the discharge type and the position of the light source, the reasonable arrangement range of the optical sensor can be preliminarily determined.

[0109] Taking phase A corona discharge as an example, such as Figure 9 As shown in the simulation model, the irradiance variation caused by a fault in the main busbar bushing on the front inner wall ranges from 0.4 to 0.6 μW, while the irradiance on the rear inner wall is only 0.1 to 0.2 μW. This indicates that the front inner wall is more sensitive to corona discharge.

[0110] Figure 10 The irradiance of each fault on the right inner wall is shown. It can be seen that the light flux on the right inner wall is significantly different when the three different phase positions of different discharge faults are different. Therefore, the discharge location of the switch cabinet can be realized by using the irradiance fingerprint feature.

[0111] S3. Based on the reasonable layout range of the daylighting units, a chi-square test is used to select a daylighting unit layout strategy that can accurately reflect the location of the fault.

[0112] 3.1. Using the chi-square test algorithm to determine the influencing factors of the discharge irradiance characteristics of the switchgear.

[0113] This invention employs the chi-square test algorithm to analyze the irradiance discrimination at different phase positions of various discharges. The inner surface is divided into several sensing points, with each sensing point having an area of ​​2 mm². 2 The irradiance data for each pixel is calculated to obtain the irradiance distribution matrix on the inner surface of the switch cabinet. For example... Figure 11 As shown, the matrices under 9 discharge conditions are compared with each other. Pixels with a deviation of less than 5% and not equal to 0 at the same position in the two matrices are considered to be similar (a matrix value of 0 indicates that this position cannot be used as a layout position for discharge lighting units, so it is not considered when counting similar pixels). The proportion of similar pixels to the total number of pixels is defined as the irradiance similarity rate.

[0114] Taking the similarity rate of irradiance on the right inner wall for various types of discharge faults as an example (e.g.) Figure 11 As shown in the figure, the intersections of rows and columns represent the irradiance similarity rates of the two types of faults. Taking the main busbar through-wall bushing phase A fault (hereinafter referred to as main A, other fault types are referred to as similar, which will not be elaborated in this invention) as an example, the similarity rates of this type of fault with main B and main C faults are 78.6% and 58.4%, respectively, while the similarity rate with other types of faults does not exceed 28.3%, indicating that the irradiance characteristics of different phases of the same type of fault have a certain degree of clustering. At the same time, regarding the influence of different discharge positions of the same type of in-phase fault on the irradiance results, combined with Figure 11 and Figure 12 The highest irradiance similarity rate (82.62%) was observed for the upper and lower faults, while the upper and right faults showed a larger difference (60.74%), although the difference was relatively small compared to non-identical or non-phase faults. The similarity rate of the light source at different locations on each of the six faces was calculated separately. The minimum similarity rate for each face was 54.12%, demonstrating that while the discharge location has some influence on the optical irradiance fingerprint, the similarity rate is relatively high compared to non-identical or non-phase faults, resulting in more similar irradiance fingerprints.

[0115] The above analysis shows that among the factors influencing the discharge irradiance characteristics of switchgear, discharge type and location have the greatest impact on irradiance, with a significantly higher similarity rate compared to dissimilar or out-of-phase faults. The shape and size of the light source have almost no effect on the irradiance characteristics of the switchgear's inner wall. Furthermore, when considering optical fault location in switchgear, faults of the same type and phase can be grouped under the same label for classification.

[0116] Once the discharge type and location are determined to have the greatest impact on irradiance, a basis for layout strategy can be provided.

[0117] 3.2 Layout Strategy of Lighting Units

[0118] The switchgear busbar compartment has a pressure relief cover on the upper outer shell to ensure that the critical pressure can be released instantly in the event of a deflagration inside the compartment, thus ensuring the safety of equipment and operators. In order for the pressure relief cover to operate normally, sensor light-collecting units cannot be installed in this position. In addition, based on the installation position and insulation distance of each component inside the busbar compartment, the approximate area where sensors can be laid out can be determined.

[0119] Taking the fiber-optic SiPM sensing system as an example, its lower limit of response power is 8.2pW, meaning the sensing system can only respond when the light energy received at the fiber optic front end exceeds this value. Therefore, the deployable area on the inner wall of the switch cabinet is divided into several light-receiving surfaces according to the fiber optic aperture. The irradiance on each light-receiving surface is simulated and calculated. If the irradiance is higher than 8.2pW, reliable discharge detection can be achieved at that point. When the same point can effectively respond to nine different fault conditions of varying types and phases, this is considered a reasonable sensor placement location. Finally, the layout positions of the light-collecting units capable of reliably detecting all types of faults are determined as follows: Figure 12 As shown, the front inner wall has the largest range for reliable detection of all types of faults, while the right inner wall and the stationary contact box are farther away, limiting the detection range.

[0120] 3.3 Analysis of Multi-Objective Layout Optimization Strategies

[0121] Based on the analysis results of step 3.2, three layout strategies are proposed for the two optimization objectives of switchgear discharge fault detection capability and location capability.

[0122] 1) Layout strategy based on detection effectiveness (Strategy 1): Based on the layout range for reliable detection of all types of faults, the average response irradiance of different fault light sources on each surface and each pixel is calculated within this range, and the point with the highest average irradiance is taken as the coordinate of the lighting unit arrangement. Under this layout strategy, the lighting unit number is denoted as Ti (i = 1, 2, ..., 5), where i represents the front, back, left, right, and bottom surface numbers, respectively.

[0123] 2) Location-based full-detection layout strategy (Strategy 2): Based on the reliable detection layout range for all types of faults, the chi-square value of different pixels on each surface is calculated within this range, and the point with the highest chi-square value is taken as the coordinate of the lighting unit arrangement on that surface. Under this layout strategy, each lighting unit is numbered Ai (i = 1, 2, ..., 5). There is a large difference in coordinates between layout strategies 1 and 2, but the characteristic positions on the rear surface overlap.

[0124] 3) Location-based incomplete detection layout strategy (Strategy 3): When considering discharge fault location, it is not necessary for all light-collecting units to respond to all faulty light sources; the lack of response from some units can also be used as a condition for determining the fault location. Therefore, a chi-square test is performed on the global irradiance distribution matrix for various types of faults. The point with the highest chi-square value on each surface is selected, and the combination of these points is tested to see if it can guarantee that there are two or more sensors capable of detecting each type of fault. If this condition is met, this layout strategy can reliably detect faults. If this condition is not met, the point with the second highest chi-square value is selected, and the test is repeated until the detection performance of the combination meets the standard. Under this layout strategy, each light-collecting unit is denoted as F. i (i = 1, 2, ..., 5). This strategy has two light-gathering units outside the layout range for reliable detection of all types of faults.

[0125] Finally, through the chi-square test of all variables, the optimal installation coordinates under the three strategies were obtained, as shown in Table 3.

[0126] Table 3 Coordinates of Three Lighting Unit Layout Strategies

[0127] Table 3Coordinates for Three Light Collection Unit Layout Strategies

[0128]

[0129] Further research was conducted on the universality and positioning accuracy of the three strategies. Simulation results of non-standard lighting units with altered shape, size, aperture angle, and installation distance were input into the model for testing. The accuracy of the output results was compared to verify the universality of the model.

[0130] Figure 13Figure (a) shows the test accuracy under various layout strategies and classification algorithms after changing the parameters of the light-gathering unit. Taking the extreme learning machine, which has the lowest overall accuracy, as an example, the localization accuracy under the three layout strategies is 88.1%, 90.2%, and 89.3%, respectively. K-nearest neighbors and BP neural networks outperform these, demonstrating that all three strategies have high universality. Analyzing the test results... Figure 13 Figures (b) to (d) show the test confusion matrices under the three layout strategies of the Extreme Learning Machine model. It can be seen that misclassifications still frequently occur between different phases of the same type of fault, and the classification accuracy for stationary contact box faults is relatively low. Therefore, when actually deploying sensors in switchgear, a more suitable sensor deployment strategy should be selected by comprehensively considering the installation conditions at the points of installation, economic factors, and the importance level of the equipment.

[0131] This invention studies the discharge optical power and spectral energy distribution characteristics of three typical discharges: corona discharge, surface discharge, and suspended spark discharge. 81.5% of the main light emission energy of each type of discharge is distributed in the ranges of 350nm–490nm and 550nm–900nm. The optical power of corona discharge is in the range of 5–30μW, the optical power of surface discharge is mostly below 1–11μW, and the optical power of suspended spark discharge is mostly in the range of 3–52μW. Based on this, the lower limit of the sensor power response is determined to be 8.2pW. Optical simulation and chi-square test were used to identify the area inside the switchgear most sensitive to partial discharge detection. Under corona discharge conditions, the irradiance variation range of the light-collecting units arranged on the front inner wall is 0.4–0.6μW, significantly higher than the 0.1–0.2μW on the rear inner wall. The average chi-square value of the detection rate of the sensing points on the front inner wall is 35.6, much higher than other areas. Three optimized layout strategies were proposed, targeting detection rate and completeness respectively. The results showed that the layout strategy with the highest detection efficiency achieved a discharge detection rate of over 90%. Although the non-completeness layout strategy sacrificed some detection rate, it significantly reduced the number of sensors required. For discharge localization, the localization accuracy under the three layout strategies was 88.1%, 90.2%, and 89.3%, respectively.

[0132] The research results of this invention provide a reference for the performance requirements and installation and deployment methods of optical partial discharge sensors in switch cabinets in terms of detection rate and economy.

[0133] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. An optimized method for deploying optical sensors for partial discharge detection in a 10kV metal-enclosed switchgear, characterized in that, Includes the following steps: S1. Conduct actual measurements on the light source characteristics of partial discharge in typical defects and obtain calibration results; The calibration results include discharge light power, discharge light intensity range, and the detection limit of the optical sensing device; S2. Based on the calibration results obtained in S1, the optical path simulation model is used to analyze the optical signal propagation characteristics and optical flux characteristics of the switch cabinet, and to determine the reasonable layout range of the light-collecting unit. S3. Based on the reasonable layout range of the lighting units, a lighting unit layout strategy that can accurately reflect the fault location is selected using the chi-square test.

2. The method for optimizing the deployment of optical sensors for partial discharge detection in a 10kV metal-enclosed switchgear according to claim 1, characterized in that, In S1, the light source characteristics of partial discharge of typical defects were measured using the established experimental platform. The experimental platform uses a total internal reflection optical integrating sphere to construct a partial discharge source calibration system. The partial discharge source calibration system includes an integrating sphere (13), a discharge power supply (6), and a high-voltage lead (1). The discharge power supply is located inside the integrating sphere (13), and the high-voltage lead (1) extends into the integrating sphere (13) and connects to the discharge power supply. An optical fiber probe (3) and a photomultiplier tube (12) are connected to the integrating sphere (13); the optical fiber probe (3) is connected to an optical fiber spectrometer (5) via an optical fiber (4); and the photomultiplier tube (12) is connected in sequence to a power meter (11), an oscilloscope (10), and a high-frequency current sensor (9). An insulating sleeve (2) is provided at the top of the integrating sphere (13), and the high voltage lead (1) passes through the insulating sleeve (2); The inner wall of the integrating sphere (13) is coated with a total internal reflection coating (7) of MgO.

3. The method for optimizing the deployment of optical sensors for partial discharge detection in a 10kV metal-enclosed switchgear according to claim 2, characterized in that, The light source characteristics of partial discharge in typical defects were measured using the established experimental platform. The specific testing methods are as follows: First, record the background light noise level and the background light noise spectral intensity inside the integrating sphere (13) to eliminate basic light interference; Then, a reference spectrum is recorded using a standard light source and a fiber optic spectrometer (5) to provide a benchmark for subsequent discharge spectrum determination; The discharge range was determined by recording the initial discharge voltage and breakdown voltage using a step-by-step voltage application method. Subsequently, starting from a predetermined initial discharge voltage, the voltage was gradually increased to the breakdown voltage, and the discharge light intensity at each voltage level was recorded. Finally, the sensor is connected to the fiber optic probe (3) to replace the fiber optic spectrometer (5) to determine the range of discharge light intensity of the sensor.

4. The optimized deployment method for optical sensors used in partial discharge detection within a 10kV metal-enclosed switchgear according to claim 3, characterized in that, The discharge optical power is calculated using the standard light source comparison method, and the expression is: Among them, t d S is the effective integration time of the spectrum. l S represents the discharge light intensity of a standard light source during the integration time. d These represent the discharge light intensity of the discharge source during the integration time.

5. The method for optimizing the deployment of optical sensors for partial discharge detection in a 10kV metal-enclosed switchgear according to claim 1, characterized in that, In S1, the process for determining the detection lower limit of the optical sensing device is as follows: A silicon photomultiplier tube is used to determine the lower limit of the received optical power. The optical power received by the silicon photomultiplier tube sensor is P. s Calculated based on the area of ​​the integrating sphere and the effective aperture of the optical fiber: In the formula: P d r is the discharge optical power. f r is the fiber aperture. s Let be the inner radius of the integrating sphere.

6. The method for optimizing the deployment of optical sensors for partial discharge detection in a 10kV metal-enclosed switchgear according to claim 1, characterized in that, In S2, the specific process of building the optical path simulation model is as follows: Based on the actual structure, dimensions, and material properties of the switchgear under test, an optical path simulation model based on TracePro was built. The simulation uses a bidirectional reflection distribution function to describe the reflection distribution characteristics of the material, and its calculation formula is as follows: Wherein, BRDF represents the reflection distribution characteristics; A is the scale constant, representing the total intensity of scattering; B and g are the reflection shape constants, representing the scattering shape; and r is the radius of the receiving cylinder in the reflection direction.

7. The optimized deployment method for optical sensors used in partial discharge detection inside a 10kV metal-enclosed switchgear according to claim 1, characterized in that, In S3, the chi-square test algorithm is used to analyze the irradiance discrimination at different phase positions of various discharges. The inner surface is divided into several sensing points, and the area of ​​each sensing point is set to X mm. 2 The irradiance data for each pixel is solved to obtain the irradiance distribution matrix on the inner surface of the switch cabinet.

8. The optimized deployment method for optical sensors used in partial discharge detection within a 10kV metal-enclosed switchgear according to claim 1, characterized in that, In S3, three layout strategies are designed. One of them is selected according to the deployment situation. The three layout strategies are as follows: Strategy 1: Layout strategy based on detection effectiveness; Strategy 2: Location-based full-detection layout strategy; Strategy 3: Location-based non-full detection layout strategy.

Images