GIS cavity and GIS internal fault prediction system and method

Abstract

The invention relates to the technical field of power system detection, in particular to a GIS cavity based on optical imaging and a GIS internal fault prediction system and method. The GIS cavity is provided with a plurality of detection point positions, each detection point position comprises more than two image acquisition points, each image acquisition point comprises a lens, a switching block, an optical fiber probe and an optical fiber connector, the top of each lens is a working part and extends into the cavity, the bottom of each lens is a connecting part, the switching block is located at the bottom of each lens, and the lenses are fixed on the wall of the cavity through the switching blocks. A containing hole and a wiring hole are formed in the middle and the two sides of the switching block respectively, the optical fiber probe and the optical fiber connector are located in the containing hole, the bottom of the optical fiber probe is connected with the optical fiber connector, and the top of the optical fiber probe extends to the lens working part. An optical fiber connected to a data center does not need to be arranged at each detection point, so that the installation cost is saved. And a plurality of optical fibers aligned to different directions are arranged in the optical fiber probe, so that dead-angle-free image acquisition can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of power system detection, and in particular to a GIS cavity based on optical imaging, a GIS internal fault prediction system and a method. Background Art

[0002] GIS (Gas Insulated Switchgear) is a high-voltage switchgear used in power systems, and is widely used in places such as substations and power plants. The main feature of GIS is that all components of high-voltage switchgear, such as brackets, disconnectors, grounding switches, current transformers, voltage transformers and busbars, are enclosed in a metal casing filled with insulating gas (usually sulfur hexafluoride, SF6). However, its closed nature also makes it more difficult to detect internal faults. The current technology mainly uses ultrasonic signals to detect micro-discharges inside, or arranges optical fibers on the outer wall of the pipeline to perform optical imaging inside and transmits optical signals to the data center for monitoring. However, ultrasonic signals cannot accurately locate the discharge position, and cannot detect faults such as local overtemperature. The use of special optical fibers such as infrared and ultraviolet optical fibers to transmit signals to long-distance GIL is too costly and difficult to apply in practice. Moreover, for densely installed GIS, the use of ultrasonic signals for detection will cause great interference. Summary of the invention

[0003] In view of the defects of the prior art, the present invention provides a GIS cavity, GIS internal fault prediction system and method, by setting a fiber optic connection port on the outer wall of the GIS pipeline, and patrolling the line by manually or robot-carrying imaging equipment, acquiring optical images at the connection port and transmitting the optical images to a data center through a network for comparison with typical fault images obtained in advance through experiments, so as to detect faults and assess risks at the same time.

[0004] In order to solve the technical problem, the technical solution adopted by the present invention is: a GIS cavity, a plurality of detection points are arranged on the GIS cavity, each detection point includes more than two image acquisition points, one of which is connected to an imaging device, called image acquisition point A, and the other image acquisition points are connected to image acquisition point A, called image acquisition point B; image acquisition point A and image acquisition point B both include a lens, an adapter block, an optical fiber probe and an optical fiber connector, the top of the lens is a working part extending into the cavity, the bottom of the lens is a connecting part, the adapter block is located at the bottom of the lens, the lens is fixed to the cavity wall through the adapter block, a receiving hole and a wiring hole are respectively opened in the middle and on both sides of the adapter block, the optical fiber probe and the optical fiber connector are located in the receiving hole, the bottom of the optical fiber probe is connected to the optical fiber connector, the top of the optical fiber probe extends to the working part of the lens, and the optical fiber passes through the wiring hole to realize the connection from image acquisition point B to image acquisition point A.

[0005] Furthermore, the optical fiber probe includes a plurality of multimode optical fibers facing in different directions, and each multimode optical fiber includes four optical fibers, namely, one infrared optical fiber, one ultraviolet optical fiber and two visible optical fibers.

[0006] Furthermore, the working portion of the lens is semicircular, and the connecting portion is a flange.

[0007] Furthermore, the accommodating hole and the wiring hole at the image acquisition point A are through holes, and the accommodating hole at the image acquisition point B is a non-through hole.

[0008] Furthermore, for GIS equipment with an installation density greater than the set requirement, multiple detection points are connected via optical fibers.

[0009] Furthermore, the detection points are arranged along the axial direction of the GIS cavity, and the image acquisition points are evenly arranged along the radial direction of the GIS cavity.

[0010] Furthermore, each detection point includes one image acquisition point A and two image acquisition points B, and the optical fiber probe includes five multimode optical fibers facing five different directions.

[0011] The present invention also discloses a GIS internal fault prediction system, including a GIS cavity, an imaging device and a data center, the GIS cavity is the above-mentioned GIS cavity, the optical fiber connector of the imaging device is connected to the optical fiber connector of the image acquisition point A, the imaging device collects ultraviolet images, infrared images and visible light images inside the GIS, and after completing the image acquisition, transmits the images to the data center through a wireless network, the data center reconstructs the received multiple multi-angle two-dimensional images into three-dimensional images through an image reconstruction algorithm, and then uses a feature extraction algorithm to compare the reconstructed three-dimensional image with a typical GIS internal fault image obtained in advance through experiments to analyze the probability of failure.

[0012] The present invention also discloses a GIS internal fault prediction method, which is implemented based on the above GIS cavity and includes the following steps: S01. Connect the optical fiber connector of the imaging device to the optical fiber connector of the image acquisition point A. The imaging device first directly acquires the ultraviolet image and the infrared image, then emits visible light into the GIS cavity through one of the two visible optical fibers, and then acquires the visible light image inside the GIS cavity through the other visible optical fiber; S02. Before the imaging device is installed in the GIS cavity, the data center first obtains ultraviolet images, infrared images, and visible light images of typical faults inside the GIS, reconstructs the collected images through deep learning methods, generates a three-dimensional image of the inside of the GIS pipeline, and stores the generated image in the database; After the image acquisition is completed, the image is transmitted to the data center via a wireless network. The data center reconstructs the received multiple multi-angle two-dimensional images into three-dimensional images through an image reconstruction algorithm; The image reconstruction process is: In each image, a feature extraction algorithm is used to extract feature points in the image. Each feature point is represented by a position coordinate (x, y) and a direction descriptor. The feature point direction descriptor is calculated using the following formula: , Where i represents the i-th feature point, n represents the number of feature points, represents the weight coefficient of the i-th feature point, Represents the gradient direction of the i-th feature point; Match the feature points in multiple images, and find the same feature point pairs under different perspectives through the nearest neighbor algorithm or the matching algorithm based on Euclidean distance. The same feature point pairs under different perspectives are called matching point pairs. The matching point pairs satisfy geometric consistency. The basic matrix F is calculated using the matching point pairs, and combined with the intrinsic parameter matrix K of the imaging camera, the essential matrix E is obtained: , in represents the intrinsic parameter matrix of camera one, represents the intrinsic parameter matrix of camera 2, The essential matrix describes the relative position of a single fiber imaging point in three-dimensional space. The essential matrix E is subjected to singular value decomposition to extract the rotation matrix R and translation vector T of a single fiber imaging point, and the spatial coordinates of the fiber imaging point are obtained. The solution that meets geometric consistency is screened out through triangulation. Combined with the imaging device posture, the three-dimensional coordinates of the matching point pairs are calculated using triangulation. Based on the three-dimensional coordinates of the matching point pairs, a multi-view stereo vision method is used to generate a dense point cloud. The dense point cloud is fitted with a Poisson surface reconstruction algorithm to obtain a complete three-dimensional model. S03, the image collected by the imaging device in real time is reconstructed in three dimensions and then compared with the three-dimensional image in the database. The comparison process is as follows: Determine the neighborhood of each feature point in the image. Each neighboring point of , calculate the geometric relationship between point pairs, first calculate the point and Point The normal vector and , calculate the point pair The geometric features between include three key features: Angle ,point point to The vector and The normal vector The angle is calculated as: , Angle ,point The normal vector and Point The normal vector The angle between them is calculated as: , Angle ,point The normal vector and Point point to The angle between the vectors is calculated as: , in , Respectively indicate points , The spatial coordinates of , Respectively indicate points , The normal vector of represents the norm of the orientation quantity; The above three angles are divided into multiple intervals, and the characteristic values ​​of each point pair are accumulated in the form of a histogram to form the SPFH feature. The FPFH is calculated by weighted summation based on the SPFH features of other points in the neighborhood. The calculation formula is: , in represents the number of points in the neighborhood, Indicate point and Point The distance between The FPFH of each point is stored as the feature information of the feature point. By comparing the feature point information of the image obtained in the test with the feature point information obtained by imaging inside the actual GIS, it is possible to analyze whether a fault occurs inside the actual GIS or predict possible faults.

[0013] Furthermore, in step S02, the simulation of different faults includes local corona discharge simulation, local overheating fault simulation and simulation of foreign matter and physical defects inside the GIS pipeline. The local corona discharge simulation collects local discharge images through ultraviolet light imaging, the local overheating fault simulation collects local overheating images through infrared light imaging, and the simulation of foreign matter and physical defects inside the GIS pipeline collects internal visible light images through visible light imaging.

[0014] Beneficial effects of the present invention: 1. It is not necessary to lay optical fiber connected to the data center for each detection point, saving installation costs.

[0015] 2. Five optical fibers pointing in different directions are arranged inside the fiber optic probe. Each image acquisition point contains three fiber optic probes, which can achieve image acquisition without blind spots.

[0016] 3. The shape of the pipeline can be flexibly designed, realizing flexible selection of monitoring points.

[0017] 4. For densely installed GIS equipment, the overall imaging of the inside of the equipment can be achieved through parallel fiber optic interfaces, which has a wider application than existing technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is the front cross-sectional view of the GIS cavity; Figure 2 It is a schematic diagram of the structure of the adapter block and the lens; Figure 3 It is a side cross-sectional view of the GIS cavity; Figure 4 It is a schematic diagram of the top view structure of the transfer block; Figure 5 It is a flowchart of the GIS internal fault prediction method; In the figure: 1. insulator, 2. outer wall of cavity, 3. guide rod, 4. image acquisition point A, 5. optical fiber, 6. bracket, 41. bolt, 42. adapter block, 43. lens, 44. optical fiber probe, 45. optical fiber adapter, 46. receiving hole, 47. wiring hole. DETAILED DESCRIPTION

[0019] In order to be able to understand the features and technical contents of the embodiments of the present disclosure in more detail, the implementation of the embodiments of the present disclosure is described in detail below in conjunction with the accompanying drawings. The attached drawings are for reference only and are not used to limit the embodiments of the present disclosure. In the following technical description, for the convenience of explanation, a full understanding of the disclosed embodiments is provided through multiple details. However, one or more embodiments can still be implemented without these details. In other cases, to simplify the drawings, well-known structures and devices can be simplified for display.

[0020] The terms "first", "second", etc. in the embodiments of the present disclosure are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchanged where appropriate, so as to describe the embodiments of the present disclosure described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions.

[0021] Unless otherwise stated, the term "plurality" means two or more.

[0022] In the embodiment of the present disclosure, the character " / " indicates that the preceding and following objects are in an "or" relationship. For example, A / B indicates: A or B.

[0023] Example 1 This embodiment discloses a GIS cavity, such as Figure 1 As shown, the GIS cavity includes a cavity outer wall 2, an insulator 1 and a guide rod 3 are arranged in the GIS cavity, and a plurality of detection points are arranged on the GIS cavity outer wall 2, each detection point includes more than two image acquisition points, one of which is connected to the imaging device, called image acquisition point A4, and the other image acquisition points are connected to the image acquisition point A4, called image acquisition point B.

[0024] like Figure 2 , 3 As shown, both image acquisition point A and image acquisition point B include a lens 43, an adapter block 42, an optical fiber probe 44 and an optical fiber connector 45. The top of the lens 43 is a semicircular working part extending into the interior of the GIS cavity. The bottom of the lens 43 is a flange connection part. The adapter block 42 is located at the bottom of the lens 43. The adapter block 42 is fixed to the cavity wall 2 by bolts 41, so that the lens 43 is pressed against the outer wall 2 of the cavity, and a sealing ring is pressed under the flange to achieve sealing. The adapter block 42 has a receiving hole 46 and a wiring hole 47 in the middle and on both sides, respectively. The optical fiber probe 44 and the optical fiber connector 45 are located in the receiving hole 46. The bottom of the optical fiber probe 44 is connected to the optical fiber connector 45. The top of the optical fiber probe 44 extends to the working part of the lens 43. The optical fiber 5 passes through the wiring hole 47 to achieve the connection between the image acquisition point B and the image acquisition point A. A bracket 6 is provided on the GIS cavity to support the optical fiber. The optical fiber probe 44 includes five multimode optical fibers facing five different directions, each of which includes four optical fibers, namely one infrared optical fiber, one ultraviolet optical fiber and two visible optical fibers.

[0025] like Figure 3 , 4 As shown, the receiving hole 46 and the wiring hole 47 of the image acquisition point A are through holes, the optical fiber connector 45 is located in the receiving hole 46, and the optical fiber connector of the imaging device is located above the optical fiber connector 45 and connected to it. The receiving hole of the image acquisition point B is a non-through hole, the optical fiber connector 45 is located in the receiving hole, and the bottom of the optical fiber probe is connected to the optical fiber 5 through the wiring hole, and is connected to the optical fiber connector of the image acquisition point A through the optical fiber 5.

[0026] For GIS equipment with an installation density greater than the set requirements, multiple detection points are connected through optical fibers, and image acquisition can be achieved by simply connecting the imaging devices together at the detection points.

[0027] In this embodiment, the detection points are arranged along the axial direction of the GIS cavity, and the image acquisition points are evenly arranged along the radial direction of the GIS cavity. Each detection point includes one image acquisition point A and two image acquisition points B, that is, there are a total of three image acquisition points evenly distributed along the radial direction of the GIS cavity.

[0028] The GIS cavity described in this embodiment is suitable for manual inspection or robot inspection. Taking robot inspection as an example, when the inspection robot moves to the measurement point, it automatically connects the optical fiber connector of the imaging device to the optical fiber connector 45 in the center of the adapter plate on the outer wall of the cavity. First, the ultraviolet image and the infrared image are directly collected, and then visible light is irradiated into the cavity through one of the two visible optical fibers, and the visible light image inside the cavity is collected through the other visible optical fiber. After the image acquisition is completed, the image is transmitted to the data center through the wireless network. The data center reconstructs the multiple multi-angle two-dimensional images obtained into three-dimensional images through the image reconstruction algorithm. Then, the feature extraction algorithm is used to compare the obtained three-dimensional image with the typical fault image of the GIS internal obtained in advance through experiments, and the probability of failure is analyzed to guide the operation and maintenance personnel to perform maintenance. In particular, for densely installed GIS equipment, multiple optical fiber interfaces can be connected through optical fibers, and image acquisition can be achieved by docking only one interface.

[0029] Example 2 The present embodiment discloses a GIS internal fault prediction system, including a GIS cavity, an imaging device and a data center. The GIS cavity is the GIS cavity described in Example 1. The optical fiber connector of the imaging device is connected to the optical fiber connector of the image acquisition point A. The imaging device acquires ultraviolet images, infrared images and visible light images inside the GIS. After completing the image acquisition, the images are transmitted to the data center through a wireless network. The data center reconstructs the received multiple multi-angle two-dimensional images into three-dimensional images through an image reconstruction algorithm, and then uses a feature extraction algorithm to compare the reconstructed three-dimensional images with typical internal fault images of the GIS obtained in advance through experiments to analyze the probability of failure.

[0030] Example 3 This embodiment discloses a GIS internal fault prediction method, such as Figure 5 As shown, the following steps are included: S01. Connect the optical fiber connector of the imaging device to the optical fiber connector of the image acquisition point A. The imaging device first directly acquires the ultraviolet image and the infrared image, then irradiates the visible light into the GIS cavity through one of the two visible optical fibers, and then acquires the visible light image inside the GIS cavity through the other visible optical fiber. After the image acquisition is completed, the image is transmitted to the data center through the wireless network. The data center reconstructs the received multiple multi-angle two-dimensional images into three-dimensional images through the image reconstruction algorithm.

[0031] S02. Before the imaging equipment is installed in the GIS cavity, the data center first obtains ultraviolet images, infrared images and visible light images of typical faults inside the GIS, reconstructs the acquired images through deep learning methods, generates a three-dimensional image of the inside of the GIS pipeline, and stores the generated image in the database. The simulations for different faults include local corona discharge simulation, local overheating fault simulation, and foreign matter and physical defect simulation inside the GIS pipeline. The local corona discharge simulation collects local discharge images through ultraviolet imaging, the local overheating fault simulation collects local overheating images through infrared imaging, and the foreign matter and physical defect simulation inside the GIS pipeline collects visible light images inside through visible light imaging.

[0032] Because the cavity involved in this method contains multiple multi-angle image acquisition fibers, the collected images will contain many overlapping areas, and the images obtained in the experiment are not completely consistent with the internal images in the actual GIS operation. Therefore, directly using unprocessed images for comparison will have a poor effect and increase the server burden. Based on this, this method uses a deep learning method to reconstruct multiple images taken at different angles in three dimensions. The image reconstruction process is as follows: In each image, a feature extraction algorithm (such as scale-invariant feature transform SIFT) is used to extract feature points in the image. Each feature point is represented by a position coordinate (x, y) and a direction descriptor. The feature point descriptor is calculated using the following formula: , Where i represents the i-th feature point, n represents the number of feature points, represents the weight coefficient of the i-th feature point, Represents the gradient direction of the i-th feature point.

[0033] Match the feature points in multiple images, and find the same feature point pairs under different perspectives through the nearest neighbor algorithm or the matching algorithm based on Euclidean distance. The same feature point pairs under different perspectives are called matching point pairs. The matching point pairs satisfy geometric consistency. The matching error is defined as: , in and is the matching point in the image, F is the basic matrix, and the matching result is optimized by minimizing the error.

[0034] The basic matrix F is calculated using the matching point pairs, and combined with the intrinsic parameter matrix K of the imaging camera, the essential matrix E is obtained: , in represents the intrinsic parameter matrix of camera one, represents the intrinsic parameter matrix of camera 2, The essential matrix describes the relative position of a single fiber imaging point in three-dimensional space. The essential matrix E is subjected to singular value decomposition to extract the rotation matrix R and translation vector T of a single fiber imaging point, and the spatial coordinates of the fiber imaging point are obtained. The solution that meets geometric consistency is screened out through triangulation.

[0035] Combined with the known position of the imaging device, the three-dimensional coordinates of the matching point pairs are calculated using triangulation: Assume that the points in the image and , whose homogeneous coordinates are expressed as and . Using the camera projection matrix and , calculate the three-dimensional point The coordinates satisfy the following equation: , By solving this equation using the least squares method, we can obtain the three-dimensional point coordinates, which is a sparse point cloud.

[0036] Based on the generation of sparse point cloud, a multi-view stereo vision method is used to generate a dense point cloud. The dense point cloud is surface fitted through the Poisson surface reconstruction algorithm to obtain a complete three-dimensional model.

[0037] S03. The image acquisition of this method is divided into two parts. One part is the typical discharge image obtained by conducting simulation tests, and the other part is the image obtained by imaging the inside of the running GIS. The images of both parts are reconstructed in three dimensions and then compared using feature extraction algorithms, specifically: Determine the neighborhood of each feature point in the image. For each point in the point cloud , we define its local neighborhood by a predefined radius r or k nearest neighbors within the neighborhood Neighborhood The size of has an important influence on the scale of feature description. A smaller radius can capture more detailed features, while a larger radius can provide global information.

[0038] For feature points Each neighboring point of , calculate the geometric relationship between point pairs, first calculate the point and Point The normal vector and , calculate the point pair The geometric features between include three key features: Angle ,point point to The vector and The normal vector The angle is calculated as: , Angle ,point The normal vector and Point The normal vector The angle between them is calculated as: , Angle ,point The normal vector and Point point to The angle between the vectors is calculated as: , in , Respectively indicate points , The spatial coordinates of , Respectively indicate points , The normal vector of represents the norm of the orientation quantity; The above three angle features describe and The geometric relationship between the three angles above is divided into multiple intervals, and the eigenvalues ​​of each point pair are accumulated in the form of a histogram to form the SPFH (Simplified Point Feature Histogram) feature. The FPFH (Fast Point Feature Histogram) is calculated by weighted summation based on the SPFH features of other points in the neighborhood. The calculation formula is: , in represents the number of points in the neighborhood, Indicate point and Point The distance between The FPFH of each point is stored as the feature information of the feature point. By comparing the feature point information of the image obtained in the test with the feature point information obtained by imaging inside the actual GIS, it is possible to analyze whether a fault occurs inside the actual GIS or predict possible faults.

[0039] In other embodiments, the method can also be implemented in a device, such as a GIS internal fault prediction device, including a processor and a memory. Optionally, the device can also include a communication interface and a bus. The processor, the communication interface, and the memory can communicate with each other through the bus. The communication interface can be used for information transmission. The processor can call the logic instructions in the memory to execute the GIS internal fault prediction method of the above embodiment.

[0040] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product.

[0041] The memory, as a computer-readable storage medium, can be used to store software programs and computer executable programs, such as program instructions / modules corresponding to the method in the embodiment of the present disclosure. The processor executes the function application and data processing by running the program instructions / modules stored in the memory, that is, realizing the GIS internal fault prediction method in the above embodiment.

[0042] The memory may include a program storage area and a data storage area, wherein the program storage area may store an operating system and an application required for at least one function; the data storage area may store data created according to the use of the terminal device, etc. In addition, the memory may include a high-speed random access memory and may also include a non-volatile memory.

[0043] The method is integrated on a computer-readable storage medium, such as a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured to execute the above-mentioned GIS internal fault prediction method. The computer-readable storage medium may be a transient computer-readable storage medium or a non-transient computer-readable storage medium.

[0044] The technical solution of the embodiment of the present disclosure can be embodied in the form of a software product, which is stored in a storage medium and includes one or more instructions for enabling a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the embodiment of the present disclosure. The aforementioned storage medium may be a non-transient storage medium, including: a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a disk or an optical disk, and other media that can store program codes, or a transient storage medium.

[0045] The above description and the accompanying drawings fully illustrate the embodiments of the present disclosure so that those skilled in the art can practice them. Other embodiments may include structural, logical, electrical, process and other changes. The embodiments represent only possible changes. Unless explicitly required, separate components and functions are optional, and the order of operation may vary. The parts and features of some embodiments may be included in or replace the parts and features of other embodiments. Moreover, the words used in this application are only used to describe the embodiments and are not used to limit the scope of protection. As used in the description in the text, unless the context clearly indicates, the singular forms of "a", "an" and "the" are intended to include plural forms as well. Similarly, the term "and / or" as used in this application refers to any and all possible combinations of one or more associated listings. In addition, when used in this application, the term "comprise" and its variants "comprises" and / or comprising refer to the presence of stated features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or groups thereof. In the absence of further restrictions, the elements defined by the sentence "including one..." do not exclude the presence of other identical elements in the process, method or device including the elements. In this article, each embodiment may focus on the differences from other embodiments, and the same and similar parts between the embodiments may refer to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method part disclosed in the embodiments, then the relevant parts can refer to the description of the method part.

[0046] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software may depend on the specific application and design constraints of the technical solution. The technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the embodiments of the present disclosure. The technicians may clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here.

Claims

1. A GIS cavity, characterized in that: There are multiple detection points on the GIS cavity, and each detection point includes more than two image acquisition points, one of which is connected to the imaging device, called image acquisition point A, and the other image acquisition points are connected to image acquisition point A, called image acquisition point B; image acquisition point A and image acquisition point B both include lenses, adapter blocks, fiber optic probes and fiber optic connectors. The top of the lens is the working part, which extends into the cavity, and the bottom of the lens is the connecting part. The adapter block is located at the bottom of the lens, and the lens is fixed to the cavity wall through the adapter block. The middle and both sides of the adapter block are respectively provided with a receiving hole and a wiring hole. The fiber optic probe and the fiber optic connector are located in the receiving hole, and the bottom of the fiber optic probe is connected to the fiber optic connector. The top of the fiber optic probe extends to the working part of the lens, and the optical fiber passes through the wiring hole to realize the connection from image acquisition point B to image acquisition point A.

2. The GIS cavity according to claim 1, characterized in that: The optical fiber probe includes a plurality of multimode optical fibers facing in different directions, and each multimode optical fiber includes four optical fibers, namely, one infrared optical fiber, one ultraviolet optical fiber and two visible optical fibers.

3. The GIS cavity according to claim 1, characterized in that: The working part of the lens is semicircular and the connecting part is a flange.

4. The GIS cavity according to claim 1, characterized in that: The receiving hole and the wiring hole of the image acquisition point A are through holes, and the receiving hole of the image acquisition point B is a non-through hole.

5. The GIS cavity according to claim 1, characterized in that: For GIS equipment with an installation density greater than the set requirements, multiple detection points are connected via optical fibers.

6. The GIS cavity according to claim 1, characterized in that: The detection points are arranged along the axial direction of the GIS cavity, and the image acquisition points are evenly arranged along the radial direction of the GIS cavity.

7. The GIS cavity according to claim 2, characterized in that: Each detection point includes one image acquisition point A and two image acquisition points B, and the optical fiber probe includes five multimode optical fibers facing five different directions.

8. A GIS internal fault prediction system, characterized by: The invention comprises a GIS cavity, an imaging device and a data center, wherein the GIS cavity is the GIS cavity as described in any one of claims 1 to 7, the optical fiber connector of the imaging device is connected to the optical fiber connector of the image acquisition point A, the imaging device acquires ultraviolet images, infrared images and visible light images inside the GIS, and after completing the image acquisition, transmits the images to the data center via a wireless network, the data center reconstructs the received multiple multi-angle two-dimensional images into three-dimensional images via an image reconstruction algorithm, and then uses a feature extraction algorithm to compare the reconstructed three-dimensional images with typical fault images inside the GIS obtained in advance through experiments, so as to analyze the probability of faults.

9. A GIS internal fault prediction method, characterized by: This method is implemented based on the GIS cavity according to any one of claims 1 to 7, and comprises the following steps: S01. Connect the optical fiber connector of the imaging device to the optical fiber connector of the image acquisition point A. The imaging device first directly acquires the ultraviolet image and the infrared image, then emits visible light into the GIS cavity through one of the two visible optical fibers, and then acquires the visible light image inside the GIS cavity through the other visible optical fiber; S02. Before the imaging device is installed in the GIS cavity, the data center first obtains ultraviolet images, infrared images, and visible light images of typical faults inside the GIS, reconstructs the collected images through deep learning methods, generates a three-dimensional image of the inside of the GIS pipeline, and stores the generated image in the database; After the image acquisition is completed, the image is transmitted to the data center via a wireless network. The data center reconstructs the received multiple multi-angle two-dimensional images into three-dimensional images through an image reconstruction algorithm; The image reconstruction process is: In each image, a feature extraction algorithm is used to extract feature points in the image. Each feature point is represented by a position coordinate (x, y) and a direction descriptor. The feature point direction descriptor is calculated using the following formula: ， Where i represents the i-th feature point, n represents the number of feature points, represents the weight coefficient of the i-th feature point, Represents the gradient direction of the i-th feature point; Match the feature points in multiple images, and find the same feature point pairs under different perspectives through the nearest neighbor algorithm or the matching algorithm based on Euclidean distance. The same feature point pairs under different perspectives are called matching point pairs. The matching point pairs satisfy geometric consistency. The basic matrix F is calculated using the matching point pairs, and combined with the intrinsic parameter matrix K of the imaging camera, the essential matrix E is obtained: ， in represents the intrinsic parameter matrix of camera one, represents the intrinsic parameter matrix of camera 2, The essential matrix describes the relative position of a single fiber imaging point in three-dimensional space. The essential matrix E is subjected to singular value decomposition to extract the rotation matrix R and translation vector T of a single fiber imaging point, and the spatial coordinates of the fiber imaging point are obtained. The solution that meets geometric consistency is screened out through triangulation. Combined with the imaging device posture, the three-dimensional coordinates of the matching point pairs are calculated using triangulation. Based on the three-dimensional coordinates of the matching point pairs, a multi-view stereo vision method is used to generate a dense point cloud. The dense point cloud is fitted with a Poisson surface reconstruction algorithm to obtain a complete three-dimensional model. S03, the image collected by the imaging device in real time is reconstructed in three dimensions and then compared with the three-dimensional image in the database. The comparison process is as follows: Determine the neighborhood of each feature point in the image. Each neighboring point of , calculate the geometric relationship between point pairs, first calculate the point and Point The normal vector and , calculate the point pair The geometric features between include three key features: Angle ,point point to The vector and The normal vector The angle is calculated as: ， Angle ,point The normal vector and Point The normal vector The angle between them is calculated as: ， Angle ,point The normal vector and Point point to The angle between the vectors is calculated as: ， in , Respectively indicate points , The spatial coordinates of , Respectively indicate points , The normal vector of represents the norm of the orientation quantity; The above three angles are divided into multiple intervals, and the characteristic values ​​of each point pair are accumulated in the form of a histogram to form the SPFH feature. The FPFH is calculated by weighted summation based on the SPFH features of other points in the neighborhood. The calculation formula is: ， in represents the number of points in the neighborhood, Indicate point and Point The distance between The FPFH of each point is stored as the feature information of the feature point. By comparing the feature point information of the image obtained in the test with the feature point information obtained by imaging inside the actual GIS, it is possible to analyze whether a fault occurs inside the actual GIS or predict possible faults.

10. The GIS internal fault prediction method according to claim 9, characterized in that: In step S02, the simulation of different faults includes local corona discharge simulation, local overheating fault simulation and simulation of foreign matter and physical defects inside the GIS pipeline. The local corona discharge simulation collects local discharge images through ultraviolet light imaging, the local overheating fault simulation collects local overheating images through infrared light imaging, and the simulation of foreign matter and physical defects inside the GIS pipeline collects internal visible light images through visible light imaging.

Images