Partial discharge detection sensor layout method, system and storage medium for high-voltage equipment

Abstract

Embodiments of the present invention provide a method, system, and storage medium for locating partial discharge detection sensors for high-voltage equipment. The method includes obtaining type and model information of multiple high-voltage equipment in a target area; determining radiation area data and a first installation area corresponding to each type and model information of the high-voltage equipment based on the type and model information and location information; the first installation area is an area where sensors can be installed, determined based on historical false detection data of the high-voltage equipment; obtaining layout information of partial discharge detection sensors based on the first installation area corresponding to the target high-voltage equipment and historical defect information and radiation area data corresponding to the target high-voltage equipment; and arranging and installing partial discharge detection sensors on a digital twin model of the target high-voltage equipment based on the layout information. This method can effectively improve the layout position accuracy of the partial discharge detection sensors, thereby improving the accuracy of partial discharge detection.

Description

Technical Field

[0001] The present invention relates to the field of, but is not limited to, power detection technology, and in particular to a method, system, and storage medium for arranging partial discharge detection sensors for high-voltage equipment. Background Art

[0002] The impact of partial discharge (PD) on high-voltage equipment is both progressive and destructive, requiring comprehensive prevention and control through online monitoring technologies, regular insulation testing, and environmental control. Early detection and treatment of PD issues are key to avoiding serious consequences, placing high demands on the accuracy of PD detection on high-voltage equipment. Currently, the installation of PD sensors on high-voltage equipment is primarily based on the experience of on-site installation engineers. Due to the varying skills of on-site installation engineers and the complex layout of high-voltage equipment on site, PD sensors installed on high-voltage equipment often experience misjudgments. Summary of the Invention

[0003] The following is a summary of the subject matter described in detail herein. This summary is not intended to limit the scope of the claims.

[0004] The main purpose of the embodiments of the present invention is to propose a method, system and storage medium for PD detection sensor layout of high-voltage equipment, which can improve the layout position accuracy of PD detection sensors, thereby improving the correctness of PD detection.

[0005] In a first aspect, an embodiment of the present invention provides a method for distributing partial discharge detection sensors for high-voltage equipment, which is applied to a power grid management system provided with digital twin models of multiple high-voltage equipment. The method includes:

[0006] Acquire type, model, and location information of a plurality of high-voltage devices in a target area;

[0007] Determine, based on each of the type and model information and the location information, the radiation area data and the first installation area of ​​the high-voltage device corresponding to the type and model information, wherein the first installation area is an area where a sensor can be installed, determined based on historical false detection data of the high-voltage device.

[0008] Based on the first installation area corresponding to the target high-voltage equipment, and according to the historical defect information corresponding to the target high-voltage equipment and the radiation area data corresponding to the other plurality of non-target high-voltage equipment, layout information of the partial discharge detection sensors is obtained;

[0009] The partial discharge detection sensors are arranged on the digital twin model of the target installed high-voltage equipment according to the layout information.

[0010] In some optional embodiments, obtaining the layout information of the partial discharge detection sensor based on the first installation area corresponding to the target high-voltage equipment and the radiation area data corresponding to multiple other non-target high-voltage equipment includes:

[0011] The first installation area, the historical defect information, and the radiation area data are input into a trained partial discharge detection sensor layout model for multimodal candidate scoring processing, and the layout information of the partial discharge detection sensor is output. The partial discharge detection sensor layout model includes an input coding layer, a multimodal fusion layer, and a spatial optimization layer.

[0012] In some optional embodiments, inputting the first installation area, the historical defect information, and the radiation area data into a trained partial discharge detection sensor layout model for multimodal candidate scoring processing, and outputting the layout information of the partial discharge detection sensor, includes:

[0013] Inputting the first installation area, the historical defect information, and the radiation area data into the input coding layer, and encoding the first installation area, the historical defect information, and the radiation area data respectively through the input coding layer to obtain a regional feature map, a defect feature vector, and a radiation feature vector;

[0014] Performing cross-modal attention fusion processing on the regional feature map, the defect feature vector, and the radiation feature vector through a multimodal fusion layer to obtain a fused feature map;

[0015] The spatial optimization layer performs candidate scoring processing on the fused feature map and the first installation area to obtain layout information of the partial discharge detection sensor.

[0016] In some optional embodiments, the spatial optimization layer includes a candidate point generation sublayer, a layout scoring sublayer, and a constraint optimization sublayer. The spatial optimization layer performs candidate scoring processing on the fused feature map and the first installation area to obtain layout information of the partial discharge detection sensor, including:

[0017] Perform candidate processing on the first installation area through the candidate point generation sublayer to obtain candidate point coordinates;

[0018] Performing feature extraction and scoring processing on the fused feature map and the candidate point coordinates through the layout scoring sublayer to obtain candidate point position scores and sensor type probabilities;

[0019] The constraint optimization sublayer performs radiation constraint processing on the candidate point coordinates, the candidate point position scores, and the sensor type probability to obtain layout information of the partial discharge detection sensor.

[0020] In some optional embodiments, obtaining the layout information of the partial discharge detection sensor based on the first installation area corresponding to the target high-voltage equipment and the radiation area data corresponding to multiple other non-target high-voltage equipment includes:

[0021] Determining a non-installation area in a digital twin model of the target-installed high-voltage equipment based on the radiation area data of a plurality of non-target-installed high-voltage equipment;

[0022] removing the first installation area according to the non-installation area to obtain a second installation area;

[0023] Layout information of the partial discharge detection sensors is determined according to the second installation area and the historical defect information, where the layout information of the partial discharge detection sensors includes the type and quantity of the partial discharge detection sensors.

[0024] In some optional embodiments, the method for generating the first installation area includes:

[0025] Acquire historical false detection data corresponding to the high-voltage device, the historical false detection data including sensor position information of the partial discharge detection sensor set on the high-voltage device and false detection cause information of the historical records corresponding to the partial discharge detection sensor;

[0026] A first installation area is determined on the digital twin model of the high-voltage equipment of the target installation according to the sensor position information, the misdetection cause information and the preset installation position information.

[0027] In some optional embodiments, determining a first installation area on the digital twin model of the target installed high-voltage equipment based on the sensor position information, the misdetection cause information, and the preset installation position information includes:

[0028] When the misdetection cause information is that first data detected by the partial discharge detection sensor corresponding to the sensor position information does not meet the partial discharge positioning calculation requirements, and second data detected by other partial discharge detection sensors meet the partial discharge positioning calculation requirements, performing area optimization processing based on the preset installation position information and the sensor position information, and determining a first installation area on the digital twin model of the target installed high-voltage equipment;

[0029] or,

[0030] When the misdetection cause information is that third data from multiple detections of the same partial discharge detection sensor corresponding to the sensor position information is erroneous, performing area optimization processing based on the preset installation position information and the sensor position information, and determining a first installation area on the digital twin model of the target high-voltage equipment;

[0031] or,

[0032] When the misdetection cause information is that in the sensor position information, there are multiple different types of partial discharge detection sensors and the fourth data detected do not meet the partial discharge positioning calculation requirements, regional optimization processing is performed according to the preset installation position information and the sensor position information, and a first installation area is determined on the digital twin model of the high-voltage equipment of the target installation.

[0033] In some optional embodiments, determining the radiation area data and the first installation area of ​​the high-voltage equipment corresponding to each type and model information according to the type and model information and the location information includes:

[0034] Searching a preset high-voltage equipment radiation data table according to each type and model information to determine the radiation range data of the high-voltage equipment corresponding to the type and model information, wherein the high-voltage equipment radiation data table is obtained by detecting the electromagnetic field radiation of the high-voltage equipment, and the high-voltage equipment radiation data table includes power frequency electromagnetic field data, radio frequency radiation data, and thermal radiation data;

[0035] Determining the radiation area data of the high-voltage equipment corresponding to the type and model information according to the radiation range data and the location information;

[0036] According to each of the type and model information, a search is performed in the device information table to determine the first installation area corresponding to the type and model information.

[0037] In some optional embodiments, the method for training the partial discharge detection sensor layout model includes:

[0038] Obtaining a sensor layout training text and a sensor layout true label corresponding to the sensor layout training text, wherein the sensor layout training text includes the first installation area, the historical defect information, and the radiation area data;

[0039] Inputting the sensor layout training text into an initial partial discharge detection sensor layout model, wherein the partial discharge detection sensor layout model includes an input encoding layer, a multimodal fusion layer, and a spatial optimization layer;

[0040] Encoding the first installation area, the historical defect information, and the radiation area data respectively through the input coding layer to obtain a regional feature map, a defect feature vector, and a radiation feature vector;

[0041] Performing cross-modal attention fusion processing on the regional feature map, the defect feature vector, and the radiation feature vector through a multimodal fusion layer to obtain a fused feature map;

[0042] Performing candidate scoring processing on the fused feature map and the first installation area through the spatial optimization layer to obtain layout information of the partial discharge detection sensor;

[0043] A multi-task loss value is calculated based on the layout information of the partial discharge detection sensor and the true label of the sensor layout to obtain a multi-task total loss value. The partial discharge detection sensor layout model is iteratively updated based on the multi-task total loss value to obtain a trained partial discharge detection sensor layout model.

[0044] In some optional embodiments, the spatial optimization layer includes a candidate point generation sublayer, a layout scoring sublayer, and a constraint optimization sublayer. The spatial optimization layer performs candidate scoring processing on the fused feature map and the first installation area to obtain layout information of the partial discharge detection sensor, including:

[0045] Perform candidate processing on the first installation area through the candidate point generation sublayer to obtain candidate point coordinates;

[0046] Performing feature extraction and scoring processing on the fused feature map and the candidate point coordinates through the layout scoring sublayer to obtain candidate point position scores and sensor type probabilities;

[0047] Performing radiation constraint processing on the candidate point coordinates, the candidate point position scores, and the sensor type probabilities through the constraint optimization sublayer to obtain layout information of the partial discharge detection sensor, where the layout information of the partial discharge detection sensor includes sensor position information, sensor type information, and sensor position score information;

[0048] The multi-task loss value calculation based on the layout information of the partial discharge detection sensor and the true label of the sensor layout to obtain the multi-task total loss value includes:

[0049] Performing loss value calculation on the sensor position information and the regional feature map to obtain a sensor position loss value;

[0050] The sensor type information and the real sensor position information in the real sensor layout label are used to calculate the type loss value to obtain the sensor type loss value;

[0051] Performing radiation constraint loss calculation on the sensor position score information and the radiation feature vector to obtain a position score loss value;

[0052] The sensor position loss value, the sensor type loss value, and the position score loss value are subjected to a total loss calculation to obtain a multi-task total loss value.

[0053] In some optional embodiments, determining the second installation area according to the first installation area corresponding to the target installed high-voltage equipment and the other plurality of radiation area data of the non-target installed high-voltage equipment includes:

[0054] Rendering a first installation area corresponding to the target installed high-voltage device, and displaying the first installation area on a digital twin model of the target installed high-voltage device;

[0055] Rendering the data of the other multiple radiation areas of the non-target installed high-voltage equipment, and displaying the radiation areas on the digital twin models of the multiple target installed high-voltage equipment respectively;

[0056] determining an overlapping portion of the plurality of radiation areas and the first installation area as a non-installation area;

[0057] The first installation area is removed according to the non-installation area to obtain a second installation area.

[0058] In a second aspect, an embodiment of the present invention provides a controller comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for arranging partial discharge detection sensors for high-voltage equipment as described in the first aspect is implemented.

[0059] In a third aspect, an embodiment of the present invention provides a power grid management system, including the controller involved in the second aspect above.

[0060] In a fourth aspect, a computer storage medium stores computer-executable instructions, wherein the computer-executable instructions are used to execute the method for arranging partial discharge detection sensors for high-voltage equipment described in the first aspect.

[0061] The beneficial effects of the present invention include: the present invention obtains the type, model and location information of multiple high-voltage devices in the target area to determine the layout of the high-voltage devices in the target area. The radiation situation between the high-voltage devices can be subsequently determined based on the layout of the high-voltage devices. Then, the radiation area data and the first installation area of ​​the high-voltage device corresponding to the type and model information are determined based on each type, model and location information. The first installation area is the area where the sensor can be installed determined based on the historical false detection data of the high-voltage device. The radiation area data and the data of the first installation area of ​​all high-voltage devices in the target area are obtained to prepare for the subsequent calculation of the layout information of the partial discharge detection sensor of the target high-voltage device. Since the first installation area is determined by the historical false detection data, some location information that is not suitable for installing a certain type of sensor can be eliminated. Therefore, when the user determines that a new or modified high-voltage device needs to be added, the layout calculation of the partial discharge detection sensor for the high-voltage device needs to be performed. At this time, the system obtains the layout information of the partial discharge detection sensor based on the first installation area corresponding to the target installed high-voltage device, the historical defect information corresponding to the target installed high-voltage device and the radiation area data corresponding to multiple other non-target installed high-voltage devices. Then, according to the layout information, the partial discharge detection sensor is arranged on the digital twin model of the target installed high-voltage device. The embodiment of the present technical solution preliminarily locates the installation position of the target high-voltage equipment where the sensor is installed through historical false detection data to obtain a first installation area, and then performs secondary positioning processing on the radiation area data brought by the high-voltage equipment near the target high-voltage equipment where the sensor is installed and the historical defect information of the target high-voltage equipment, and finally determines the layout information of the partial discharge detection sensor. That is, the layout information of the partial discharge detection sensor is determined by fully considering the historical situation of the high-voltage equipment where the sensor is installed and the radiation situation of the high-voltage equipment near it, which can effectively improve the layout position accuracy of the partial discharge detection sensor, thereby improving the correctness of partial discharge detection.

[0062] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present invention. The purposes and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 This is a flowchart of the steps of a method for arranging partial discharge detection sensors for high-voltage equipment provided by an embodiment of the present invention;

[0064] Figure 2 2 is a schematic diagram of a layout model of a partial discharge detection sensor provided by an embodiment of the present invention;

[0065] Figure 3is a schematic diagram of a training method for a partial discharge detection sensor layout model provided by an embodiment of the present invention;

[0066] Figure 4 This is a schematic diagram of a controller provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0067] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0068] It should be noted that although the device schematics illustrate functional module divisions and the flowcharts illustrate logical sequences, in certain circumstances, the steps shown or described may be performed in a sequence that differs from the module divisions in the device or the sequence in the flowcharts. The terms "first," "second," and the like in the specification, claims, or accompanying drawings are used to distinguish similar items and are not necessarily used to describe a specific sequence or precedence.

[0069] The impact of partial discharge (PD) on high-voltage equipment is both progressive and destructive, requiring comprehensive prevention and control through online monitoring technologies, regular insulation testing, and environmental control. Early detection and treatment of PD issues are key to avoiding serious consequences, placing high demands on the accuracy of PD detection on high-voltage equipment. Currently, the installation of PD sensors on high-voltage equipment is primarily based on the experience of on-site installation engineers. Due to the varying skills of on-site installation engineers and the complex layout of high-voltage equipment on site, PD sensors installed on high-voltage equipment often experience misjudgments.

[0070] To solve the above-mentioned problems, the present application provides a method, system and storage medium for arranging partial discharge detection sensors for high-voltage equipment.

[0071] In the present application, a method, system and storage medium for arranging partial discharge detection sensors for high-voltage equipment are provided, which are described in detail one by one in the following embodiments.

[0072] like Figure 1As shown, an embodiment of the present invention provides a method for arranging partial discharge detection sensors of high-voltage equipment, which is applied to a power grid management system, wherein the power grid management system is provided with a digital twin model of the high-voltage equipment, and the power grid management system is provided with a partial discharge detection sensor layout interface on which multiple partial discharge sensors are arranged on the digital twin model of the high-voltage equipment. Through the partial discharge detection sensor layout interface, a comprehensive judgment can be made based on the radiation conditions of other high-voltage equipment in the environment in which the target high-voltage equipment is located and the detection conditions of the historical sensor layout of the target high-voltage equipment, thereby producing a relatively reasonable sensor layout scheme, and respectively arranging multiple partial discharge sensors at different positions on the digital twin model of the high-voltage equipment to improve the layout position accuracy of the partial discharge detection sensors, thereby improving the correctness of partial discharge detection.

[0073] It should be noted that the partial discharge sensor can be set inside or outside the digital twin model of the high-voltage equipment, and this embodiment does not specifically limit it.

[0074] It should be noted that the multiple partial discharge sensors may be evenly distributed or distributed in different regions according to the actual layout requirements, which is not specifically limited in this embodiment.

[0075] Specifically, the features of a digital twin model for high-voltage equipment may include: a 1:1 3D model of the high-voltage equipment's exterior, capable of being displayed in 3D within the software; for any coordinate point within the model's 3D scene, the model can independently determine whether the point is inside or outside the high-voltage equipment; and for any coordinate point within the high-voltage equipment, the model can independently determine which high-voltage component is located within the high-voltage equipment at that location. Modeling methods may include using 3D modeling tools (such as SolidWorks) to create a 3D model of the high-voltage equipment based on drawings or on-site measurements. During modeling, the high-voltage equipment's housing, outer surface of the pipeline, or casing is modeled to a thickness of 0.001 meters, and the interior of the pipeline also needs to be modeled. The 3D model of the high-voltage equipment is spatially digitized.

[0076] It should be noted that the high-voltage equipment may be a gas-insulated switchgear, a transformer, or a high-voltage distribution cabinet, etc., and this embodiment does not specifically limit it.

[0077] Partial discharge monitoring for high-voltage equipment utilizes transformer partial discharge monitoring and automatic location technology, based on intelligent, simultaneous data collection from multiple sensors. By integrating data from ultra-high frequency (UHF), high frequency (HF), and ultrasonic sensors to construct a digital twin model of the transformer and build a multi-sensor time difference of arrival fingerprint library, this technology enables simultaneous electrical-electrical, acoustic-electrical, and acoustic-acoustic monitoring, optimizing partial discharge location results. Enhanced algorithms for partial discharge monitoring and judgment enable a variety of sensor layouts for high-voltage equipment, extending beyond pure ultrasonic or high-frequency / UHF sensor deployments.

[0078] It should be noted that the UHF, HF, and ultrasonic synchronous monitoring technology primarily utilizes synchronous sampling technology to assign synchronized clocks with better than nanosecond accuracy to each UHF, HF, and ultrasonic monitoring channel within the monitoring device. This ensures synchronized timestamps on the signal waveforms detected by different monitoring channels, enabling precise comparison of waveform arrival times and meeting the prerequisite for accurate positioning. Furthermore, by collecting data on different types of partial discharge patterns detected by different types of sensors, the data characteristics of abnormal signals in different frequency bands generated by partial discharge defects within various transformers are obtained, allowing for multi-source fusion identification of partial discharge types. Utilizing a fully automatic intelligent positioning algorithm based on UHF "electric-electric" positioning, ultrasonic and high-frequency "acoustic-electric" positioning, and ultrasonic "acoustic-acoustic" positioning, when multiple sensors detect partial discharge signals, the algorithm can simultaneously perform online automatic positioning based on multiple principles. By comparing the positioning accuracy of UHF "electric-electric," ultrasonic and high-frequency "acoustic-electric," and ultrasonic "acoustic-acoustic" positioning for different defect types, the algorithm automatically determines which principle's positioning result should be used as the final positioning result after the defect type is successfully identified.

[0079] Applied to the above-mentioned power grid management system, the method for arranging partial discharge detection sensors for high-voltage equipment includes:

[0080] S100: Acquire type, model, and location information of multiple high-voltage devices in a target area.

[0081] Specifically, in the same distribution room in the power grid management system, various high-voltage equipment, such as transformers, high-voltage distribution cabinets, high-voltage switches, and gas-insulated switchgear, may be installed. Therefore, when a certain high-voltage equipment is updated, replaced, or added, or the radiation area of ​​the newly added or changed high-voltage equipment changes, it may affect the sensor detection of other high-voltage equipment in the distribution room. Therefore, it is necessary to first obtain the type and model information of multiple high-voltage equipment in the target area. Due to the different structures in the distribution room, the types and locations of the high-voltage equipment arranged there are different. Therefore, it is necessary to obtain the type, model, and location information of multiple high-voltage equipment in the target area near the high-voltage equipment where the sensor is installed or changed.

[0082] S200, determining the radiation area data and the first installation area of ​​the high-voltage equipment corresponding to each type and model information according to each type and model information and location information, wherein the first installation area is an area where the sensor can be installed determined according to historical false detection data of the high-voltage equipment.

[0083] Specifically, the radiation range data of different types and models of high-voltage equipment are all different. The radiation range data of high-voltage equipment is obtained by engineers or through measurement robots performing radiation range data detection on the high-voltage equipment in operation. A data table corresponding to the radiation range data is stored in the power grid management system, and the data table corresponding to the radiation range data is preset in the power grid management system. In addition, the power grid management system also stores equipment information tables for various high-voltage equipment. The information table also provides information on the first installation area where the partial discharge detection sensor is recommended to be installed. Then, combined with the location information of each high-voltage equipment, the radiation range data of a certain high-voltage equipment can be obtained relative to the radiation range data of other high-voltage equipment. After setting up multiple digital twin models of high-voltage equipment, the radiation range data can be generated based on the radiation range data on the digital twin model of the high-voltage equipment corresponding to the radiation range data. A radiation ring image is generated and displayed in a fusion manner with the digital twin model of the high-voltage equipment. Then, for other high-voltage equipment within the radiation range of the radiation ring image of the digital twin model of the high-voltage equipment, the radiation range data of the high-voltage equipment relative to other high-voltage equipment can be obtained.

[0084] It should be noted that if a high-voltage device is located at the same location point within the radiation range of multiple other high-voltage devices, the radiation area data of the location point can be generated based on the superposition of the radiation range data of multiple other high-voltage devices, or the highest value of the radiation range data of multiple other high-voltage devices can be taken as the radiation area data. This embodiment does not make any specific limitations on this.

[0085] It should be noted that different measuring instruments can be selected according to different radiation types. For example, for industrial frequency electromagnetic field measurements, an electric field strength meter for measuring electric field strength or a magnetic field strength meter for measuring magnetic induction strength can be used; for radio frequency radiation measurements, a spectrum analyzer for analyzing the frequency and intensity of high-frequency electromagnetic waves or a radio frequency field strength meter for measuring radio frequency radiation power density can be used; for thermal radiation measurements, an infrared thermal imager for detecting the surface temperature distribution of the equipment or a temperature sensor for contact temperature measurement can be used.

[0086] In some optional embodiments, step S200 may include searching from a preset high-voltage equipment radiation data table according to each type and model information to determine the radiation range data of the high-voltage equipment corresponding to the type and model information. The high-voltage equipment radiation data table is obtained by detecting the electromagnetic field radiation conditions of the high-voltage equipment. The high-voltage equipment radiation data table includes industrial frequency electromagnetic field data, radio frequency radiation data and thermal radiation data. The radiation area data of the high-voltage equipment corresponding to the type and model information is then determined based on the radiation range data and position information; searching from the equipment information table according to each type and model information to determine the first installation area corresponding to the type and model information.

[0087] Among them, the data of the first installation area can also be based on the suggestions in the equipment information table. During the daily operation of a large number of high-voltage equipment, the data detected by the partial discharge detection sensor may contain false detection data, that is, historical false detection data. The data of the first installation area can also be further corrected according to the frequency and deviation of such data to obtain new data of the first installation area, and the data can be updated in the equipment information table.

[0088] It should be noted that the method for generating the first installation area may include: obtaining historical false detection data corresponding to the high-voltage equipment, the historical false detection data including the sensor position information of the partial discharge detection sensor set on the high-voltage equipment and the false detection cause information of the historical records corresponding to the partial discharge detection sensor, and then determining the first installation area based on the sensor position information, the false detection cause information and the position information of the high-voltage equipment.

[0089] In some optional embodiments, when performing distance conversion on partial discharge signals of high-voltage equipment obtained by multiple partial discharge detection sensors of different or same types, when the deviation between the distance between the partial discharge signal calculated by a certain partial discharge detection sensor and the distance between the partial discharge signal calculated by other partial discharge detection sensors and the partial discharge detection sensor exceeds 20% during the positioning judgment, then the data detected by the partial discharge detection sensor this time can be determined as erroneous detection data and recorded; then, if the partial discharge detection sensor continues to have erroneous detection data in the future, the area covered by the partial discharge detection sensor and the nearby area can be defined as areas requiring correction and eliminated from the original first installation area.

[0090] It should be noted that, because different types of partial discharge detection sensors have different anti-interference capabilities and abilities to monitor partial discharge signals, the first installation areas corresponding to different types of partial discharge detection sensors may be different. However, considering the complexity of data storage and data operations, the first installation areas corresponding to different types of partial discharge detection sensors may also be set to be the same.

[0091] In some optional embodiments, when the misdetection cause information is that the first data detected by the partial discharge detection sensor corresponding to the sensor position information does not meet the partial discharge positioning calculation requirements, and the second data detected by other partial discharge detection sensors meets the partial discharge positioning calculation requirements, area optimization processing is performed according to the preset installation position information and the sensor position information to determine the first installation area.

[0092] In some optional embodiments, when the misdetection cause information is that the third data of multiple detections of the same partial discharge detection sensor corresponding to the sensor position information is erroneous, area optimization processing is performed based on the preset installation position information and the sensor position information to determine the first installation area.

[0093] In some optional embodiments, when the reason for the misdetection is that in the sensor position information, there are multiple different types of partial discharge detection sensors and the fourth data detected do not meet the partial discharge positioning calculation requirements, regional optimization processing is performed based on the preset installation position information and the sensor position information to determine the first installation area.

[0094] S300 , based on a first installation area corresponding to a target installed high-voltage device, and according to historical defect information corresponding to the target installed high-voltage device and radiation area data corresponding to multiple other non-target installed high-voltage devices, obtain layout information of partial discharge detection sensors.

[0095] Specifically, the first installation area corresponding to the target installed high-voltage equipment is the location information where the sensor is recommended or can be installed. Then, the first installation area can be used to eliminate sensor layout design problems existing in some high-voltage equipment that has historically been misdetected by sensors. Then, based on the historical defect information corresponding to the target installed high-voltage equipment and the radiation area data brought by other high-voltage equipment near the installation site, the specific situation on site is further constrained, thereby effectively improving the accuracy of the layout information of the partial discharge detection sensor.

[0096] In some optional embodiments, a non-installation area is determined in a digital twin model of a target-installed high-voltage device based on radiation area data of multiple non-target-installed high-voltage devices; then, a first installation area is removed based on the non-installation area to obtain a second installation area; and then, layout information of the partial discharge detection sensor is determined based on the second installation area and historical defect information. The layout information of the partial discharge detection sensor includes the type and quantity of the partial discharge detection sensor.

[0097] Specifically, according to the radiation area data of multiple non-target installed high-voltage equipment, combined with the location information of the non-target installed high-voltage equipment and the target installed high-voltage equipment, the non-installation area can be determined in the digital twin model of the target installed high-voltage equipment. The non-installation area is the area that is relatively heavily radiated by the non-target installed high-voltage equipment. It can be understood that the area can be decomposed according to the radiation height in the digital twin model of the non-target installed high-voltage equipment based on the radiation area data, and displayed in a ring shape. For example, the high radiation area is a red ring area close to the center of the non-target installed high-voltage equipment, the medium radiation area is a yellow ring area outside the high radiation area, and the low radiation area is a green ring area outside the medium radiation area. Different types of partial discharge detection sensors have different anti-interference capabilities. Different types of partial discharge detection sensors The corresponding acceptable radiation areas are different. Then, the area overlapping with the non-target installed high-voltage equipment in the digital twin model in the first installation area according to the radiation area can be determined as the radiation situation received by the partial discharge detection sensor, so that the non-installation area can be determined according to the type of partial discharge detection sensor required in the area. The first installation area will then be removed according to the non-installation area, that is, the non-installation area is removed to obtain the second installation area. The layout information of the partial discharge detection sensor is determined based on the second installation area and the historical defect information. The layout information of the partial discharge detection sensor includes the type and quantity of the partial discharge detection sensor. For the second installation area where there is historical defect information and the distance meets the first distance threshold, the number or type of partial discharge detection sensors can be appropriately increased to solve the inaccuracy problem caused by radiation.

[0098] In some optional embodiments, the first installation area corresponding to the target installed high-voltage equipment is first rendered, and the first installation area is displayed in the digital twin model of the target installed high-voltage equipment; then the other multiple radiation area data of the non-target installed high-voltage equipment are rendered, and the radiation areas are respectively displayed in the digital twin models of the multiple target installed high-voltage equipment; then the overlapping parts of the multiple radiation areas and the first installation area are determined as non-installation areas; the first installation area is removed according to the non-installation area to obtain the second installation area.

[0099] In some optional embodiments, referring to Figure 2 , the first installation area, historical defect information and radiation area data are input into the trained partial discharge detection sensor layout model for multimodal candidate scoring processing, and the layout information of the partial discharge detection sensor is output. The partial discharge detection sensor layout model includes an input coding layer 210, a multimodal fusion layer 220 and a spatial optimization layer 230, wherein the historical defect information is the defect information caused by aging or damage of the target high-voltage equipment. The defect information of the high-voltage equipment will be reflected in the form of partial discharge, that is, the location where the partial discharge exists is associated with the location corresponding to the defect information, and different types of defect information have different corresponding position correlations.

[0100] Specifically, the first installation area, historical defect information and radiation area data are input into the trained partial discharge detection sensor layout model for multimodal candidate scoring processing, and the layout information of the partial discharge detection sensor is output, including: first inputting the first installation area, historical defect information and radiation area data into the input coding layer 210, and encoding the first installation area, historical defect information and radiation area data respectively through the input coding layer 210 to obtain a regional feature map, a defect feature vector and a radiation feature vector; then performing cross-modal attention fusion processing on the regional feature map, the defect feature vector and the radiation feature vector through the multimodal fusion layer 220 to obtain a fused feature map; and then performing candidate scoring processing on the fused feature map and the first installation area through the spatial optimization layer 230 to obtain the layout information of the partial discharge detection sensor.

[0101] It should be noted that the partial discharge detection sensor layout model in this embodiment is a multimodal graph neural network, and the input data includes the first installation area, historical defect information, and radiation area data, wherein the historical defect information is represented by a two-dimensional heat map, which is used to represent the probability or frequency of defects at each location of the equipment. The first installation area; the radiation area data is represented by a graph structure, the nodes include the target high-voltage equipment and other surrounding high-voltage equipment, and the edges represent the radiation impact with intensity weights. In addition, the radiation area corresponding to each radiation source device can be represented by polygonal coordinates, but when input into the model, it can be converted into a radiation heat map of the same size as the defect heat map, where each pixel value represents the radiation intensity received at that location, or the graph structure data, node features, and adjacency matrix can be directly used; the first installation area is usually a binary mask with a size similar to the defect heat map of the historical defect information, 1 indicates that it can be installed, and 0 indicates that it cannot be installed.

[0102] The input coding layer 210 includes three encoders: a defect heat map encoder 211, a radiation map encoder 212, and an installation area encoder 213. The defect heat map encoder 211 uses a convolutional neural network to extract features from historical defect information. By setting several convolutional layers and pooling layers, it finally flattens or uses global average pooling to obtain a defect feature vector of the historical defect information. The radiation map encoder 212 superimposes the radiation areas of the high-voltage equipment of each radiation source to generate a radiation heat map, and then uses a convolutional neural network to extract the radiation feature vector of the radiation area data. The installation area encoder 213 also uses a convolutional neural network to extract features to obtain a regional feature map. The multimodal fusion layer 220 is used to fuse the regional feature map, the defect feature vector, and the radiation feature vector. The regional feature map, the defect feature vector, and the radiation feature vector are first spliced ​​together, and then a fully connected layer is used to reduce the dimension or integrate the features of the spliced ​​data to obtain a fused feature map. The spatial optimization layer 230 may include a candidate point generation sublayer 231, a layout scoring sublayer 232, and a constraint optimization sublayer 233. The spatial optimization layer 230 then performs candidate scoring processing on the fused feature map and the first installation area to obtain layout information for the partial discharge detection sensor, including: first, performing candidate processing on the first installation area through the candidate point generation sublayer 231 to obtain candidate point coordinates; then, performing feature extraction and scoring processing on the fused feature map and candidate point coordinates through the layout scoring sublayer 232 to obtain candidate point location scores and sensor type probabilities; and then, performing radiation constraint processing on the candidate point coordinates, candidate point location scores, and sensor type probabilities through the constraint optimization sublayer 233 to obtain layout information for the partial discharge detection sensor. The neural network model of the technical solution of this embodiment achieves an anti-interference optimized layout in a complex multi-device environment by integrating the radiation impact relationship and historical defect conditions between devices, while satisfying engineering installation constraints, effectively improving the layout position accuracy of the partial discharge detection sensor, and thereby improving the accuracy of partial discharge detection.

[0103] In some optional embodiments, the partial discharge detection sensor layout model includes an input encoding layer 210, a multimodal fusion layer 220, and a spatial optimization layer 230. The spatial optimization layer 230 includes a candidate point generation sublayer 231, a layout scoring sublayer 232, and a constraint optimization sublayer 233. The input data of the partial discharge detection sensor layout model include: defect heat map (matrix [H, W]); radiation data (which can be a heat map [H, W] or a graph structure (node ​​feature matrix and adjacency matrix); and installable area (binary mask [H, W]. The output data is sensor layout information: a list containing K elements, each element is a tuple (x, y, sensor_type), indicating the type and layout location of each sensor.

[0104] The model data processing for the partial discharge detection sensor layout model is as follows: defect heat map -> CNN encoder -> defect feature vector; radiation data -> map encoder -> radiation feature vector; installation area -> CNN encoder -> installable feature vector. Next, the three feature vectors are concatenated through a multimodal fusion layer 220 -> fused feature vector; the fused feature vector plus candidate point position encoding -> fully connected network (weight sharing) -> score and type probability for each candidate point; the candidate point with the highest score is then selected (constraints taken into account) and the sensor layout information is output.

[0105] It should be noted that in actual projects, the installable area may already include information such as safe distances. Therefore, when generating candidate points, we ensure that the distance between them is greater than the minimum installation distance. This eliminates the need to consider distance constraints during scoring. The final selection is simply sorted by score, as the candidate points already meet the minimum distance.

[0106] It's important to note that before inputting into the model, candidate points can be generated within the installable area. This can be done using a uniform grid sampling method (e.g., one point every 0.1 meter) or adaptive sampling based on the shape of the installable area (e.g., dense at the edges, sparse in the center). After candidate points are generated, their coordinates can be converted to positional encodings (e.g., normalized to 0-1 and then embedded as vectors).

[0107] In some optional embodiments, historical defect information [H, W]-->CNN (convolutional layer, pooling layer)-->global average pooling-->defect feature vector [D]; radiation area data: (node ​​feature matrix [N, F], adjacency matrix [N, N])-->GCN (2 layers)-->target device node embedding [D]; first installation area [H, W]-->CNN (simple layers)-->radiation feature vector [D]; three feature vectors are concatenated [3*D]-->fully connected layer (dimensionality reduced to D)-->fused feature map [D]; for each candidate point: candidate point coordinates (x, y)-->position encoding (fully connected layer)-->position feature vector [D]; fused feature vector [D] and position feature vector [D] are concatenated-->[2*D]-->fully connected layer-->output: installation probability (0 to 1) and type probability (4 categories); then the installation probability of each candidate point is sorted, and the K points with the highest probability are selected.

[0108] It should be noted that the model outputs the installation probability and type probability for each candidate point. A probability threshold (such as 0.5) is set to determine installation, but this may result in more than K points. Therefore, the K points with the highest probability are selected (K is the preset number of sensors). If there are fewer than K points, all are selected; if there are more than K points, the first K are selected. These K points are then checked to see if they meet the minimum distance constraint (if not, points with lower probabilities are removed until they do). Finally, the locations and types of these points are output.

[0109] In some optional embodiments, referring to Figure 3 , a training method for a partial discharge detection sensor layout model, including:

[0110] Step S310: Acquire a sensor layout training text and a sensor layout true label corresponding to the sensor layout training text, wherein the sensor layout training text includes a first installation area, historical defect information, and radiation area data;

[0111] Step S320: inputting the sensor layout training text into an initial partial discharge detection sensor layout model, where the partial discharge detection sensor layout model includes an input coding layer, a multimodal fusion layer, and a spatial optimization layer;

[0112] Step S321: encoding the first installation area, historical defect information, and radiation area data respectively through the input coding layer to obtain a regional feature map, a defect feature vector, and a radiation feature vector;

[0113] Step S322: performing cross-modal attention fusion processing on the regional feature map, the defect feature vector, and the radiation feature vector through a multimodal fusion layer to obtain a fused feature map;

[0114] Step S323 , performing candidate scoring processing on the fused feature map and the first installation area through the spatial optimization layer to obtain layout information of the partial discharge detection sensor;

[0115] In step S330 , a multi-task loss value is calculated based on the layout information of the partial discharge detection sensor and the true label of the sensor layout to obtain a multi-task total loss value, and the partial discharge detection sensor layout model is iteratively updated based on the multi-task total loss value to obtain a trained partial discharge detection sensor layout model.

[0116] Specifically, data preparation begins: primarily, three types of input data (the first installation area, historical defect information, and radiation area data) and label data (sensor layout location and type) are prepared. Three encoders (a defect heat map encoder, a radiation map encoder, and an installation area encoder) are used to encode the three input data sets. Multimodal fusion is then performed on the three encoded region feature maps, defect feature vectors, and radiation feature vectors. The spatial optimization layer then calculates the predicted partial discharge sensor layout information. Loss calculations are then performed on the partial discharge sensor layout information, including position loss, type loss, and radiation constraint loss. Backpropagation and optimization are performed based on the loss calculation results, and gradient descent is used to update the weights. This iterative update of the partial discharge sensor layout model is performed based on the total multi-task loss value, resulting in a trained partial discharge sensor layout model.

[0117] It should be noted that during training, labeled data is prepared, namely, the optimal sensor layout (location and type) for each high-voltage device. This labeled data can be a manually designed layout or a layout obtained through simulation. The labeled data takes the form of a binary label (whether a sensor is installed) and a sensor type label (if it is a positive example) for each candidate point. Alternatively, the layout problem can be formulated as a multi-label classification problem, predicting whether a sensor is installed and the type of sensor installed for each candidate point. Regarding the loss function, since each candidate point is independent, a binary cross-entropy loss (whether installed) or a multi-class cross-entropy loss (sensor type) can be used for each candidate point. Since only K sensors are required, a ranking loss (such as the TopK loss) or a global loss (such as the ensemble prediction loss) can be used. However, for simplicity, the installation decision for each candidate point can be treated as a binary classification, using a cross-entropy loss with weighting, which is suitable for situations with a small number of positive examples. For type prediction, the cross-entropy loss can be calculated only for positive examples (installation points). Constraint violation penalties can also be added (such as a penalty for sensors being too close to each other).

[0118] It's important to note that each candidate point in the training data has a label (0 or 1, indicating whether it's installed) and a sensor type (if installed). Therefore, two losses can be defined: installation loss (binary cross entropy or focal loss) and type loss (multi-class cross entropy (calculated only for installation points). The total loss = installation loss + α * type loss. Since the number of sensors K is fixed in practical deployments, it's possible to use K as an input and control the number in the model output. However, K is typically predetermined, so it can be fixed.

[0119] S400: Layout and set partial discharge detection sensors on the digital twin model of the target installed high-voltage equipment according to the layout information.

[0120] Specifically, partial discharge detection sensors are arranged on the digital twin model of the target installed high-voltage equipment according to the layout information. The display interface after layout can display the target installed high-voltage equipment, the partial discharge detection sensors set on the high-voltage equipment, and the radiation range map of the non-installed high-voltage equipment. Some partial discharge detection sensors set near the position between two radiation levels can be marked, and the changed or newly added partial discharge detection sensors can also be marked, so that the installation decision maker can intuitively see the situation of the changed partial discharge detection sensors, which can facilitate decision-making and modification.

[0121] It should be noted that the marked partial discharge detection sensor can also be marked with the sensor type and radiation resistance data, and other data can be added. The data can be displayed permanently or after selecting the icon corresponding to the partial discharge detection sensor. This embodiment does not make any specific restrictions on this.

[0122] The present invention obtains the type, model and location information of multiple high-voltage devices in the target area to determine the layout of the high-voltage devices in the target area. The radiation situation between the high-voltage devices can be subsequently determined based on the layout of the high-voltage devices. Then, the radiation area data and the first installation area of ​​the high-voltage device corresponding to the type and model information are determined according to each type, model and location information. The first installation area is the area where the sensor can be installed determined based on the historical false detection data of the high-voltage device. The radiation area data and the data of the first installation area of ​​all high-voltage devices in the target area are obtained for subsequent calculation of the layout information of the partial discharge detection sensor of the target high-voltage device in preparation. Since the first installation area is determined by historical false detection data, some location information that is not suitable for installing a certain type of sensor can be eliminated. Then, when the user determines that a new or modified high-voltage device needs to be added, the layout calculation of the partial discharge detection sensor for the high-voltage device needs to be performed. At this time, the system obtains the layout information of the partial discharge detection sensor based on the first installation area corresponding to the target installed high-voltage device, the historical defect information corresponding to the target installed high-voltage device and the radiation area data corresponding to multiple other non-target installed high-voltage devices. Then, according to the layout information, the partial discharge detection sensor is arranged on the digital twin model of the target installed high-voltage device. The embodiment of the present technical solution preliminarily locates the installation position of the target high-voltage equipment where the sensor is installed through historical false detection data to obtain a first installation area, and then performs secondary positioning processing on the radiation area data brought by the high-voltage equipment near the target high-voltage equipment where the sensor is installed and the historical defect information of the target high-voltage equipment, and finally determines the layout information of the partial discharge detection sensor. That is, the layout information of the partial discharge detection sensor is determined by fully considering the historical situation of the high-voltage equipment where the sensor is installed and the radiation situation of the high-voltage equipment near it, which can effectively improve the layout position accuracy of the partial discharge detection sensor, thereby improving the correctness of partial discharge detection.

[0123] As shown in Figure 4, Figure 4 1 shows a block diagram of a controller 1000 according to an embodiment of the present application. The components of the controller 1000 include, but are not limited to, a memory 1200 and a processor 1100. The processor 1100 and the memory 1200 are connected via a bus, and the memory 1200 is used to store data.

[0124] The controller 1000 also includes an access device that enables the controller 1000 to communicate via one or more networks. Examples of these networks include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or a combination of communication networks such as the Internet. The access device may include one or more of any type of network interface (e.g., a network interface card (NIC)) of wired or wireless, such as an IEEE 802.11 wireless local area network (WLAN) wireless interface, a World Wide Interoperability for Microwave Access (Wi-MAX) interface, an Ethernet interface, a universal serial bus (USB) interface, a cellular network interface, a Bluetooth interface, a near field communication (NFC) interface, and the like.

[0125] The controller 1000 can be any type of stationary or mobile electronic device, including a mobile computer or mobile electronic device (e.g., a tablet computer, a personal digital assistant, a laptop computer, a notebook computer, a netbook, etc.), a mobile phone (e.g., a smartphone), a wearable electronic device (e.g., a smart watch, smart glasses, etc.), or other types of mobile devices, or a stationary electronic device such as a desktop computer or a PC. The controller 1000 of the smart chair can also be a mobile or stationary server.

[0126] The processor 1100 is used to execute computer executable instructions of a method for arranging partial discharge detection sensors for high-voltage equipment.

[0127] The above is a schematic diagram of a computing device according to this embodiment. It should be noted that the technical solution of this computing device is based on the same concept as the technical solution of the method for arranging partial discharge detection sensors for high-voltage equipment described above. For details not described in detail in the technical solution of the computing device, please refer to the description of the technical solution of the method for arranging partial discharge detection sensors for high-voltage equipment described above.

[0128] According to one embodiment of the present application, a power grid management system is also provided. The power grid management system is equipped with a controller 1000, or the power grid management system and the controller 1000 are connected via communication, so that the power grid management system can locate partial discharge sources through the controller 1000. It should be noted that the technical solution of the computing device and the technical solution of the method for arranging partial discharge detection sensors for high-voltage equipment described above are based on the same concept. For details not described in detail in the technical solution of the computing device, please refer to the description of the technical solution of the method for arranging partial discharge detection sensors for high-voltage equipment described above.

[0129] An embodiment of the present application also provides a storage medium, which is a computer-readable storage medium and stores a computer program. When the computer program is executed by a processor, the above-mentioned method for arranging partial discharge detection sensors for high-voltage equipment is implemented.

[0130] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may include a memory remotely located relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned networks include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof. The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and may be located in one place, or may be distributed to multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment.

[0131] Those skilled in the art will appreciate that all or some of the steps and systems described above can be implemented as software, firmware, hardware, or any combination thereof. Some or all of the physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on computer-readable media, which may include computer storage media (or non-transitory media) and communication media (or transient media). As is well known to those skilled in the art, the term computer storage media encompasses volatile and non-volatile, removable, and non-removable media implemented in any method or technology for storing information, such as computer-readable instructions, data structures, program modules, or other data. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disks (DVDs) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. Furthermore, as is well known to those skilled in the art, communication media typically includes computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0132] The above is a specific description of the preferred implementation of the present application, but the present application is not limited to the above implementation mode. Technical personnel familiar with the art can also make various equivalent modifications or substitutions under the shared conditions that do not violate the spirit of the present application. These equivalent modifications or substitutions are all included in the scope defined by the claims of the present application.

Claims

1. A method for arranging partial discharge detection sensors for high-voltage equipment, characterized in that: Applied to a power grid management system, the power grid management system being provided with digital twin models of multiple high-voltage devices, the method comprising: Acquire type, model, and location information of a plurality of high-voltage devices in a target area; Determine, based on each of the type and model information and the location information, the radiation area data and the first installation area of ​​the high-voltage device corresponding to the type and model information, wherein the first installation area is an area where a sensor can be installed, determined based on historical false detection data of the high-voltage device. Based on the first installation area corresponding to the target high-voltage equipment, and according to the historical defect information corresponding to the target high-voltage equipment and the radiation area data corresponding to the other plurality of non-target high-voltage equipment, layout information of the partial discharge detection sensors is obtained; Arrange and set the partial discharge detection sensor on the digital twin model of the target installed high-voltage equipment according to the layout information; The method for determining the first installation area includes: obtaining historical false detection data corresponding to the high-voltage equipment, the historical false detection data including sensor position information of the partial discharge detection sensor set on the high-voltage equipment and false detection cause information recorded in the history corresponding to the partial discharge detection sensor; determining the first installation area on a digital twin model of the high-voltage equipment at the target installation based on the sensor position information, the false detection cause information, and the preset installation position information; The determining, according to each of the type and model information and the location information, the radiation area data and the first installation area of ​​the high-voltage equipment corresponding to the type and model information includes: Searching a preset high-voltage equipment radiation data table according to each type and model information to determine the radiation range data of the high-voltage equipment corresponding to the type and model information, wherein the high-voltage equipment radiation data table is obtained by detecting the electromagnetic field radiation of the high-voltage equipment, and the high-voltage equipment radiation data table includes power frequency electromagnetic field data, radio frequency radiation data, and thermal radiation data; Determine the radiation area data of the high-voltage equipment corresponding to the type and model information based on the radiation range data and the location information; and search the equipment information table based on each type and model information to determine the first installation area corresponding to the type and model information.

2. The method for arranging partial discharge detection sensors for high-voltage equipment according to claim 1, characterized in that: The obtaining, based on the first installation area corresponding to the target-installed high-voltage equipment and according to historical defect information corresponding to the target-installed high-voltage equipment and radiation area data corresponding to multiple other non-target-installed high-voltage equipment, layout information of the partial discharge detection sensor includes: The first installation area, the historical defect information, and the radiation area data are input into a trained partial discharge detection sensor layout model for multimodal candidate scoring processing, and the layout information of the partial discharge detection sensor is output. The partial discharge detection sensor layout model includes an input coding layer, a multimodal fusion layer, and a spatial optimization layer.

3. The method for arranging partial discharge detection sensors for high-voltage equipment according to claim 2, characterized in that: The step of inputting the first installation area, the historical defect information, and the radiation area data into a trained partial discharge detection sensor layout model for multimodal candidate scoring processing, and outputting the layout information of the partial discharge detection sensor, comprises: Inputting the first installation area, the historical defect information, and the radiation area data into the input coding layer, and encoding the first installation area, the historical defect information, and the radiation area data respectively through the input coding layer to obtain a regional feature map, a defect feature vector, and a radiation feature vector; Performing cross-modal attention fusion processing on the regional feature map, the defect feature vector, and the radiation feature vector through a multimodal fusion layer to obtain a fused feature map; The spatial optimization layer performs candidate scoring processing on the fused feature map and the first installation area to obtain layout information of the partial discharge detection sensor.

4. The method for arranging partial discharge detection sensors for high-voltage equipment according to claim 3, characterized in that: The spatial optimization layer includes a candidate point generation sublayer, a layout scoring sublayer, and a constraint optimization sublayer. The spatial optimization layer performs candidate scoring processing on the fused feature map and the first installation area to obtain layout information of the partial discharge detection sensor, including: Perform candidate processing on the first installation area through the candidate point generation sublayer to obtain candidate point coordinates; Performing feature extraction and scoring processing on the fused feature map and the candidate point coordinates through the layout scoring sublayer to obtain candidate point position scores and sensor type probabilities; The constraint optimization sublayer performs radiation constraint processing on the candidate point coordinates, the candidate point position scores, and the sensor type probability to obtain layout information of the partial discharge detection sensor.

5. The method for arranging partial discharge detection sensors for high-voltage equipment according to claim 1, characterized in that: The obtaining, based on the first installation area corresponding to the target-installed high-voltage equipment and according to historical defect information corresponding to the target-installed high-voltage equipment and radiation area data corresponding to multiple other non-target-installed high-voltage equipment, layout information of the partial discharge detection sensor includes: Determining a non-installation area in a digital twin model of the target-installed high-voltage equipment based on the radiation area data of a plurality of non-target-installed high-voltage equipment; removing the first installation area according to the non-installation area to obtain a second installation area; Layout information of the partial discharge detection sensors is determined according to the second installation area and the historical defect information. The layout information of the partial discharge detection sensors includes types and quantities of the partial discharge detection sensors.

6. The method for arranging partial discharge detection sensors for high-voltage equipment according to claim 1, characterized in that: The determining, based on the sensor position information, the misdetection cause information, and the preset installation position information, a first installation area on the digital twin model of the target high-voltage equipment includes: When the misdetection cause information is that first data detected by the partial discharge detection sensor corresponding to the sensor position information does not meet the partial discharge positioning calculation requirements, and second data detected by other partial discharge detection sensors meet the partial discharge positioning calculation requirements, performing area optimization processing based on the preset installation position information and the sensor position information, and determining a first installation area on the digital twin model of the target installed high-voltage equipment; or, When the misdetection cause information is that third data from multiple detections of the same partial discharge detection sensor corresponding to the sensor position information is erroneous, performing area optimization processing based on the preset installation position information and the sensor position information, and determining a first installation area on the digital twin model of the target high-voltage equipment; or, When the misdetection cause information is that in the sensor position information, there are multiple different types of partial discharge detection sensors and the fourth data detected do not meet the partial discharge positioning calculation requirements, regional optimization processing is performed according to the preset installation position information and the sensor position information, and a first installation area is determined on the digital twin model of the high-voltage equipment of the target installation.

7. The method for arranging partial discharge detection sensors for high-voltage equipment according to claim 2, characterized in that: The training method of the partial discharge detection sensor layout model includes: Obtaining a sensor layout training text and a sensor layout true label corresponding to the sensor layout training text, wherein the sensor layout training text includes the first installation area, the historical defect information, and the radiation area data; Inputting the sensor layout training text into an initial partial discharge detection sensor layout model, wherein the partial discharge detection sensor layout model includes an input encoding layer, a multimodal fusion layer, and a spatial optimization layer; Encoding the first installation area, the historical defect information, and the radiation area data respectively through the input coding layer to obtain a regional feature map, a defect feature vector, and a radiation feature vector; Performing cross-modal attention fusion processing on the regional feature map, the defect feature vector, and the radiation feature vector through a multimodal fusion layer to obtain a fused feature map; Performing candidate scoring processing on the fused feature map and the first installation area through the spatial optimization layer to obtain layout information of the partial discharge detection sensor; A multi-task loss value is calculated based on the layout information of the partial discharge detection sensor and the true label of the sensor layout to obtain a multi-task total loss value. The partial discharge detection sensor layout model is iteratively updated based on the multi-task total loss value to obtain a trained partial discharge detection sensor layout model.

8. The method for arranging partial discharge detection sensors for high-voltage equipment according to claim 7, characterized in that: The spatial optimization layer includes a candidate point generation sublayer, a layout scoring sublayer, and a constraint optimization sublayer. The spatial optimization layer performs candidate scoring processing on the fused feature map and the first installation area to obtain layout information of the partial discharge detection sensor, including: Perform candidate processing on the first installation area through the candidate point generation sublayer to obtain candidate point coordinates; Performing feature extraction and scoring processing on the fused feature map and the candidate point coordinates through the layout scoring sublayer to obtain candidate point position scores and sensor type probabilities; Performing radiation constraint processing on the candidate point coordinates, the candidate point position scores, and the sensor type probabilities through the constraint optimization sublayer to obtain layout information of the partial discharge detection sensor, where the layout information of the partial discharge detection sensor includes sensor position information, sensor type information, and sensor position score information; The multi-task loss value calculation based on the layout information of the partial discharge detection sensor and the true label of the sensor layout to obtain the multi-task total loss value includes: Performing loss value calculation on the sensor position information and the regional feature map to obtain a sensor position loss value; The sensor type information and the real sensor position information in the real sensor layout label are used to calculate the type loss value to obtain the sensor type loss value; Performing radiation constraint loss calculation on the sensor position score information and the radiation feature vector to obtain a position score loss value; The sensor position loss value, the sensor type loss value, and the position score loss value are subjected to a total loss calculation to obtain a multi-task total loss value.

9. A power grid management system, characterized in that: The method comprises a controller, wherein the controller comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for arranging partial discharge detection sensors for high-voltage equipment according to any one of claims 1 to 8 is implemented.

10. A computer storage medium, characterized in that The computer storage medium stores computer-executable instructions, and the computer-executable instructions are used to execute the method for arranging partial discharge detection sensors for high-voltage equipment according to any one of claims 1 to 8.

Images