Power transmission line fault feature and risk analysis method based on digital twinning

Abstract

The invention relates to the field of power transmission, and discloses a power transmission line fault feature and risk analysis method based on digital twinning, which comprises the following steps: acquiring cable environment data, cable displacement data and cable operation data through a plurality of data acquisition sensors, and mapping the data to a pre-constructed cable pipe gallery digital twinning model; power transmission line fault features are generated, and a power transmission line risk analysis result is generated; according to the power transmission line risk analysis result, generating an inspection task and a corresponding virtual inspection task; executing a virtual inspection task in the cable pipe gallery digital twin model through the digital inspection equipment to verify the virtual inspection task, and if verification is passed, sending the inspection task to the inspection equipment to execute the inspection task; and correcting the power transmission line risk analysis result according to the inspection image of the inspection equipment to obtain a power transmission fault judgment result. Therefore, the safety risk of personnel and equipment in the inspection process is reduced.

Description

Technical Field

[0001] This invention relates to the field of power transmission, and more specifically to a method for analyzing the fault characteristics and risks of power transmission lines based on digital twins. Background Technology

[0002] High-voltage cable tunnels are critical connection nodes in the high-voltage power transmission process, so their status awareness and fault location are extremely important for the operation and maintenance of high-voltage power grids.

[0003] The following technical problems often exist in existing fault analysis and troubleshooting processes:

[0004] First, safety risks often arise during fault location and risk elimination, such as water seepage and electric leakage threatening the personal safety of inspection personnel. If inspection robots are used directly, due to the complexity of the spatial layout and terrain within the integrated cable tunnel, personnel need to use the camera mounted on the inspection robot for auxiliary judgment, which affects the inspection efficiency of the inspection robot.

[0005] Secondly, due to the complexity and diversity of pipelines within the cable tunnel, faults or influencing factors in other pipelines (such as water pipes and gas pipes) can interfere with the fault diagnosis of the transmission line during fault location and analysis, thereby reducing the accuracy of fault analysis. Summary of the Invention

[0006] The summary section of this invention provides a brief overview of the concepts, which will be described in detail in the detailed description section below. This summary section is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0007] This invention proposes a method for analyzing the fault characteristics and risks of transmission lines based on digital twins, in order to solve one or more of the technical problems mentioned in the background section above.

[0008] This invention provides a method for analyzing the fault characteristics and risks of transmission lines based on digital twins, including:

[0009] Cable environmental data, cable displacement data, and cable operation data are collected by various data acquisition sensors installed in the cable tunnel and mapped to a pre-built digital twin model of the cable tunnel.

[0010] Based on cable environmental data, cable displacement data, cable operation data, and cable tunnel inspection channel information, transmission line fault characteristics are generated, and based on the transmission line fault characteristics, transmission line risk analysis results are generated.

[0011] Based on the risk analysis results of transmission lines, inspection tasks and corresponding virtual inspection tasks are generated.

[0012] The virtual inspection task is performed in the digital twin model of the cable tunnel by digital inspection equipment to verify the virtual inspection task. If the verification is successful, the inspection task is sent to the inspection equipment to execute the inspection task.

[0013] The risk analysis results of the transmission line are corrected based on the inspection images of the inspection equipment to obtain the transmission fault determination results.

[0014] Optionally, multiple data acquisition sensors include multiple temperature sensors and multiple gas concentration sensors. The multiple temperature sensors and multiple gas concentration sensors are evenly distributed in the cable duct. Each temperature sensor is used to detect the temperature value of the corresponding cable duct area, and each gas concentration sensor is used to detect the gas concentration of the corresponding cable duct area.

[0015] Based on cable environmental data, cable displacement data, cable operation data, and cable tunnel inspection information, transmission line fault characteristics are generated. Based on these fault characteristics, transmission line risk analysis results are then generated, including:

[0016] Based on the temperature values ​​collected by the corresponding temperature sensors, the temperature change rate of each cable tunnel area is calculated, and the temperature change rates of each cable tunnel area at the same time are spliced ​​together to obtain the temperature change characteristics; the temperature gradient values ​​of adjacent cable tunnel areas are calculated to obtain the temperature gradient characteristics.

[0017] The temperature characteristics are obtained by combining the temperature change characteristics and the temperature gradient characteristics;

[0018] Based on cable displacement data, determine the displacement characteristics in the cable gallery;

[0019] Determine the gas concentration characteristics in the cable tunnel based on the gas concentration.

[0020] Based on the inspection channel information, determine the type of cable tunnel;

[0021] Based on temperature characteristics, displacement characteristics, gas concentration characteristics, and pipe gallery type, fault characteristics of transmission lines are generated;

[0022] Based on the characteristics of transmission line faults, risk analysis results are generated for transmission lines, including the location of the risk, the type of risk, and the level of risk.

[0023] Optionally, based on the fault characteristics of the transmission line, generate transmission line risk analysis results, including:

[0024] The fault features of the transmission line are input into an encoding network with an attention mechanism to obtain the encoded features;

[0025] The encoded features are input into multiple task headers to obtain the risk analysis results of the transmission line. The multiple task headers include the cable risk level output task header, the risk type output task header, the risk factor contribution output task header, the risk location output task header, and the cross-pipeline correlation analysis result output task header.

[0026] Optionally, the inspection equipment can be an inspection robot or an inspection personnel terminal. Each inspection robot has a corresponding digital robot in the cable tunnel digital twin model and data mapping is achieved. Each inspection personnel has a corresponding digital inspection personnel in the cable tunnel digital twin model and data mapping is achieved.

[0027] Optionally, based on the transmission line risk analysis results, inspection tasks and corresponding virtual inspection tasks are generated, including:

[0028] Based on the risk type and risk level, the type of inspection equipment is determined, and the inspection path is generated based on the risk location. This includes: if the risk level is low to medium and the risk type is unspecified, the inspection personnel terminal is determined as the inspection equipment; if the risk level is high and the risk type is specified, the inspection robot is determined as the inspection equipment; and path planning is performed based on the risk location to obtain multiple candidate inspection paths.

[0029] The inspection equipment type, each candidate inspection path, and the inspection speed constitute an inspection task, and each inspection task is mapped to a corresponding virtual inspection task.

[0030] Optionally, a virtual inspection task can be performed in the digital twin model of the cable tunnel using digital inspection equipment to verify the virtual inspection task. If the verification is successful, the inspection task is sent to the inspection equipment for execution, including:

[0031] By using digital inspection equipment to execute each virtual inspection task in the digital twin model of the cable tunnel, the path verification results corresponding to each virtual inspection task are obtained. The path verification results include whether there are any obstacles, the number of obstacles, and the inspection time.

[0032] The path verification result is verified by pre-set verification rules. If the verification passes, the corresponding inspection task is sent to the inspection equipment to execute the inspection task.

[0033] Optionally, the risk analysis results of the transmission line can be corrected based on the inspection images from the inspection equipment to obtain the transmission fault determination results, including:

[0034] Acquire inspection images taken after the inspection equipment reaches the risk location. The inspection images include infrared images and visible light images.

[0035] Image recognition was performed on infrared and visible light images respectively to obtain image recognition results;

[0036] The risk analysis results of the transmission line are corrected based on the image recognition results to obtain the transmission fault determination results.

[0037] Optionally, the method for analyzing the fault characteristics and risks of transmission lines based on digital twins of the present invention further includes:

[0038] When the inspection equipment performs its inspection task, the corresponding digital inspection equipment moves synchronously in real time to record inspection process data.

[0039] The present invention has the following beneficial effects:

[0040] 1. Improved the safety, intelligence, and efficiency of utility tunnel operation and maintenance. Specifically, by constructing a digital twin model of the cable tunnel and executing virtual inspection tasks within this model, the scientific and rational nature of the inspection plan is ensured. After the inspection equipment completes its tasks, it combines infrared and visible light images for image recognition and risk analysis to accurately determine transmission faults. Simultaneously, the digital inspection equipment records inspection process data in real time, achieving full-process digital management. The overall solution enables multi-dimensional monitoring, visual analysis, intelligent planning, and dynamic verification of transmission line risks, significantly improving the safety, intelligence, and efficiency of utility tunnel operation and maintenance.

[0041] 2. Improved accuracy of fault analysis: Specifically, by deploying sensors for temperature, gas concentration, and other parameters in the cable tunnel, real-time environmental monitoring is achieved, and multimodal fault features such as temperature, displacement, gas concentration, and tunnel type are extracted and fused. These fault features are input into an attention-based coding network, and the risk level, type, location, and contribution are output through a multi-task head, enabling multi-angle, fine-grained, and interpretable risk analysis. Through multi-source sensor acquisition, multimodal feature fusion, and deep coding with attention mechanisms and multi-task output, not only is the accuracy and comprehensiveness of risk identification improved, but also quantitative, localized, and causally interpretable risk analysis results are provided for operation and maintenance, thereby achieving safe and intelligent management of cable tunnel operation. Attached Figure Description

[0042] The above and other features, advantages, and aspects of the various embodiments of the present invention will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0043] Figure 1 This is a flowchart of the method for analyzing the fault characteristics and risks of transmission lines based on digital twins according to the present invention;

[0044] Figure 2 This is a schematic diagram of the cable gallery structure in the digital twin-based transmission line fault feature and risk analysis method of the present invention;

[0045] Figure 3 This is a flowchart of the inspection feedback and result correction process for the transmission line fault characteristics and risk analysis method based on digital twins of the present invention.

[0046] Figure 4 This is a flowchart of the multi-task-head-based risk analysis process of the transmission line fault characteristics and risk analysis method based on digital twins of the present invention. Detailed Implementation

[0047] The invention will now be described in more detail with reference to the accompanying drawings. While some embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the invention. It should be understood that the drawings and embodiments of the invention are for illustrative purposes only and are not intended to limit the scope of protection of the invention.

[0048] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0049] It should be noted that the concepts of "first" and "second" mentioned in this invention are only used to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0050] It should be noted that the terms "a" and "a plurality of" used in this invention are illustrative rather than restrictive. Those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0051] The names of messages or information exchanged between the various devices of this invention are for illustrative purposes only and are not intended to limit the scope of these messages or information.

[0052] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0053] like Figure 1 As shown, the method for analyzing the fault characteristics and risks of transmission lines based on digital twins of the present invention includes the following steps:

[0054] Step 101: Collect cable environmental data, cable displacement data, and cable operation data through various data acquisition sensors installed in the cable tunnel, and map them into a pre-built digital twin model of the cable tunnel;

[0055] In practice, cable tunnels refer to a general term for underground or above-ground passages, boxes, or corridors for laying cables, including facilities such as pipes, supports, inspection wells, entrances and exits, and access routes. Figure 2 This is a schematic diagram of the cable tunnel structure in the digital twin-based transmission line fault characteristic and risk analysis method of this invention. Multiple data acquisition sensors refer to devices installed within the cable tunnel for monitoring and collecting various types of data. These sensors include, but are not limited to, temperature sensors, light sensors, gas concentration sensors, displacement sensors, etc. Based on this, cable environmental data includes various parameters of the environment within the cable tunnel, such as temperature, humidity, light intensity, and gas concentration. Cable displacement data refers to the change in the cable's position within the tunnel, including cable displacement values. Cable operation data refers to the electrical parameters of the cable during operation, including voltage and local current changes. Depending on actual needs, distributed fiber optic temperature measurement systems (DTS), partial discharge detection instruments, cable sheath circulating current sensors, and fiber optic strain sensors can also be used to collect data. The distributed fiber optic temperature measurement system is used to collect temperature data. Partial discharge detection instruments are used to collect cable operation data. Partial discharge is a discharge phenomenon that occurs in an insulator under a strong electric field and is an early sign of cable insulation aging and faults. These sensors can detect these minute discharge signals. Cable sheath circulating current sensors (such as current transformers) are also used to collect cable operating data. Cable sheath circulating current is the current generated by electromagnetic induction in high-voltage cables during operation. Abnormal circulating current values ​​usually indicate sheath grounding faults or other electrical problems. Fiber optic strain sensors are used to collect cable displacement data. Additionally, partial cable operating data can be obtained through power monitoring systems and combined with partial cable operating data collected by sensors to obtain complete cable operating data. Power monitoring systems are real-time operation monitoring and control systems for power grids or distribution systems, used to collect and monitor the operating condition data of power equipment (such as cables).

[0056] Based on this, the implementing entity of the digital twin-based transmission line fault characteristic and risk analysis method of this invention (e.g., a comprehensive online monitoring system for cable tunnels) can map various collected data to a pre-constructed digital twin model of the cable tunnel. The digital twin model of the cable tunnel is constructed using BIM (Building Information Modeling), GIS (Geographic Information System), and 3D laser scanning technology to create a visualized 3D model of the cable tunnel. Then, engines such as Unity are used to render the digital twin model of the cable tunnel, thereby realizing the mapping relationship between the cable tunnel and its digital twin model. The comprehensive online monitoring system for cable tunnels is an intelligent monitoring platform for the operating environment and equipment status of cable tunnels.

[0057] As an example, a visualized 3D model of a cable tunnel can be constructed using the following steps: First, a 3D laser scanner is used to perform a full-range scan of the cable tunnel to obtain high-precision point cloud data. Then, BIM software is used to create an information model containing all pipelines, supports, and ancillary facilities. This information model is then integrated with a GIS system, enabling a geographic mapping between the model space and the actual space. Finally, Unity is used as the final digital twin visualization and interaction platform. The BIM model is imported into the engine via FBX (FilmBox) / IFC (Industry Foundation Classes) format and given materials and dynamic interactive capabilities. During the data mapping stage, data collected by sensors can be bound to corresponding locations in the model via the data interface API (Application Programming Interface).

[0058] Step 102: Based on cable environmental data, cable displacement data, cable operation data, and cable tunnel inspection channel information, generate transmission line fault characteristics, and based on the transmission line fault characteristics, generate transmission line risk analysis results.

[0059] In practice, as an example, multiple data acquisition sensors include multiple temperature sensors and multiple gas concentration sensors. The multiple temperature sensors and multiple gas concentration sensors are evenly distributed in the cable duct. Each temperature sensor is used to detect the temperature value of the corresponding cable duct area, and each gas concentration sensor is used to detect the gas concentration of the corresponding cable duct area.

[0060] Based on this, the temperature change rate of each cable tunnel area is calculated according to the temperature values ​​collected by the corresponding temperature sensors. The temperature change rates of all cable tunnel areas at the same time are then concatenated to obtain temperature change characteristics. The temperature gradient values ​​of adjacent cable tunnel areas are calculated to obtain temperature gradient characteristics. Here, the temperature gradient value can be a temperature difference, and the temperature change rate is the temperature change rate between the current time point and the previous time point. The temperature change characteristics and temperature gradient characteristics are combined to obtain temperature characteristics. Based on cable displacement data, displacement characteristics in the cable tunnel are determined; based on gas concentration, gas concentration characteristics in the cable tunnel are determined. Based on inspection channel information, the tunnel type of the cable tunnel is determined; based on temperature characteristics, displacement characteristics, gas concentration characteristics, and tunnel type, transmission line fault characteristics are generated. Based on this, the transmission line fault characteristics are input into an encoding network with an attention mechanism to obtain encoded characteristics. The encoded characteristics are then input into multiple task heads to obtain transmission line risk analysis results. These task heads include a cable risk level output task head, a risk type output task head, a risk factor contribution output task head, a risk location output task head, and a cross-pipeline correlation analysis result output task head.

[0061] Step 103: Based on the risk analysis results of the transmission line, generate inspection tasks and corresponding virtual inspection tasks.

[0062] Step 104: Perform a virtual inspection task in the digital twin model of the cable tunnel using a digital inspection device to verify the virtual inspection task. If the verification is successful, send the inspection task to the inspection device to execute the inspection task.

[0063] In some embodiments, the inspection equipment is an inspection robot or an inspection personnel terminal. Each inspection robot has a corresponding digital robot in the cable tunnel digital twin model and data mapping is implemented. Each inspection personnel has a corresponding digital inspection personnel in the cable tunnel digital twin model and data mapping is implemented. Here, inspection equipment refers to the physical equipment that actually performs the inspection task, including inspection robots and inspection personnel terminals. An inspection robot is an automated device that can move automatically or remotely in a cable tunnel and carries multiple sensors (such as visible light cameras and infrared thermal imagers), suitable for inspections in high-risk or complex areas. An inspection personnel terminal refers to a handheld device, such as a tablet or smartphone, used by inspection personnel to receive tasks, view data, and perform manual operations. Digital inspection equipment refers to the virtual counterparts of the inspection robot and inspection personnel terminal in the cable tunnel digital twin model, namely, digital robots and digital inspection personnel. They can simulate the movement, path, and operation of real equipment to pre-verify the feasibility of tasks in a virtual environment.

[0064] Based on this, the cable tunnel integrated online monitoring system performs conditional judgments according to risk type and risk level to determine the type of inspection equipment. It then generates inspection paths based on the risk location, including: if the risk level is low to medium and the risk type is unspecified, the inspection personnel terminal is identified as the inspection equipment. If the risk level is high and the risk type is specified, the inspection robot is identified as the inspection equipment. Specified types refer to those fault types that are pre-set as high-risk or require special handling, such as "partial discharge," "excessive gas concentration," and "abnormal sheath circulation." Once these types of risks are identified by the model, it means that higher-level inspection equipment (such as inspection robots) is needed to handle them to ensure personnel safety or rapid response. Unspecified types refer to risks that do not belong to the "specified types," which are generally considered relatively low risks, such as minor temperature increases or small displacement changes. For these risks, inspection personnel can handle them using terminals. Path planning is performed based on the risk location, resulting in multiple candidate inspection paths. An inspection task is composed of the type of inspection equipment, each candidate inspection path, and the inspection speed. Each inspection task is mapped to a corresponding virtual inspection task. In practice, after determining the type of inspection equipment, the integrated online monitoring system for cable tunnels uses a path planning algorithm (such as the A* algorithm) to plan multiple feasible routes from the starting point (such as the entrance or robot charging station) to the risk location in a digital twin model, with the risk location as the endpoint. Candidate inspection paths refer to the multiple alternative routes generated for the inspection task after the path planning algorithm has run. These paths typically differ in terms of length, safety, or the areas they traverse.

[0065] Based on this, the determined inspection equipment type (e.g., "inspection robot"), multiple candidate inspection paths (e.g., path A, path B), and inspection speed (e.g., 2 m / s) are combined, with each path combined with the equipment type and speed to form an independent task. As an example, Task 1: Inspection equipment type is "inspection robot," inspection path is "path A," and inspection speed is "2 m / s." Task 2: Inspection equipment type is "inspection robot," inspection path is "path B," and inspection speed is "1.5 m / s" (possibly because path B has more curves). Subsequently, an identical virtual inspection task is created in the digital twin model for each generated inspection task. For example, if Task 1 involves dispatching a real robot, the cable tunnel integrated online monitoring system will create an instruction in the digital twin model to have the digital robot travel along "path A" at a speed of "2 m / s." Here, inspection speed refers to the recommended travel speed for the inspection equipment (robot or personnel) on a specific path. Virtual inspection tasks are virtual versions of physical inspection tasks within the digital twin model of cable tunnels. They contain the same instructions as the physical tasks but are executed by digital inspection equipment in a virtual environment. The type of inspection equipment refers to the category of inspection tools selected by the integrated online monitoring system for cable tunnels to dispatch for a specific task, including inspection robots or personnel terminals.

[0066] In some embodiments, each virtual inspection task is executed in a digital twin model of a cable tunnel using digital inspection equipment, yielding path verification results for each virtual inspection task. These results include the presence or absence of obstacles, the number of obstacles, and the inspection time. In practice, within the digital twin environment, a "digital robot" or "digital inspector" corresponding to the inspection equipment performs simulated inspections along the task path. During execution, the equipment's movement, perception, and interaction along the path are simulated, including speed changes, path turns, and spatial maneuverability. Specifically, a path planning algorithm is invoked to perform virtual walking / movement in the 3D tunnel model; a simulation engine (such as Unity) is used to detect collisions and obstacles along the path; and the estimated inspection time is calculated based on the equipment's set inspection speed. The path verification result refers to the evaluation report generated after the virtual inspection task is executed in the digital twin model. This result includes three key indicators: the presence or absence of obstacles, the number of obstacles, and the inspection time. The presence or absence of obstacles assesses whether there are any objects or areas on the path that could obstruct the inspection equipment's passage, such as pipes, cable supports, or narrow passages. The number of obstacles is the specific number of obstacles on the path. The inspection time is the total time required to simulate completing the path.

[0067] Based on this, the path verification results are validated using pre-defined verification rules. If the verification passes, the corresponding inspection task is sent to the inspection device for execution. The verification rules are a pre-defined set of criteria used to determine whether the virtual inspection task meets the executable conditions. For example, the verification rules may state that there must be no obstacles on the path, or the number of obstacles must be less than or equal to a threshold; the inspection time must be less than a predetermined time limit (e.g., 30 minutes); and risk points must be within the path's coverage area. In practice, the executing entity uses logical judgments (if-else or a rule engine) to compare the path verification results obtained in the previous step (including the number of obstacles and the inspection time) with the pre-defined verification rules. If all conditions are met, it is considered passed; otherwise, it is considered failed. Only when the virtual inspection task passes all verification rules will the executing entity send the corresponding real inspection task (e.g., an inspection task containing path A) to the corresponding inspection device (e.g., a real inspection robot) via the network. If virtual verification fails, the executing entity may replan the path or notify maintenance personnel for manual intervention.

[0068] Step 105: Correct the risk analysis results of the transmission line based on the inspection images of the inspection equipment to obtain the transmission fault judgment result.

[0069] In some embodiments, such as Figure 3The diagram illustrates the inspection feedback and result correction flowchart of the digital twin-based transmission line fault characteristic and risk analysis method of the present invention. The process involves acquiring inspection images (including infrared and visible light images) taken by the inspection equipment after it reaches the risk location; performing image recognition on the infrared and visible light images to obtain the image recognition results; and correcting the transmission line risk analysis results based on the image recognition results to obtain the transmission line fault determination result. In practice, the inspection equipment integrates a visible light camera and an infrared thermal imager. When the inspection equipment reaches the risk location, the executing entity automatically or remotely controlled by a person simultaneously captures visible light and infrared images of the area. If it is an inspection robot, it will reach the designated risk location using GPS or a visual positioning system, then automatically activate the camera and thermal imager to take pictures, and transmit the images to the monitoring center via a wireless network. If it is an inspection personnel terminal, when personnel arrive at the designated location, the terminal will prompt them to use its built-in camera and thermal imager (or connected external devices) to take pictures. The image data captured by the inspection robot or the inspection personnel terminal is transmitted to the executing entity in real time or in batches through various communication methods. Inspection images refer to on-site photographs taken by inspection equipment (whether robots or human terminals) using their onboard cameras after reaching a risk location. These images are crucial for subsequent precise location and fault diagnosis. Infrared images are thermal imaging images that can display the temperature distribution on an object's surface. In power transmission line inspections, infrared images are particularly important for detecting hot spots (hot patches) at cables or joints, as these are early signs of many electrical faults. Visible light images are ordinary images that we can see with the naked eye. They are used to provide visual information on-site, such as physical damage to cable sheaths, corrosion of supports, and the presence of obstacles or unusual objects in the surrounding area.

[0070] Based on this, two independent image recognition models are pre-trained: one for processing infrared images and one for processing visible light images. These models are typically based on a convolutional neural network (CNN) architecture and are trained using a large number of labeled cable tunnel images. The infrared image recognition model is trained to identify hot spots, temperature anomalies, etc. The visible light image recognition model is trained to identify physical damage or anomalies such as cable sheath damage, support corrosion, external object intrusion, and water accumulation. The pre-training process for the image recognition models involves collecting a large amount of image data related to cable tunnels, including infrared images (thermal imaging data under different operating conditions) and visible light images (cable surface, supports, environment, etc.). The data needs to cover various scenarios: normal state, overheating, cracks, displacement, obstacles, water stains, smoke, etc. The collected images are labeled with categories such as "normal," "hot spot," "damaged," "loose," and "obstacle," or with specific target locations. Data preprocessing: Image enhancement: rotation, scaling, contrast enhancement, noise addition, etc., to improve model robustness; Infrared image normalization: mapping temperature gradients to uniform grayscale values; Dataset partitioning: training set (70%), validation set (20%), test set (10%). Based on different image types and recognition tasks, select appropriate neural network models: Infrared images are suitable for object detection models (YOLOv8) to detect and locate hotspots; visible light images are suitable for image classification + object detection models (ResNet+YOLO) to identify damage, obstacles, displacement, etc. If simultaneous processing of infrared and visible light is required, a multimodal fusion model (such as a two-stream CNN) can be used. At the start of training, load pre-trained model weights, which can significantly speed up training and improve model performance. Adjust parameters such as learning rate, batch size, and optimizer to find the optimal training configuration. Input the labeled dataset into the model for iterative training. During training, the model continuously adjusts its weights based on the difference between the predicted results and the true labels (i.e., the loss function) until the model performance reaches expectations. Evaluate the model's performance on independent test sets, typically using metrics such as accuracy and precision. If the model's performance is unsatisfactory, you can go back to the previous steps, such as collecting more data, adjusting the model architecture, or retraining. After training, deploy the model.

[0071] Based on this, the infrared and visible light images transmitted back from the inspection equipment are input into the corresponding image recognition models, and the image recognition results are output. The image recognition results are the specific output obtained by the image recognition model after analyzing the infrared and visible light images. The image recognition results are aligned with the transmission line risk analysis results (alignment by location, type, and time dimension). A fusion algorithm (such as rule fusion, weighted voting, Bayesian inference, etc.) is used to synthesize the results. Correction rules can be as follows: if the image recognition detects a "hot spot" and the temperature exceeds a threshold, the risk level is corrected to "high risk"; if the model predicts "gas leakage" but the image recognition does not detect smoke, the risk level is downgraded or marked as "pending verification"; if the model predicts a non-specified type: displacement risk, but the image recognition detects damage, the risk type is corrected to a non-specified type: damage. After correction, the final transmission line fault determination result is output as the basis for scheduling and maintenance. The transmission line fault determination result is the final conclusion regarding the transmission line fault. It is a comprehensive judgment based on the combination of model analysis results (prediction) and inspection image results (verification). It is generally more reliable than relying solely on sensor data.

[0072] In some embodiments, the above method may further include:

[0073] When the inspection equipment performs its inspection task, the corresponding digital inspection equipment moves synchronously in real time to record inspection process data. In practice, after receiving the real-time data transmitted by the inspection equipment, the executing entity immediately updates the state of the corresponding digital inspection equipment in the digital twin model. Throughout the synchronization process, the executing entity continuously records all dynamic data of the digital inspection equipment and stores this data in a database, forming complete inspection process data. Specifically, when the real-time synchronous motion inspection equipment moves within the cable tunnel, the virtual position and posture of its corresponding digital inspection equipment (e.g., a digital robot) in the digital twin model are also updated synchronously. This is typically achieved through the equipment's positioning system (such as UWB, inertial navigation, or visual SLAM) and data interface. Inspection process data refers to all data recorded by the executing entity during synchronous motion, including but not limited to: Real-time equipment position: the precise coordinates of the equipment within the tunnel; Movement trajectory: the actual route the equipment travels from the starting point to the end point; Time consumed: the total time to complete the task; Data along the route: all data collected by the equipment's sensors (such as cameras and gas sensors) during the journey.

[0074] These embodiments enhance the safety, intelligence, and efficiency of utility tunnel operation and maintenance. Specifically, by constructing a digital twin model of the cable tunnel and executing virtual inspection tasks within this model, the scientific rationality of the inspection plan is ensured. After the inspection equipment completes its tasks, it combines infrared and visible light images for image recognition and risk analysis to accurately determine transmission faults. Simultaneously, the digital inspection equipment records inspection process data in real time, achieving full-process digital management. The overall solution enables multi-dimensional monitoring, visual analysis, intelligent planning, and dynamic verification of transmission line risks, significantly improving the safety, intelligence, and efficiency of utility tunnel operation and maintenance.

[0075] In some embodiments, to further address the second technical problem described in the background section, namely, "due to the complex and diverse pipelines within cable ducts, faults or influencing factors in other pipelines (such as water pipes and gas pipes) can interfere with fault judgment during transmission line fault location and analysis, thereby reducing the accuracy of fault analysis," in some embodiments of the present invention, multiple data acquisition sensors include multiple temperature sensors and multiple gas concentration sensors. These sensors are uniformly distributed within the cable duct. Each temperature sensor detects the temperature value of a corresponding cable duct area, and each gas concentration sensor detects the gas concentration of a corresponding cable duct area. The temperature sensor, used to detect the temperature at a specific monitoring point within the cable duct, can be a thermocouple, a thermistor, or an infrared thermometer. The gas concentration sensor, used to detect the concentration of certain gases (such as carbon monoxide and methane) in the duct environment, is generally based on electrochemical, semiconductor, or infrared principles and outputs a signal corresponding to the gas concentration. The temperature value refers to the ambient temperature parameter detected in real-time by the temperature sensor at a specific location within the cable duct. The gas concentration refers to the proportion of a certain type of gas component in a unit volume of air within a specific area of ​​the cable duct.

[0076] Based on cable environmental data, cable displacement data, cable operation data, and cable tunnel inspection information, transmission line fault characteristics are generated. Based on these fault characteristics, transmission line risk analysis results are then generated, including:

[0077] Step 1: Calculate the temperature change rate of each cable tunnel area based on the temperature values ​​collected by the corresponding temperature sensors, and stitch together the temperature change rates of each cable tunnel area at the same time to obtain the temperature change characteristics; calculate the temperature gradient values ​​of adjacent cable tunnel areas to obtain the temperature gradient characteristics.

[0078] In some embodiments, the integrated online monitoring system for cable tunnels can periodically acquire temperature values ​​from each temperature sensor. For example, it can acquire values ​​every 5 minutes. Then, the temperature change rate is obtained by subtracting the temperature value from the previous temperature value at the current moment and dividing by the time interval. For instance, suppose a temperature sensor measures 30 degrees Celsius at 10:00 and 31 degrees Celsius at 10:05. The temperature change rate at 10:05 is (31 minus 30, then divided by 5, resulting in 0.2 degrees Celsius / minute). The cable tunnel integrated online monitoring system performs this calculation on all deployed temperature sensors. The temperature change rate data calculated by all temperature sensors at the same time (e.g., 10:05) are arranged and combined according to their position in the cable tunnel to form a data vector or array. This array represents the temperature change characteristics. If there are 100 temperature sensors in the tunnel, the cable tunnel integrated online monitoring system will arrange the temperature change rates calculated by these 100 sensors at 10:05 into an array containing 100 values, such as [0.2, 0.1, 0.3, ..., 0.5]. Then, the cable tunnel integrated online monitoring system will iterate through all adjacent temperature sensors and calculate their temperature difference at the same time. For example, assuming sensor A and sensor B are adjacent, at 10:05, A measures a temperature of 40 degrees Celsius, and B measures a temperature of 4... 5 degrees Celsius. Therefore, the temperature gradient between them is 45 minus 40, resulting in 5 degrees Celsius. The integrated online monitoring system for cable tunnels performs this calculation on all adjacent sensor pairs. Based on this, the temperature gradient values ​​calculated by all adjacent sensor pairs are concatenated to form a data vector or array, which is the temperature gradient feature. The rate of temperature change refers to how quickly the temperature value of a certain area changes over time. For example, if the temperature rises by 5 degrees Celsius in one hour, its rate of temperature change is 5 degrees Celsius per hour. This is an important indicator for measuring abnormal temperature changes. The temperature gradient value refers to the temperature difference between adjacent areas. The temperature change feature is a comprehensive set of data obtained by concatenating the temperature change rate data of all cable tunnel areas at the same moment; it reflects the temperature change trend of the entire tunnel at a certain point in time. The temperature gradient feature is a comprehensive set of data obtained by calculating the temperature gradient values ​​of all adjacent cable tunnel areas; it reveals the heat distribution and potential hotspot locations within the entire tunnel.

[0079] Step two: Combine the temperature change characteristics and temperature gradient characteristics to obtain the temperature characteristics;

[0080] In some embodiments, the integrated online monitoring system for cable tunnels sequentially concatenates the previously generated temperature change features (one array) and temperature gradient features (another array) to form a single long array. For example, a temperature change feature array A = [0.2, 0.1, ...] containing 100 values ​​was previously obtained. Simultaneously, a temperature gradient feature array B = [5, 2, ...] containing 99 values ​​was also obtained. By concatenating these arrays, you will obtain a temperature feature array C = [0.2, 0.1, ..., 5, 2, ...] containing 199 values.

[0081] Step 3: Determine the displacement characteristics in the cable gallery based on the cable displacement data;

[0082] In some embodiments, cable displacement data collected simultaneously from all displacement sensors are stitched together to form a data vector or array. This array is the displacement feature. For example, suppose 50 displacement sensors are installed in a cable tunnel. At a certain point in time, the integrated online monitoring system for the cable tunnel obtains the corresponding displacement values ​​from these 50 sensors, for example, [0.1mm, 0.2mm, -0.1mm, ..., 0.5mm]. This array containing 50 values ​​is the displacement feature used for subsequent analysis. The displacement feature is a dataset describing the displacement of all cables in the cable tunnel. This feature is an important indicator for assessing cable mechanical stress or potential damage risk.

[0083] Step 4: Determine the gas concentration characteristics in the cable tunnel based on the gas concentration.

[0084] In some embodiments, multiple gas concentration sensors are evenly distributed within the cable tunnel. These sensors continuously monitor the gas concentration in their respective areas. The integrated online monitoring system for cable tunnels stitches together the data acquired from all sensors at the same time to form a gas concentration feature array. Assuming 20 gas concentration sensors are installed, the integrated online monitoring system for cable tunnels stitches together the data from these sensors (e.g., methane concentration) into an array, such as [5ppm, 6ppm, 20ppm,...].

[0085] Step 5: Determine the type of cable tunnel based on the inspection channel information;

[0086] In some embodiments, the integrated online monitoring system for cable tunnels extracts inspection channel information from the BIM / GIS model. This inspection channel information refers to the layout, dimensions, width, and height of the internal passageways of the cable tunnel, as well as structured data such as obstacles, bends, and slopes. This information determines the passage capability of inspection equipment (such as inspection robots or personnel). Based on this, the system automatically analyzes the inspection channel information according to preset rules and assigns a type label to the tunnel. This label, or its coded value, is the tunnel type characteristic. The tunnel type is a classification of cable tunnels based on the inspection channel information. For example, it can be classified as "narrow," "spacious," "multi-level," or "complex" based on its spatial size and complexity. Different tunnel types may correspond to different inspection difficulties and risk factors.

[0087] Step 6: Generate transmission line fault characteristics based on temperature characteristics, displacement characteristics, gas concentration characteristics, and pipe gallery type;

[0088] In some embodiments, the executing entity concatenates the arrays of temperature features, displacement features, and gas concentration features prepared in the previous steps, along with the encoded value representing the tunnel type, sequentially to form a unified feature vector containing all the information. For example, previously we obtained a temperature feature array containing 199 values, a displacement feature array containing 50 values, a gas concentration feature array containing 20 values, and the tunnel type (e.g., "narrow" is encoded as [0,1]). Concatenating these data together yields a longer transmission line fault feature vector, such as [temperature data..., displacement data..., gas concentration data...,0,1].

[0089] Step 7: Based on the fault characteristics of the transmission line, generate the risk analysis results for the transmission line, including:

[0090] The fault characteristics of the transmission line are input into an encoding network with an attention mechanism to obtain the encoded features. The encoded features are then input into multiple task heads to obtain the risk analysis results of the transmission line. The multiple task heads include a cable risk level output task head, a risk type output task head, a risk factor contribution output task head, a risk location output task head, and a cross-pipeline correlation analysis result output task head.

[0091] The results of the transmission line risk analysis include the location of the risk, the type of risk, and the level of risk.

[0092] In some embodiments, such as Figure 4The diagram illustrates a multi-task-head-based risk analysis flowchart for the digital twin-based transmission line fault feature and risk analysis method of this invention. Attention mechanisms are a technique used in deep learning models that allows the model to focus more on the parts that have a greater impact on the final result when processing data. In this invention, it can automatically identify which features (such as temperature, displacement, gas concentration, etc.) are most critical for predicting specific risks (such as overheating, insulation faults) and assign these features higher weights. The encoding network is a neural network structure primarily responsible for learning and transforming the input raw feature data (i.e., transmission line fault features), compressing it into a more abstract and compact representation, i.e., encoded features. This process removes redundant information and extracts the most important patterns and correlations in the data. Based on this, the concatenated transmission line fault feature vector is input into the encoding network with an attention mechanism. The encoding network can consist of multiple fully connected layers, with Dropout and activation functions (such as ReLU) embedded between layers to prevent overfitting. The attention mechanism learns and adjusts the weights of each input feature to highlight those features most helpful for prediction. For example, suppose an encoding network receives a transmission line fault feature vector containing hundreds of data points. During the learning process, it discovers that temperature gradient data is particularly important for predicting overheating risk, and therefore assigns higher attention weights to these data. Ultimately, the network outputs a compressed vector as the encoded features. These encoded features are a series of data generated by the encoding network. They are a high-level abstraction and compression of the original fault features, containing all the key information, but in a form more conducive to subsequent task processing. The encoded features output by the encoding network are then sent to multiple independent task heads. These task heads are specially trained sub-networks for their respective tasks. In a multi-task learning model, a task head refers to an independent sub-network connected after the encoding network, specifically designed to complete a particular task. Each task head receives the same encoded features as input but is responsible for outputting different results. Multiple task heads refer to multiple parallel task heads that can simultaneously learn from a single encoded feature and output various risk analysis results, such as risk level and risk type. Multiple task headers include cable risk level output task header, risk type output task header, risk factor contribution output task header, risk location output task header, and cross-pipeline correlation analysis result output task header.

[0093] In some embodiments, the risk level output task head is a classification network that determines whether a risk is "high," "medium," or "low" based on encoded features. The risk type output task head is also a classification network used to identify fault types, such as "overheating," "insulation fault," or "mechanical damage." The risk location output task head is also a classification task used to precisely locate the specific location of the risk. The risk factor contribution output task head can utilize the attention weights in the encoding network, normalize them, and directly output the risk factor contribution to show the contribution of different features to the risk. The cross-pipeline correlation analysis result output task head is a regression task head that can output the risk scores of other pipelines, thereby analyzing their correlation with the transmission line risk. Through this multi-task learning approach, the model can simultaneously complete multiple analysis tasks and output comprehensive transmission line risk analysis results, including risk location, risk type, and risk level. The transmission line risk analysis result is a comprehensive output, derived by analyzing the fault characteristics of the transmission line (such as temperature, displacement, and operational data), and it includes a comprehensive assessment of potential fault risks, specifically including the location, type, and level of the risk. A risk location refers to the specific physical location where a transmission line may experience a fault or an anomaly, such as a specific area within a cable tunnel or at a particular cable support. The risk type indicates the type or nature of the potential fault; for example, "partial discharge," "excessive gas concentration," or "abnormal sheath current" are specified types. The risk level measures the severity of the risk. According to the document, risk levels can be categorized as "high risk" and "medium-low risk." Different risk levels will affect subsequent inspection task decisions; for example, high-risk tasks are typically assigned to inspection robots, while medium-low risk tasks may be assigned to inspection personnel.

[0094] These embodiments improve the accuracy of fault analysis. Specifically, by deploying sensors such as temperature and gas concentration sensors in the cable tunnel, real-time environmental monitoring is achieved, and multimodal fault features such as temperature, displacement, light intensity, gas concentration, and tunnel type are extracted and fused. These fault features are input into an attention-based coding network, and the risk level, type, location, and contribution are output through a multi-task head, enabling multi-angle, fine-grained, and interpretable risk analysis. Through multi-source sensor acquisition, multimodal feature fusion, and deep coding with attention mechanisms and multi-task output, not only is the accuracy and comprehensiveness of risk identification improved, but also quantitative, localized, and causally interpretable risk analysis results are provided for operation and maintenance, thereby achieving safe and intelligent management of cable tunnel operation.

[0095] In some embodiments, to facilitate the operation and loading of the digital twin model of the cable tunnel, improve loading efficiency, and reduce operating costs, different operating modes can be configured for the digital twin model of the cable tunnel. For example, operating modes may include, but are not limited to: low-energy operation mode, full-scale operation mode, and fault diagnosis mode. Each operating mode is configured with the digital entities to be presented and the display effect data corresponding to each digital entity (corresponding one-to-one with the actual equipment in the cable tunnel), and different operating modes are configured with different digital entities.

[0096] Based on this, when troubleshooting is required, the digital twin model of the cable tunnel enters the troubleshooting mode. In the troubleshooting mode, all digital entities are fully loaded and the dynamic effect data of all digital entities are unloaded. This can reduce the resource consumption during the troubleshooting process while ensuring the accuracy of the troubleshooting, and also make it easier to identify the current mode and avoid misoperation.

[0097] The above description is merely a selection of preferred embodiments of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention is not limited to specific combinations of the above-described technical features, but also includes other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in this invention.

Claims

1. A method for analyzing the fault characteristics and risks of transmission lines based on digital twins, characterized in that, include: Cable environmental data, cable displacement data, and cable operation data are collected by various data acquisition sensors installed in the cable tunnel and mapped to a pre-built digital twin model of the cable tunnel. Based on the cable environment data, cable displacement data, cable operation data, and the inspection channel information of the cable tunnel, transmission line fault characteristics are generated, and transmission line risk analysis results are generated based on the transmission line fault characteristics. Based on the risk analysis results of the transmission lines, inspection tasks and corresponding virtual inspection tasks are generated. The virtual inspection task is executed in the digital twin model of the cable tunnel by the digital inspection equipment to verify the virtual inspection task. If the verification is successful, the inspection task is sent to the inspection equipment to execute the inspection task. The risk analysis results of the transmission line are corrected based on the inspection images from the inspection equipment to obtain the transmission fault determination results.

2. The method for analyzing the fault characteristics and risks of transmission lines based on digital twins according to claim 1, characterized in that, The multiple data acquisition sensors include multiple temperature sensors and multiple gas concentration sensors, which are evenly distributed in the cable tunnel. Each temperature sensor is used to detect the temperature value of the corresponding cable tunnel area, and each gas concentration sensor is used to detect the gas concentration of the corresponding cable tunnel area. The process involves generating transmission line fault characteristics based on the cable environment data, cable displacement data, cable operation data, and the inspection channel information of the cable tunnel, and then generating transmission line risk analysis results based on these fault characteristics, including: Based on the temperature values ​​collected by the corresponding temperature sensors, the temperature change rate of each cable tunnel area is calculated, and the temperature change rates of each cable tunnel area at the same time are spliced ​​together to obtain the temperature change characteristics; the temperature gradient values ​​of adjacent cable tunnel areas are calculated to obtain the temperature gradient characteristics. The temperature characteristics are obtained by combining the temperature change characteristics and the temperature gradient characteristics; Based on cable displacement data, determine the displacement characteristics in the cable gallery; Determine the gas concentration characteristics in the cable tunnel based on the gas concentration. Based on the inspection channel information, determine the type of cable tunnel; Based on the temperature characteristics, displacement characteristics, gas concentration characteristics, and pipe gallery type, transmission line fault characteristics are generated; Based on the fault characteristics of the transmission line, a risk analysis result for the transmission line is generated, which includes the risk location, risk type, and risk level.

3. The method for analyzing the fault characteristics and risks of transmission lines based on digital twins according to claim 2, characterized in that, The step of generating transmission line risk analysis results based on the fault characteristics of the transmission line includes: The fault features of the transmission line are input into an encoding network with an attention mechanism to obtain encoded features; The encoded features are input into multiple task heads to obtain the transmission line risk analysis results. The multiple task heads include a cable risk level output task head, a risk type output task head, a risk factor contribution output task head, a risk location output task head, and a cross-pipeline correlation analysis result output task head.

4. The method for analyzing the fault characteristics and risks of transmission lines based on digital twins according to claim 3, characterized in that, The inspection equipment is an inspection robot or an inspection personnel terminal. Each inspection robot has a corresponding digital robot in the digital twin model of the cable tunnel and data mapping is achieved. Each inspection personnel has a corresponding digital inspection personnel in the digital twin model of the cable tunnel and data mapping is achieved.

5. The method for analyzing the fault characteristics and risks of transmission lines based on digital twins according to claim 4, characterized in that, The step of generating inspection tasks and corresponding virtual inspection tasks based on the risk analysis results of the transmission lines includes: Based on the risk type and risk level, the type of inspection equipment is determined, and the inspection path is generated based on the risk location. This includes: if the risk level is low to medium and the risk type is unspecified, the inspection personnel terminal is determined as the inspection equipment; if the risk level is high and the risk type is specified, the inspection robot is determined as the inspection equipment; and path planning is performed based on the risk location to obtain multiple candidate inspection paths. The inspection equipment type, each candidate inspection path, and the inspection speed constitute an inspection task, and each inspection task is mapped to a corresponding virtual inspection task.

6. The method for analyzing the fault characteristics and risks of transmission lines based on digital twins according to claim 5, characterized in that, The virtual inspection task is executed in the digital twin model of the cable tunnel using a digital inspection device to verify the virtual inspection task. If the verification is successful, the inspection task is sent to the inspection device for execution, including: Each virtual inspection task is executed in the digital twin model of the cable tunnel by digital inspection equipment, and the path verification result corresponding to each virtual inspection task is obtained. The path verification result includes whether there are any obstacles, the number of obstacles, and the inspection time. The path verification result is verified by a pre-set verification rule. If the verification passes, the corresponding inspection task is sent to the inspection equipment to execute the inspection task.

7. The method for analyzing the fault characteristics and risks of transmission lines based on digital twins according to claim 6, characterized in that, The step of correcting the risk analysis results of the transmission line based on the inspection images from the inspection equipment to obtain the transmission fault determination result includes: The inspection equipment acquires inspection images taken after reaching the risk location, including infrared images and visible light images. Image recognition was performed on infrared and visible light images respectively to obtain image recognition results; The risk analysis results of the transmission line are corrected based on the image recognition results to obtain the transmission fault determination results.

8. The method for analyzing the fault characteristics and risks of transmission lines based on digital twins according to claim 7, characterized in that, Also includes: When the inspection equipment performs the inspection task, the corresponding digital inspection equipment moves synchronously in real time to record inspection process data.

Images