Railway power equipment monitoring system

Abstract

The invention provides a railway power equipment monitoring system, and relates to the technical field of power equipment monitoring. The system comprises a data acquisition module for acquiring operation data of power equipment; the data analysis module constructs a random forest network model, optimizes the random network model by using a pigeon optimization algorithm to obtain an optimal model, and inputs the operation data into the optimal model to obtain a fault type and a traveling wave signal; the fault distance measurement module locates a fault position to obtain coordinate data; the inspection module performs inspection according to the coordinate data to obtain an inspection report; and the task management module receives the operation data, the inspection report and the fault type, and distributes tasks to workers. According to the method, the gradient of the dynamic topology semantic map is calculated through the composite potential field function, the optimal moving direction is obtained, the DLite algorithm is used, path selection of the algorithm is corrected according to the optimal moving direction in cooperation with monitoring of actual obstacle changes, and the dynamic environment adaptability is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power equipment monitoring, and particularly to a railway power equipment monitoring system. Background Art

[0002] The railway power equipment monitoring system is an important means to ensure the safe, stable and efficient operation of railway power facilities. With the rapid development of global railway transportation, especially the popularization of high-speed railways and urban rail transit, the reliability and safety of power equipment have become increasingly important. These equipment include substations, catenary equipment, distribution equipment, power supply systems, and related monitoring and protection devices. To ensure the normal operation of these power equipment, it is particularly necessary to implement an effective monitoring system. The evolution of power equipment monitoring technology has been continuously progressing with the development of the railway industry. Traditional monitoring methods usually rely on manual inspections and regular maintenance. Although they can ensure the operation of equipment to a certain extent, they are inefficient, have a long response time, and it is difficult to achieve real-time monitoring. The emergence of modern monitoring systems, with the help of modern sensor technology and information communication technology, has realized the full-cycle and all-round monitoring of power equipment. The basis of the power equipment monitoring system is sensor technology. Sensors can collect the operation data of equipment in real time, including parameters such as current, voltage, temperature, and frequency. These parameters are key indicators reflecting the health status of the equipment and are crucial for judging whether the equipment is in a normal working state.

[0003] Traditional monitoring methods have the defects of low efficiency, low monitoring accuracy, fragmented monitoring, and poor adaptability to dynamic environments. Summary of the Invention

[0004] The present invention provides a railway power equipment monitoring system to solve the defects of fragmented monitoring and poor adaptability to dynamic environments in the prior art.

[0005] On the one hand, the present invention provides a railway power equipment monitoring system, which includes: A data acquisition module for acquiring the operation data of power equipment.

[0006] A data analysis module that constructs a random forest network model, optimizes the random forest network model using the pigeon optimization algorithm to obtain an optimal model, and inputs the operation data into the optimal model to obtain the fault type and traveling wave signal.

[0007] A fault location ranging module for calculating the operation data to obtain a traveling wave signal, and positioning the fault location in the overhead power line by combining with the traveling wave principle and the traveling wave signal to obtain coordinate data.

[0008] An inspection module for performing inspections according to the coordinate data to obtain an inspection report.

[0009] The task management module is used to receive the operation data, inspection report and fault type, and assign tasks to staff.

[0010] According to the railway power equipment monitoring system provided by the present invention, the step of constructing a random forest network model includes: Construct multiple decision trees, set the maximum depth of the decision trees and the minimum number of samples required for classification points, and obtain an initial network model.

[0011] Acquire experimental data, and use the experimental data to train the initial network to obtain a random forest network model.

[0012] According to the railway power equipment monitoring system provided by the present invention, the step of optimizing the random network model using the pigeon optimization algorithm includes: The mean square error is used as the fitness function of the random forest network model.

[0013] Initialize the pigeon population, use Tent chaotic mapping to generate individual pigeon positions, and each individual pigeon represents a set of hyperparameters of the random forest network model.

[0014] Calculate the fitness value of each individual pigeon, use the navigation strategy to update the speed of each individual pigeon, and update the position of each individual pigeon.

[0015] The fitness value of each individual pigeon after the update is calculated. If the fitness value of the individual pigeon is higher than the preset fitness threshold, the individual with the highest fitness value is selected as the optimal hyperparameter of the model. Otherwise, the pigeon population continues to iterate until the fitness value is higher than the preset fitness threshold or the maximum number of iterations is reached, and the individual with the highest fitness value is selected as the optimal hyperparameter of the model.

[0016] According to the railway power equipment monitoring system provided by the present invention, the data analysis module includes: a cutting unit and a reading unit, wherein the cutting unit is used to cut off the fault line from the normal operating power line according to the fault type, and the reading unit is used to record the power consumption of the power equipment.

[0017] According to the railway power equipment monitoring system provided by the present invention, the step of obtaining coordinate data includes: The fault traveling wave arrival time independently recorded at each node is combined with the traveling wave signal to obtain a multi-node time difference sequence.

[0018] An overdetermined system of equations is established by combining the time difference equations of each node.

[0019] The multi-node time difference sequence is used in conjunction with a least squares residual optimization algorithm to solve the overdetermined equations to obtain coordinate data.

[0020] According to the railway power equipment monitoring system provided by the present invention, the specific steps for obtaining the inspection report include: Collect temperature data of the detection area and take pictures of the detection area to obtain original pictures. And perform filtering processing on the original pictures to obtain denoised images.

[0021] Convert the denoised images from the RGB color space to the HSV color space to obtain converted images.

[0022] Use the threshold method to perform color segmentation on the converted images to obtain binary images.

[0023] Use the contour detection method to identify the object contours in the binary images to obtain a contour list.

[0024] Use the moments of the contours to calculate the center point coordinates of the contour list, and mark the foreign objects invading the boundary according to the center point coordinates of the contour list to obtain invasion labels. Analyze based on the temperature data and the invasion labels to obtain the inspection report.

[0025] According to the railway power equipment monitoring system provided by the present invention, the inspection module includes: an environment acquisition unit and a path planning unit. The environment acquisition unit is used to collect environment data in real time. The path planning unit is used to generate an optimal path according to the environment data.

[0026] According to the railway power equipment monitoring system provided by the present invention, the steps for generating the optimal path include: Generate a dynamic topological semantic map according to the environment data.

[0027] Construct a composite potential field function that fuses multiple objectives. The composite potential field function includes: a target gravitational potential field function, a static obstacle repulsive potential field function, a dynamic obstacle prediction repulsive potential field function, and an energy consumption potential field function.

[0028] Use the composite potential field function to calculate the gradient of the dynamic topological semantic map to obtain the best moving direction.

[0029] Initialize the starting point and the ending point according to the dynamic topological semantic map, use the DLite algorithm, select a path according to the map, and for the global path, correct the path selection of the algorithm according to the best moving direction in cooperation with monitoring the actual changes of obstacles to obtain the optimal path.

[0030] According to the railway power equipment monitoring system provided by the present invention, the steps for generating the dynamic topological semantic map include: Obtain the 3D point cloud data and RGB-D images of the environment in real time.

[0031] The RGB-D image is segmented at the pixel level to obtain obstacle categories. The 3D point cloud data is denoised to obtain processed cloud data.

[0032] The obstacle category is mapped to the processed cloud data through a coordinate transformation matrix to obtain a 3D point cloud map with semantic labels.

[0033] Static objects in the 3D point cloud map with semantic labels are extracted to form static nodes, dynamic objects are extracted, and the current position, speed, acceleration and threat coefficient assigned to the corresponding objects are recorded to form dynamic nodes, and a dynamic topological semantic map is obtained.

[0034] According to the railway power equipment monitoring system provided by the present invention, the task management module includes: an input unit and a cancellation unit. A reminder unit. The input unit is used to manually input tasks. The cancellation unit is used to mark processed tasks. The reminder unit is used to pop up a window reminder for unprocessed tasks according to a preset time.

[0035] The railway power equipment monitoring system provided by the present invention constructs a random forest network model, optimizes the random network model using the pigeon optimization algorithm, obtains the optimal model, inputs the operating data into the optimal model, obtains the fault type, and improves the accuracy of fault classification. The arrival time of the fault traveling wave independently recorded by each node is used in conjunction with the traveling wave signal to obtain a multi-node time difference sequence, and an overdetermined equation group is established by combining the time difference equations of each node. The multi-node time difference sequence is used in conjunction with the least squares residual optimization algorithm to solve the overdetermined equation group to obtain coordinate data, thereby improving the accuracy of obtaining coordinate data and reducing the amount of calculation. A dynamic topological semantic map is generated according to environmental data, thereby improving the accuracy of obstacle resolution and solving the problems of dynamic obstacle avoidance lag and high resource consumption. The gradient of the dynamic topological semantic map is calculated using the composite potential field function to obtain the optimal moving direction, and the DLite algorithm is used to select a path according to the map, and the path selection of the actual obstacle change correction algorithm is coordinated with the optimal moving direction to obtain the optimal path. The success rate of obstacle avoidance and the applicability to dynamic environments are improved, while reducing energy consumption and inspection time. Inspections are carried out according to the optimal path, and inspection reports are obtained through analysis. The fault types and inspection reports are collected through the task management module, assigned to staff, and tasks are tracked, which reduces the fragmentation of monitoring and improves the efficiency of problem solving. It solves the defects of the existing traditional monitoring methods, such as low efficiency, low accuracy, fragmented monitoring, and poor adaptability to dynamic environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] To more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can also be obtained based on these drawings.

[0037] Figure 1 It is a schematic flowchart of the railway power equipment monitoring system provided by the embodiment of the present invention; Figure 2 It is a schematic flowchart of the pigeon optimization algorithm of the railway power equipment monitoring system provided by the embodiment of the present invention; Figure 3 It is a schematic flowchart of generating the optimal path of the railway power equipment monitoring system provided by the embodiment of the present invention. Detailed implementation manners

[0038] To make the objectives, technical solutions, and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions in the present invention in conjunction with the accompanying drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

[0039] As Figures 1 - 3 shown, the railway power equipment monitoring system provided by the embodiment of the present invention includes: A data acquisition module for acquiring the operation data of power equipment.

[0040] In this embodiment, the operation data includes data such as voltage, current, and temperature. A high-precision digital temperature sensor can be installed at key device nodes to monitor the device temperature in real time. The Hall effect sensor can be used for non-contact measurement of current and voltage to avoid affecting the original circuit. Continuous acquisition of key parameters such as temperature and current can be performed once per second. When the monitored value exceeds the threshold, high-frequency acquisition such as 10 times per second is immediately started to capture the abnormal process. Communication with the host computer or cloud platform is carried out through the TCP / IP protocol, supporting Gigabit Ethernet to ensure the stability of large data volume transmission. For data that needs to be uploaded in real time, 4G or 5G networks can be used to ensure the timeliness and reliability of data transmission. Digital filtering algorithms such as Kalman filtering are applied to the collected raw data to remove noise interference and improve data accuracy. Outliers and duplicate values in the operation data are removed, and the Lagrange interpolation method is used to supplement the missing values in the operation data. The moving average method can be used to reduce data fluctuations by calculating the average value of multiple data points, thereby smoothing the curve. Methods such as random sampling, stratified sampling, or clustering sampling can also be used to reduce the data volume while retaining important information and reducing the impact of noise on the analysis results. Efficient data structures such as hash tables or tree structures can be used to identify and delete duplicate data records to ensure the uniqueness and accuracy of the data. For outliers in the measured data, outliers can be identified through the quartiles of the data, and data points outside the range of Q1 - 1.5×IQR or Q3 + 1.5×IQR are regarded as outliers. When supplementing the missing values, the data points near the missing values are selected as the interpolation points. Generally, it is recommended to select 5 data points before and after to ensure the accuracy of interpolation.

[0041] The data analysis module constructs a random forest network model, optimizes the random forest network model using the pigeon optimization algorithm to obtain the optimal model, and inputs the operation data into the optimal model to obtain the fault type.

[0042] The steps for constructing the random forest network model include: Construct multiple decision trees and set the maximum depth of each decision tree and the minimum number of samples required for the classification point to obtain the initial network model.

[0043] Obtain experimental data and use the experimental data to train the initial network to obtain the random forest network model.

[0044] In this embodiment, the experimental data includes different operation data and corresponding fault types.

[0045] The steps for optimizing the random network model using the pigeon optimization algorithm include: Use the mean square error as the fitness function of the random forest network model.

[0046] Initialize the pigeon population and generate the positions of pigeon individuals using the Tent chaotic map. Each pigeon individual represents a set of hyperparameters of a random forest network model.

[0047] Update the velocity of each pigeon individual using the navigation strategy and update the position of each pigeon individual.

[0048] In this embodiment, the navigation strategy includes a map update phase and a compass update phase. In the map update phase, some pigeon individuals are randomly selected for velocity update, and in the compass update phase, the velocities of the remaining pigeon individuals are updated. The expression formula for the map update phase is:

[0049] is the velocity updated for the individual, is the current velocity of the individual, randomly select the vector difference of a row of individuals, is a random number between 0 and 1.

[0050] The expression formula for the compass update phase is:

[0051] Among them, is the velocity updated for the individual, is the current velocity of the individual, is the compass learning rate, is the vector difference between the current individual position and the optimal individual position, is a random number between 0 and 1.

[0052] The expression formula for updating the position of the pigeon individual is:

[0053] Among them, is the position of the pigeon individual after update, the current position of the pigeon individual, is the velocity of the i-th pigeon in the t-th generation.

[0054] Calculate the fitness value of each updated pigeon individual. If the fitness value of the pigeon individual is higher than the preset fitness threshold, select the individual with the highest fitness value as the optimal hyperparameter of the model; otherwise, continue to iterate until the fitness value is higher than the preset fitness threshold or the maximum number of iterations is reached, and select the individual with the highest fitness value as the optimal hyperparameter of the model.

[0055] The data analysis module includes: a cutting unit and a receiving unit. The cutting unit is used to cut off the faulty line from the normally operating power line according to the fault type. The receiving unit is used to record the power consumption of the power equipment.

[0056] In this embodiment, if a fault type is detected, such as overload, short circuit or equipment failure, the cutting unit will start immediately and quickly cut off the connection between the faulty line and the normally operating power line. This can effectively prevent the spread of faults and protect the safety of power equipment and users. Users can understand the current cutting status of the system at any time through the visual interface so as to make corresponding adjustments in time. The function of the meter reading unit is to conduct an in-depth analysis of the use of electric energy equipment. This unit will continuously monitor and calculate the power consumption and operating stability of each power-consuming equipment. Through the collected operating data, the meter reading unit can provide detailed reports on the efficiency of the power-consuming equipment. These reports not only show the real-time power consumption data of the equipment, but also generate trend charts of historical data to help users better understand the power consumption pattern and make reasonable power management decisions. The meter reading unit can also predict the possible performance changes of the equipment in the future by mining historical data, and warn users in advance to pay attention to possible potential problems. In this way, proactive measures can be taken before the problem breaks out to avoid unnecessary economic losses.

[0057] The fault distance measurement module is used to calculate the operation data to obtain a traveling wave signal, and locate the fault position in the power overhead line by using the traveling wave principle and the traveling wave signal to obtain coordinate data.

[0058] In this embodiment, the fault distance measurement module only receives the voltage and current data in the operation data for calculation.

[0059] The steps to obtain coordinate data include: The multi-node time difference sequence is obtained by using the fault traveling wave arrival time recorded independently at each node in combination with the traveling wave signal.

[0060] In this embodiment, it is assumed that multiple nodes such as nodes A, B and C are deployed in the power system on a transmission line, and high-precision current and voltage measuring equipment is installed on each node. When a fault occurs, each node immediately records the absolute time of arrival of the downlink wave. Through these times, the time difference between the nodes can be calculated. An overdetermined set of equations containing the node time difference is established. According to the relationship between the time difference and the distance between the nodes. The overdetermined set of equations can utilize the information of multiple nodes to improve the accuracy of fault location.

[0061] The overdetermined equations are established by combining the time difference equations of each node. The overdetermined equations are solved by using the multi-node time difference sequence combined with the least squares residual optimization algorithm to obtain the coordinate data.

[0062] In this embodiment, the least - squares residual optimization algorithm can reasonably use multiple measurement values to improve the accuracy of fault location calculation. This method greatly reduces the error probability caused by single observation by making full use of multiple observation data from different nodes. It has stronger robustness to the noise and outliers of individual data points. Even if there are certain deviations in the measurement data of a certain node, the overall result can still accurately reflect the location of the fault point, which helps to avoid incorrect fault location.

[0063] An inspection module for performing inspections according to coordinate data to obtain an inspection report.

[0064] In this embodiment, the inspection report includes fault coordinates, fault types, and content.

[0065] The specific steps for obtaining the inspection report include: Collect the temperature data of the detection area and take pictures of the detection area to obtain the original pictures. And perform filtering processing on the original pictures to obtain denoised images.

[0066] In this embodiment, for the temperature data, within a predetermined detection area, the unmanned aerial vehicle flies to a specified height and uses an infrared sensor to monitor the ground temperature in real - time. The data is transmitted through the ground control station, and the temperature distribution at different positions is recorded. Multiple images of the detection area are also taken using a high - definition camera. Each image should cover different perspectives and heights to ensure comprehensive information acquisition. A Gaussian filter can be used to remove Gaussian noise in the images, smooth the images, and retain edge information. The obtained images not only provide basic data for subsequent processing but also provide rich visual information for direct observation and analysis, which helps to understand the regional state.

[0067] Convert the denoised image from the RGB color space to the HSV color space to obtain a converted image.

[0068] In this embodiment, the HSV color space separates color information into hue, saturation, and value, which is convenient for color recognition and segmentation. Compared with the RGB space, HSV is more robust to light changes and can improve the accuracy of subsequent processing.

[0069] Use the threshold method to perform color segmentation on the converted image to obtain a binary image.

[0070] Use the contour detection method to identify the object contours in the binary image to obtain a contour list.

[0071] In this embodiment, contour detection can accurately identify the target objects in the image and provide the shape and boundary information of the objects.

[0072] Calculate the center point coordinates of the contour list using the moments of the contour, and mark the foreign objects that invade the boundary according to the center point coordinates of the contour list to obtain the invasion boundary label. Analyze according to the temperature data and the invasion boundary label to obtain the inspection report.

[0073] The inspection module includes: an environment acquisition unit and a path planning unit. The acquisition unit is used to collect environmental data in real time. The path planning unit is used to generate the optimal path according to the environment.

[0074] In this embodiment, the generated optimal path is used to provide for the drone to perform inspections.

[0075] The steps for generating the optimal path include: Generate a dynamic topological semantic map according to the environmental data.

[0076] Construct a composite potential field function that integrates multiple objectives. The composite potential field function includes: a target gravitational potential field function, a static obstacle repulsive potential field function, a dynamic obstacle prediction repulsive potential field function, and an energy consumption potential field function.

[0077] In this embodiment, the expression formula of the composite potential field function is:

[0078] Among them, is the overall potential field, is the target gravitational potential field, is the static obstacle repulsive potential field, is the dynamic obstacle prediction repulsive potential field, is the energy consumption potential field.

[0079] The expression formula of the target gravitational potential field function is:

[0080] Among them, is the target gravitational potential field, is the target gravitational intensity coefficient, is the current position of the drone, is the target position.

[0081] The expression formula of the static obstacle repulsive potential field function is:

[0082] Among them, the target gravitational potential field, is the repulsive force coefficient adjusted according to the obstacle type, is the position of the i-th obstacle, is the current position of the drone.

[0083] The expression formula of the dynamic obstacle prediction repulsive potential field function is:

[0084] Among them, is the dynamic repulsive force intensity coefficient, is the weight factor of each dynamic obstacle, is the position of the i-th dynamic obstacle at a future moment predicted based on Kalman filtering, is the current position of the UAV.

[0085] The expression formula of the energy consumption potential field function is:

[0086] Among them, is the energy consumption potential field, is the energy consumption intensity coefficient, is the gravitational acceleration, is the flight speed, is the acceleration.

[0087] The gradient of the dynamic topological semantic map is calculated using the composite potential field function to obtain the optimal movement direction.

[0088] The starting point and the ending point are initialized according to the dynamic topological semantic map. The DLite algorithm is used to select a path according to the map and the global path. According to the optimal movement direction, the path selection of the algorithm is corrected by monitoring the actual change of obstacles to obtain the optimal path.

[0089] The steps of generating the dynamic topological semantic map include: Obtain the 3D point cloud data and RGB-D images of the environment in real time.

[0090] In this embodiment, an RGB-D camera can be used to fix it on the UAV. For the obtained original 3D point cloud data, a denoising algorithm can be applied. Commonly used ones include voxel grid filtering and statistical filtering, etc. Taking statistical filtering as an example, a window radius can be set, and the noise points are determined by calculating the average distance of the neighboring points of each point. After the denoising process, the quality of the point cloud data is significantly improved, reducing the error analysis caused by noise and enhancing the accuracy of subsequent data processing.

[0091] Perform pixel-level segmentation on the RGB-D image to obtain the obstacle categories. Perform denoising processing on the 3D point cloud data to obtain the processed cloud data.

[0092] In this embodiment, deep learning models such as U-Net and MaskR-CNN can be used to perform pixel-level segmentation on the real-time captured RGB image. Pixel-level segmentation can clearly identify different objects in the image, providing a solid foundation for subsequent object recognition and environmental understanding.

[0093] Map the obstacle categories to the processed cloud data through a coordinate transformation matrix to obtain a 3D point cloud map with semantic labels.

[0094] In this embodiment, through the coordinate transformation matrix of the camera, the obstacle category results obtained by pixel-level segmentation are mapped to the processed 3D point cloud data. This matrix is generally obtained through calibration and can convert RGB pixel coordinates into 3D space coordinates. The obtained point cloud map with semantic labels provides higher-level information for understanding the environment and facilitates subsequent analysis and decision-making.

[0095] Extract static objects from the 3D point cloud map with semantic labels to form static nodes, extract dynamic objects, record the current position, speed, acceleration, and assign threat coefficients to the corresponding objects to form dynamic nodes, and obtain a dynamic topological semantic map.

[0096] In this embodiment, when extracting dynamic objects, motion targets can be detected to output the center coordinates, speed, and acceleration of the bounding box. For the segmented bird / vehicle targets, the displacement vector is calculated by the optical flow method. When extracting static objects, the visual segmentation results are projected onto the point cloud through the calibration matrix to supplement the labels in the uncovered areas. The records of static targets include: position, category. For the threat coefficient of birds, the trajectory can be predicted by using the social force model for the collective behavior of birds through dynamic obstacle threat modeling to obtain the threat coefficient.

[0097] The task management module includes: an input unit, a cancellation unit, and a reminder unit. The input unit is used for manual input of tasks. The cancellation unit is used to mark the processed tasks. The reminder unit is used to pop up reminders for unprocessed tasks according to the preset time.

[0098] In this embodiment, the main function of the input unit is to allow users to input task information manually. Users can input key information such as the title, detailed description, priority, and due date of the task in this unit. The system can also provide a drop-down menu for users to select the task category. In addition, the input unit can set up a voice input function, enabling users to add tasks quickly, which is particularly convenient when used on mobile devices. The cancellation unit is mainly used to mark the processed tasks, providing users with a clear task completion status. When users complete a certain task, they can mark the task as processed by clicking the "Complete" button, thus removing it from the to-do list. To prevent accidental operations, a confirmation dialog box can also be popped up before users cancel the task. At the same time, users can choose whether to save the completed tasks to the history for future viewing and review. The function of the reminder unit is to provide users with time reminders for unprocessed tasks. This function pops up reminders according to the preset time settings to ensure that users do not miss important due dates. The reminder time can also be customized for each task, for example, one day in advance. In addition, the system can also send dynamic updates related to unprocessed tasks to users via push notifications, enabling users to always keep track of the task progress. In the pop-up reminder, the system not only provides task information but also can give suggestions or motivational words to help users boost their work motivation.

[0099] Example 1: The data acquisition module has a current signal sampling rate of 10 kHz. It detects that the current in phase A suddenly increases from 200 A to 1500 A and transmits the abnormal data to the data analysis module and the fault location module in real time. After inputting the current data, the model outputs the fault type as "phase A short circuit fault", the amplitude of the traveling wave head = 1.2 kV, and the propagation speed = 2.8×10^8 m / s. The time differences of the fault traveling waves recorded by each monitoring node are: Δt1 = 12 μs, Δt2 = 18 μs. An overdetermined equation system is established, and the least squares residual algorithm is used to solve for the fault point coordinates, which are then transmitted to the inspection module. For the static nodes of the dynamic topological semantic map, the tower around the fault point has a λ value of 0.8, and the tree has a λ value of 0.3. For the dynamic node, the speed of the bird flock is 5 m / s, and λ = 0.5 + 0.1×5 = 1.0. The DLite algorithm generates a global path and makes local adjustments to avoid high-threat areas. It takes a high-definition image showing that the crack length on the insulator surface is > 10 cm and there is no overhanging foreign object. The local temperature is measured to be 92°C, and an inspection report is generated. The task type is: emergency repair, and the content is: replace the faulty insulator, check the wire connection and coordinate position. After the maintenance personnel complete the on-site processing, they mark the task as "completed" through the mobile device, and the system automatically archives the inspection report and the maintenance record.

[0100] The railway power equipment monitoring system provided in this embodiment constructs a random forest network model, optimizes the random network model using the pigeon optimization algorithm, obtains the optimal model, inputs the operating data into the optimal model, obtains the fault type, and improves the accuracy of fault classification. The arrival time of the fault traveling wave independently recorded by each node is combined with the traveling wave signal to obtain a multi-node time difference sequence, and an overdetermined equation group is established by combining the time difference equations of each node. The multi-node time difference sequence is used to cooperate with the least squares residual optimization algorithm to solve the overdetermined equation group to obtain coordinate data, thereby improving the accuracy of obtaining coordinate data and reducing the amount of calculation. A dynamic topological semantic map is generated according to environmental data, which improves the accuracy of obstacle resolution and solves the problems of dynamic obstacle avoidance lag and high resource consumption. The gradient of the dynamic topological semantic map is calculated using the composite potential field function to obtain the optimal moving direction, and the DLite algorithm is used to select the path according to the map, and the path selection of the actual obstacle change correction algorithm is coordinated with the optimal moving direction to obtain the optimal path. The success rate of obstacle avoidance and the applicability to dynamic environments are improved, while reducing energy consumption and inspection time. Inspections are carried out according to the optimal path, and inspection reports are obtained through analysis. The fault types and inspection reports are collected through the task management module, assigned to staff, and tasks are tracked, which reduces the fragmentation of monitoring and improves monitoring efficiency. It solves the defects of the existing traditional monitoring methods, such as low efficiency, high safety hazards, fragmented monitoring, and poor adaptability to dynamic environments.

[0101] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0102] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0103] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A railway power equipment monitoring system, characterized in that, The system includes: A data acquisition module, used to acquire the operation data of the power equipment; A data analysis module constructs a random forest network model, optimizes the random forest network model using a pigeon optimization algorithm to obtain an optimal model, inputs the operating data into the optimal model, and obtains a fault type; A fault distance measurement module is used to calculate the operation data to obtain a traveling wave signal; and locate the fault position in the power overhead line by using the traveling wave principle and the traveling wave signal to obtain coordinate data; An inspection module, used to perform inspections according to the coordinate data and obtain an inspection report; The task management module is used to receive the operation data, inspection report and fault type, and assign tasks to staff.

2. The railway power equipment monitoring system according to claim 1, characterized in that The step of constructing a random forest network model includes: Constructing multiple decision trees, and setting the maximum depth of the decision trees and the minimum number of samples required for classification points to obtain an initial network model; Acquire experimental data, and use the experimental data to train the initial network to obtain a random forest network model.

3. The railway power equipment monitoring system according to claim 1, characterized in that The steps of optimizing the random forest network model using the pigeon optimization algorithm include: The mean square error is used as the fitness function of the random forest network model; Initialize the pigeon population and use Tent chaotic mapping to generate individual pigeon positions. Each individual pigeon represents a set of hyperparameters of the random forest network model. Use the navigation strategy to update the speed of each individual pigeon and the position of each individual pigeon; The fitness value of each individual pigeon after the update is calculated. If the fitness value of the individual pigeon is higher than the preset fitness threshold, the individual with the highest fitness value is selected as the optimal hyperparameter of the model. Otherwise, the pigeon population continues to iterate until the fitness value is higher than the preset fitness threshold or the maximum number of iterations is reached, and the individual with the highest fitness value is selected as the optimal hyperparameter of the model.

4. The railway power equipment monitoring system according to claim 1, characterized in that, The data analysis module includes: a cutting unit and a metering unit; the cutting unit is used to cut off the faulty line from the normally operating power line according to the fault type; the metering unit is used to record the power consumption of the power equipment.

5. The railway power equipment monitoring system according to claim 1, wherein The steps to obtain coordinate data include: The multi-node time difference sequence is obtained by using the fault traveling wave arrival time recorded independently at each node in combination with the traveling wave signal; An overdetermined system of equations is established by combining the time difference equations of each node; The multi-node time difference sequence is used in conjunction with a least squares residual optimization algorithm to solve the overdetermined equations to obtain coordinate data.

6. The railway power equipment monitoring system according to claim 1, characterized in that, The specific steps of obtaining the inspection report include: Collect temperature data of the detection area and take a photo of the detection area to obtain an original image; and filter the original image to obtain a denoised image; Convert the denoised image from the RGB color space to the HSV color space to obtain a converted image; Performing color segmentation on the converted image using a threshold method to obtain a binary image; Using a contour detection method to identify the contours of objects in the binary image, and obtaining a contour list; Calculate the center point coordinates of the contour list using the moments of the contours, and mark the foreign objects invading the boundary based on the center point coordinates of the contour list to obtain invasion boundary labels; analyze based on the temperature data and the invasion boundary labels to obtain an inspection report.

7. The railway power equipment monitoring system according to claim 1, characterized in that, The inspection module includes: an environment acquisition unit and a path planning unit; the environment acquisition unit is used to collect environmental data in real time; the path planning unit is used to generate an optimal path according to the environmental data.

8. The railway power equipment monitoring system according to claim 7, wherein The step of generating the optimal path includes: Generate a dynamic topological semantic map according to the environmental data; Construct a composite potential field function that integrates multiple objectives; the composite potential field function includes: a target gravitational potential field function, a static obstacle repulsive potential field function, a dynamic obstacle prediction repulsive potential field function, and an energy consumption potential field function; Calculate the gradient of the dynamic topological semantic map using the composite potential field function to obtain the best moving direction; Initialize the starting point and the ending point according to the dynamic topological semantic map, use the DLite algorithm to select a path according to the map, and correct the path selection of the algorithm according to the best moving direction in cooperation with monitoring the changes of actual obstacles to obtain an optimal path.

9. The railway power equipment monitoring system according to claim 8, wherein, The step of generating a dynamic topological semantic map includes: Obtain the 3D point cloud data and RGB-D images of the environment in real time; Perform pixel-level segmentation on the RGB-D images to obtain obstacle categories; perform denoising processing on the 3D point cloud data to obtain processed cloud data; Map the obstacle categories into the processed cloud data through a coordinate transformation matrix to obtain a 3D point cloud map with semantic labels; Extract static objects in the 3D point cloud map with semantic labels to form static nodes, extract dynamic objects, and record the current position, speed, acceleration, and threat coefficient assigned to the corresponding objects to form dynamic nodes, thereby obtaining a dynamic topological semantic map.

10. The railway power equipment monitoring system according to claim 1, wherein The task management module includes: an input unit, a cancellation unit, and a reminder unit; the input unit is used for manual input of tasks; the cancellation unit is used to mark the processed tasks; the reminder unit is used to pop up a reminder for unprocessed tasks according to a preset time.

Images