A substation autonomous inspection and foreign matter cleaning collaborative control system and method

Abstract

The application discloses a substation autonomous inspection and foreign matter cleaning cooperative control system and method, and belongs to the technical field of intelligent inspection and foreign matter cleaning. The system comprises: a multi-modal detection module that acquires images, point clouds and multi-source sensing data, and identifies foreign matter coordinates and visual features; a spatial positioning module that performs spatial registration and obtains foreign matter three-dimensional coordinates; a risk decision module that fuses multi-dimensional information and evaluates foreign matter risk levels; a cleaning execution module that generates cleaning paths and execution parameters and issues them; and an inspection execution module that plans inspection paths and actions based on states and constraint conditions, and completes subsequent equipment inspection tasks. The scheme enables the system to achieve clear division of labor and accurate control in multiple stages such as foreign matter identification, spatial positioning, risk assessment, cleaning instruction generation and post-cleaning inspection, improves the expansibility and overall response efficiency of the system, and thus guarantees efficient, safe and stable execution of the substation autonomous inspection and foreign matter cleaning tasks.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of intelligent inspection and foreign matter cleaning, and particularly relates to a substation autonomous inspection and foreign matter cleaning collaborative control system and method. BACKGROUND

[0002] With the intelligent development of the power system, the operation safety and efficiency of the substation as the core node of the power grid are of great concern. There are a large number of high-voltage electrical equipment in the substation. If foreign matters (such as animals, floating objects, and vegetation intrusion) appear in the surrounding environment, serious accidents such as short circuit, discharge, and even fire may be caused. The traditional manual inspection method is limited by high labor cost, high operation risk, and many blind spots in detection, and cannot meet the real-time and accuracy requirements of the modern power grid.

[0003] Nowadays, the artificial inspection and foreign matter cleaning are performed as follows: first, an inspection route is planned in advance and personnel are assigned, the personnel enter the substation area with detection tools, the equipment state and the surrounding environment are checked through visual observation and instrument measurement, the position, type, and potential risk of the foreign matter are recorded after the foreign matter is found, it is judged whether the foreign matter needs to be cleaned immediately according to experience, the foreign matter is removed using a manual tool, and the cleaning effect is checked again to confirm the cleaning effect and record the data.

[0004] However, the whole process relies on manual operation, the efficiency is limited by the physical strength and experience of the personnel, and the missed inspection or incomplete cleaning is caused by fatigue or negligence; the data recording and reporting are delayed or have errors due to human factors, and lack of real-time feedback mechanism, and it is difficult to quickly respond to the dynamically changing environmental risks; for complex or hidden foreign matters (such as conductive bodies embedded in the gaps of the equipment), the manual cleaning is difficult. SUMMARY

[0005] The substation autonomous inspection and foreign matter cleaning collaborative control system and method provided by the embodiments of the application solve the problems that the existing foreign matter cleaning and intelligent inspection rely on manual operation throughout the process, the efficiency is limited by the physical strength and experience of the personnel, the missed inspection or incomplete cleaning is caused by fatigue or negligence, the data recording and reporting are delayed or have errors due to human factors, and lack of real-time feedback mechanism, and it is difficult to quickly respond to the dynamically changing environmental risks; for complex or hidden foreign matters (such as conductive bodies embedded in the gaps of the equipment), the manual cleaning is difficult.

[0006] In a first aspect, the embodiments of the application provide a substation autonomous inspection and foreign matter cleaning collaborative control system, which comprises:

[0007] A multi-modal detection module is configured to acquire two-dimensional image data, laser radar point cloud data, and multi-modal sensor data of the substation, and identify the two-dimensional coordinate positions of various foreign matters and the first foreign matter visual feature description in the two-dimensional image data by using a convolutional neural network target detection algorithm.

[0008] a spatial positioning module configured to perform spatial registration processing according to the laser radar point cloud data and the two-dimensional coordinate positions to obtain first three-dimensional spatial coordinates of each foreign object;

[0009] a risk decision module configured to obtain spatial layout information of the electrical equipment, input the first three-dimensional spatial coordinates, the first foreign object visual feature description, the spatial layout information, and the multi-modal sensor data into a preset foreign object risk level scoring model to obtain first foreign object risk level scores of each foreign object;

[0010] a cleaning execution module configured to determine first foreign object operation requirements of each foreign object according to the first foreign object risk level scores and a preset cleaning strategy library, obtain first current positions, first attitude angles, and first motion state information of the inspection robot, determine a cleaning path of the inspection robot and first cleaning execution parameters for each foreign object according to the first three-dimensional spatial coordinates, the first current positions, the first attitude angles, the first motion state information, and the first foreign object operation requirements, and send the cleaning path and the first cleaning execution parameters for each foreign object to the inspection robot;

[0011] an inspection execution module configured to, if receiving cleaning completion information of the inspection robot, obtain second current positions, second attitude angles, and second motion state information of the inspection robot, obtain a spatial topology structure and safety constraint conditions of the substation, and determine an inspection path of the inspection robot and inspection actions for each electrical equipment according to the second current positions, the second attitude angles, the second motion state information, the spatial topology structure, the spatial layout information, and the safety constraint conditions, and send the inspection path and the inspection actions for each electrical equipment to the inspection robot.

[0012] In a second aspect, an embodiment of the present application provides a substation autonomous inspection and foreign object cleaning cooperative control method, and the method comprises:

[0013] obtaining two-dimensional image data, laser radar point cloud data, and multi-modal sensor data of a substation, and identifying two-dimensional coordinate positions and first foreign object visual feature descriptions of each foreign object in the two-dimensional image data by using a convolutional neural network target detection algorithm;

[0014] performing spatial registration processing according to the laser radar point cloud data and the two-dimensional coordinate positions to obtain first three-dimensional spatial coordinates of each foreign object;

[0015] obtaining spatial layout information of the electrical equipment, inputting the first three-dimensional spatial coordinates, the first foreign object visual feature description, the spatial layout information, and the multi-modal sensor data into a preset foreign object risk level scoring model to obtain first foreign object risk level scores of each foreign object;

[0016] According to the first foreign matter risk level score and a preset cleaning strategy library, a first foreign matter operation requirement of each foreign matter is determined, a first current position, a first attitude angle and first motion state information of the inspection robot are obtained, a cleaning path of the inspection robot and a first cleaning execution parameter for each foreign matter are determined according to the first three-dimensional space coordinate, the first current position, the first attitude angle, the first motion state information and the first foreign matter operation requirement, and the cleaning path and the first cleaning execution parameter for each foreign matter are sent to the inspection robot.

[0017] If the cleaning completion information of the inspection robot is received, a second current position, a second attitude angle and second motion state information of the inspection robot are obtained, a space topology structure and a safety constraint condition of the transformer substation are obtained, an inspection path of the inspection robot and an inspection action for each electrical equipment are determined according to the second current position, the second attitude angle, the second motion state information, the space topology structure, the space layout information and the safety constraint condition, and the inspection path and the inspection action for each electrical equipment are sent to the inspection robot.

[0018] In a third aspect, an electronic device is provided, which includes a processor, a memory, and a program or instructions stored in the memory and executable on the processor, and the program or instructions, when executed by the processor, implement the steps of the method of the second aspect.

[0019] In a fourth aspect, a readable storage medium is provided, which stores a program or instructions, and the program or instructions, when executed by a processor, implement the steps of the method of the second aspect.

[0020] In the embodiments of the present application, the system is made to realize clear division of labor and accurate control in multiple stages such as foreign matter identification, space positioning, risk assessment, cleaning instruction generation and post-cleaning inspection, and the expansibility and overall response efficiency of the system are improved, so as to ensure efficient, safe and stable execution of the transformer substation autonomous inspection and foreign matter cleaning task. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 is a structural schematic diagram of a transformer substation autonomous inspection and foreign matter cleaning collaborative control system provided by an embodiment of the present application;

[0022] Figure 2 is a structural schematic diagram of a transformer substation autonomous inspection and foreign matter cleaning collaborative control system provided by an embodiment of the present application;

[0023] Figure 3 is a flowchart of a transformer substation autonomous inspection and foreign matter cleaning collaborative control method provided by an embodiment of the present application;

[0024] Figure 4is a structural schematic diagram of an electronic device provided in Embodiment Four of the present application. DETAILED DESCRIPTION

[0025] In order to make the objects, technical solutions and advantages of the present application clearer, the specific embodiments of the present application are further described in detail below with reference to the drawings. It should be understood that the specific embodiments described herein are only used to explain the present application and not to limit the present application. In addition, it should be noted that, for the convenience of description, only parts related to the present application are shown in the drawings, rather than all contents. Before discussing the example embodiments in more detail, it should be mentioned that some example embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations (or steps) as sequential processes, many of the operations can be implemented in parallel, concurrently or simultaneously. In addition, the order of the operations can be rearranged. The processes can be terminated when the operations are completed, but can also have additional steps not included in the drawings. The processes can correspond to methods, functions, procedures, subroutines, etc.

[0026] The technical solutions in the embodiments of the present application will be described clearly below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art belong to the scope of protection of the present application.

[0027] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are generally of a kind and do not limit the number of objects, for example, the first object can be one or more. In addition, "and / or" in the specification and claims means at least one of the connected objects, and the character " / ", generally represents a "or" relationship between the associated objects before and after.

[0028] The RSMC chip, the multi-stage starting method of the chip and the Beidou communication navigation device provided in the embodiments of the present application will be described in detail below with reference to the drawings, through specific embodiments and application scenarios.

[0029] Embodiment One

[0030] Figure 1 is a structural schematic diagram of a transformer station autonomous inspection and foreign matter cleaning collaborative control system provided in Embodiment One of the present application. As shown in Figure 1 , specifically includes the following:

[0031] The multi-modal detection module 101 is configured to acquire two-dimensional image data, laser radar point cloud data and multi-modal sensor data of the substation, and identify two-dimensional coordinate positions of each foreign object and a first foreign object visual feature description in the two-dimensional image data by using a convolutional neural network object detection algorithm.

[0032] The spatial positioning module 102 is configured to perform spatial registration processing according to the laser radar point cloud data and the two-dimensional coordinate positions, to obtain a first three-dimensional spatial coordinate of each foreign object.

[0033] The risk decision module 103 is configured to acquire spatial layout information of the electrical equipment, input the first three-dimensional spatial coordinate, the first foreign object visual feature description, the spatial layout information and the multi-modal sensor data into a preset foreign object risk level scoring model, and obtain a first foreign object risk level score of each foreign object.

[0034] The cleaning execution module 104 is configured to determine a first foreign object operation requirement of each foreign object according to the first foreign object risk level score and a preset cleaning strategy library, acquire a first current position, a first attitude angle and a first motion state information of the inspection robot, determine a cleaning path of the inspection robot and a first cleaning execution parameter for each foreign object according to the first three-dimensional spatial coordinate, the first current position, the first attitude angle, the first motion state information and the first foreign object operation requirement, and send the cleaning path and the first cleaning execution parameter for each foreign object to the inspection robot.

[0035] The inspection execution module 105 is configured to acquire a second current position, a second attitude angle and a second motion state information of the inspection robot if the cleaning completion information of the inspection robot is received, acquire a spatial topology structure and a safety constraint condition of the substation, and determine an inspection path of the inspection robot and an inspection action for each electrical equipment according to the second current position, the second attitude angle, the second motion state information, the spatial topology structure, the spatial layout information and the safety constraint condition, and send the inspection path and the inspection action for each electrical equipment to the inspection robot.

[0036] In the embodiment, the substation can be a kind of key facilities in the power system, used to realize the functions of voltage transformation, power distribution and control protection, etc. In the present scheme, the substation is the target area of the inspection and cleaning operation, including a plurality of electrical equipment (such as transformers, switch devices, busbars, lightning arresters, etc.) and auxiliary facilities such as access paths, fences, ground, etc.

[0037] The two-dimensional image data can be image information data collected by an image sensor (such as a high-definition camera) and expressed in the form of a pixel matrix, usually including an RGB image or a grayscale image. In the present scheme, the two-dimensional image data is used to perform target detection, image change recognition, and region feature extraction, and is one of the main data sources for visual analysis.

[0038] The laser radar point cloud data can refer to the spatial three-dimensional point set data obtained by laser scanning by a laser radar device, each point containing spatial coordinates and possible reflection intensity attributes.

[0039] The multi-modal sensor data can refer to perception data collected by different types of sensors (such as infrared, temperature and humidity, ultrasonic, gas detection, vibration sensors, etc.) and describing the state of the environment or object.

[0040] The convolutional neural network target detection algorithm can be a deep learning model based on a convolutional neural network (CNN) and is used to detect the location and category of target objects in an image. Typical representatives include YOLO, Faster R-CNN, SSD, etc. In the present scheme, it is used to detect foreign objects in the substation from two-dimensional image data, including the extraction of two-dimensional coordinate positions and feature information.

[0041] The foreign object can refer to an abnormal object that does not belong to the body of the substation equipment or its attached structure, which may enter the station due to wind blowing, animal intrusion, falling, etc., such as plastic bags, bird nests, tools, animals, branches, etc.

[0042] The two-dimensional coordinate position can refer to the pixel-level position of the foreign object in the image coordinate system, usually represented in the form of a center point coordinate or a top-left corner and a bottom-right corner coordinate of a bounding box. It is one of the output results of the target detection algorithm and is used for subsequent spatial registration and three-dimensional positioning calculation.

[0043] The first foreign object visual feature description can refer to the feature vector or semantic description information extracted by the convolutional neural network when identifying the foreign object, which can represent the appearance attributes of the foreign object, such as color distribution, shape contour, texture pattern, etc.

[0044] The substation scene can be captured in real time by deploying high-definition visible light industrial cameras on inspection robots or fixed monitoring devices, set a frame rate (such as 1-5 frames per second), obtain image data covering the entire substation area, and form the two-dimensional image data. The image data is in RGB format, with a resolution of 1920x1080 or higher, and is accompanied by a timestamp for subsequent synchronization with point cloud data and sensor data. A three-dimensional laser radar (such as Velodyne, Livox, etc.) installed on a robot or fixed platform is used to scan the spatial environment of the substation in a multi-beam and 360° rotation manner, and obtain the laser radar point cloud data. Each sample point in the point cloud data records its spatial three-dimensional coordinates (X, Y, Z) and reflection intensity value, which can be used to reconstruct the geometric structure of the equipment and the spatial distribution of foreign objects in the substation. By deploying other sensor components (such as infrared thermal imagers, ultrasonic range finders, temperature and humidity meters, air quality sensors, etc.) in the inspection system, non-image data reflecting the substation environment and equipment operating state are synchronously collected, forming the multi-modal sensor data. These data can be used to enhance the accuracy of foreign object judgment and subsequent risk level determination. The two-dimensional image data is input into a trained convolutional neural network (CNN) target detection model, such as YOLOv5, Faster R-CNN or RetinaNet. The model structure includes:

[0045] Backbone network (such as CSPDarknet, ResNet) for extracting image features;

[0046] Neck network (such as FPN, PAN) for multi-scale feature fusion;

[0047] Head network outputs the boundary box position, class label and confidence of the foreign object.

[0048] The network output contains the rectangular boundary box and center point coordinates of each detected foreign object, forming the two-dimensional coordinate position of the foreign object.

[0049] Based on the boundary box output by the detection model, the RoI Align region alignment technology is used to extract the depth feature vector of each foreign object region in the intermediate feature map of CNN, which is used to represent the color distribution, texture details and edge contour of the foreign object, and constitutes the first foreign object visual feature description. This description is used for subsequent foreign object classification, risk scoring and cleaning strategy matching steps.

[0050] The first three-dimensional spatial coordinates can be calculated by spatially aligning the two-dimensional coordinates of the foreign object on the image plane with the three-dimensional point cloud data collected by the laser radar, establishing a pixel-point cloud mapping relationship, and thus obtaining the three-dimensional coordinates representing the actual position of the foreign object in the substation scene.

[0051] First, the extrinsic matrix between the camera and the LiDAR is obtained using methods such as Zhang's method or LiDAR-Camera Calibration Toolkit (e.g. Kalibr), i.e. the rotation matrix R (3x3) and the translation vector T (3x1) between the LiDAR coordinate system and the image coordinate system. These extrinsic parameters are used to transform the radar points from the radar coordinate system to the camera coordinate system. The intrinsic matrix K of the camera, including the focal length fx, fy and the principal point coordinates cx, cy, is obtained to describe the projection process from three-dimensional space to two-dimensional image plane. This matrix is usually obtained by a chessboard image calibration method. For each point X, Y, Z (in the LiDAR coordinate system) in the LiDAR point cloud data, the following conversion steps are used to map it to the corresponding pixel point (u, v) in the image coordinate system:

[0052] First, the point cloud data is converted to the camera coordinate system by rotation and translation:

[0053]

[0054] Then project the three-dimensional point to the image plane:

[0055]

[0056] where (X, Y, Z) is the coordinate of the LiDAR point cloud data; R is the rotation matrix (3x3), which represents the directional transformation of rotating the point in the radar coordinate system to the camera coordinate system; T is the translation vector (3x1), which represents the offset position of the origin of the radar coordinate system relative to the camera coordinate system; , , is the three-dimensional coordinate in the camera coordinate system (i.e. the point cloud point is converted to the camera's perspective).

[0057] Then establish a one-to-one mapping relationship between the projection pixel position (u, v) of all laser points in the image and its corresponding three-dimensional coordinate (X, Y, Z) to form an image pixel index table. This index table can be used to quickly find the three-dimensional point corresponding to any image point. For the two-dimensional coordinate positions of each foreign object identified in the convolutional neural network target detection algorithm, find the corresponding three-dimensional point of the closest (u, v) point in the above index table. To enhance robustness, search for valid point clouds within a 3x3 or 5x5 neighborhood window around the pixel coordinate to avoid matching failures caused by occlusion or sparse point clouds. Through the matching result, the three-dimensional position (X, Y, Z) of each foreign object in the actual environment of the substation is extracted as its first three-dimensional spatial coordinate.

[0058] The electrical equipment can refer to various high-voltage or low-voltage devices that undertake functions such as power transmission, conversion, control, measurement, and protection in a substation, mainly including transformers: for voltage conversion between high voltage and low voltage; circuit breakers / switching devices: for controlling the on-off of the circuit; cable bridges, bus ducts, busbars: for transmitting current; voltage transformers, current transformers: for measurement and protection; capacitors, reactors: for reactive power compensation; arresters, grounding devices: for protection; communication and monitoring devices: for transmitting state information; ring network cabinets, switch cabinets, complete power distribution cabinets, etc.

[0059] The spatial layout information refers to the three-dimensional spatial arrangement structure information of the electrical equipment inside the substation, including device coordinates, the installation position of each electrical equipment in the three-dimensional space, such as (X, Y, Z) under the world coordinate system; device size and bounding box information, used to determine whether foreign objects are close to or block the device; device category, such as circuit breakers, transformers, busbars, etc., which can correspond to different sensitive areas and risk levels; installation relationship and topological structure, such as some devices are connected in groups and need to be paid attention to multiple points at the same time during cleaning; maintenance / channel restriction information, which areas are high-voltage danger zones, whether the robot is allowed to enter, how many meters away from the boundary the robot needs to slow down, etc.

[0060] The pre-set foreign object risk level scoring model can be a risk assessment model designed according to historical experience or expert rules, which is used to assign a risk level score to each foreign object target detected in the substation.

[0061] The first foreign object risk level score can refer to the risk score result of a certain foreign object target obtained by the above-mentioned model in the current identification period, which reflects the potential threat degree of the foreign object to the safe operation of the substation.

[0062] First, the spatial layout information of the electrical equipment is derived from the BIM model (Building Information Modeling) or GIS system of the substation, which is usually represented in a standard three-dimensional model (such as IFC format) or JSON / XML structured format, and the content includes the spatial position coordinates (Xi, Yi, Zi) of various types of equipment (such as transformers, circuit breakers, switch cabinets, etc.), physical dimensions (length, width, height), installation direction, type number, operating voltage level, and logical connection relationship with surrounding equipment (topological constraints). In order to match the three-dimensional space coordinate system actually used by the cleaning robot, a unified space coordinate transformation matrix (such as quaternion rotation and Euclidean translation based on homogeneous transformation) is used to register the layout data to the world coordinate system used by the robot. Then the first three-dimensional space coordinates, the first foreign object visual feature description, the spatial layout information, and the multi-modal sensor data are combined into a multi-dimensional feature vector wherein, is the first three-dimensional space coordinates, a first foreign object visual feature description; spatial layout information; multimodal sensor data. Finally, the multi-dimensional feature vector is input into a pre-trained foreign object risk level scoring model, which can be a lightweight neural network (such as MLP or Transformer structure); a tree model (such as XGBoost or LightGBM); or a weighted model of rule + scoring factor linear combination. The model output is the first foreign object risk level score.

[0063] To build a foreign object risk level scoring model with generalization ability, a high-quality historical training dataset needs to be constructed first. Each sample data in the training dataset consists of the following four types of input elements and a label:

[0064] a first three-dimensional spatial coordinate (i.e. the three-dimensional position of the historical foreign object );

[0065] a first foreign object visual feature description (usually an image semantic vector extracted by a pre-trained deep convolutional neural network, such as ResNet50 or YOLOv5 , representing the category, texture, size, etc. of the foreign object);

[0066] spatial layout information (including the spatial relationship between the foreign object and the key electrical equipment, such as distance , directional angle, whether it crosses the safety boundary, etc.);

[0067] multimodal sensor data (such as temperature distribution of infrared thermal imaging, wind speed and direction, electromagnetic interference strength, etc. environmental parameters, uniformly represented as );

[0068] The training label is the risk level label of each historical foreign object sample after expert scoring or accident data backtracking labeling, which is usually a continuous risk score (such as ∈[0,1]).

[0069] For each historical sample, the above four types of data are encoded into a unified format input feature vector: All vector features are normalized, such as using Min-Max normalization or Z-score standardization, to ensure that different dimensions have consistent weights in model input. Different model structures are selected according to the type of scoring label: if the label is a continuous risk score (regression task), a multilayer perceptron (MLP) can be used: a fully connected neural network or XGBoost, LightGBM, etc. Gradient boosting decision tree, suitable for small and medium-sized data; if the label is a risk level (classification task), a classification network with cross-entropy loss function can be used; or SVM + PCA dimensionality reduction method. The optimization method used in training can choose Adam or SGD, and the loss function is: for regression task: minimize mean square error (MSE); for classification task: minimize cross-entropy loss (Cross Entropy); through multiple rounds of training and cross-validation, to ensure the generalization performance of the model in different scenarios.

[0070] After training, the prediction accuracy and robustness of the model are evaluated using an independent test set, using indicators including regression indicators: , MAE, RMSE; classification indicators: accuracy, recall, F1-score, confusion matrix. If the model performs poorly in a certain scenario, oversampling / undersampling; data augmentation (such as image rotation, brightness disturbance); model integration (such as Bagging or Boosting) can be used to further optimize. Finally, the trained foreign object risk level scoring model is deployed to edge devices or background servers to realize fast risk scoring of newly identified foreign objects for subsequent cleaning strategy driving and task scheduling.

[0071] The pre-set cleaning strategy library can refer to a set of parameter rules defined and stored in advance to guide the inspection robot to perform foreign object cleaning operations.

[0072] The first foreign object operation requirement can refer to a specific set of cleaning parameters and control requirements selected from the cleaning strategy library based on the current identification information and risk level score for each foreign object. Including the specified cleaning method (such as air jet / adsorption / brushing); the required angle adjustment range of the end effector; the safe approach distance; the cleaning intensity, cleaning time; the requirement for robot posture stability (such as whether to slow down and stabilize the landing); whether the cleaning can be completed in stages (for example, first dust removal and then adsorption).

[0073] The inspection robot can be a mobile intelligent robot platform in the system, which has the capabilities of automatic navigation, visual recognition, foreign object cleaning, and path planning.

[0074] The first current position can refer to the current spatial position coordinates of the inspection robot measured based on a navigation positioning system (such as GPS, SLAM, or odometer + laser radar fusion positioning) when the cleaning task instruction is received.

[0075] The first attitude angle can refer to the direction state or attitude information of the robot at this time, such as pitch angle, yaw angle, and roll angle (Pitch, Yaw, Roll), which is used to describe the current orientation of the robot.

[0076] The first motion state information can refer to the dynamic parameter information of the inspection robot at present, including the current speed (linear speed and angular speed), the current acceleration, whether the current is in the state of turning, accelerating, or decelerating, the current path curvature or driving stability.

[0077] The cleaning path can refer to the three-dimensional space path trajectory planned between the current position of the robot (the first current position) and the position of the foreign matter (the first three-dimensional space coordinates) based on requirements such as obstacle avoidance, shortest path, and safe distance. The path is calculated by the navigation system based on path planning algorithms such as A*, D*, RRT, PRM, and Bezier curve fitting.

[0078] The first cleaning execution parameter can refer to a set of specific cleaning action parameters that the inspection robot needs to perform before and after reaching each foreign matter, which is used to control the cleaning executor to complete the cleaning task. It can include: cleaning method (jet, clamping, adsorption, etc.), cleaning trigger distance and trigger angle, cleaning execution intensity (torque, suction force, pressure, etc.), action duration (such as holding the cleaning state time), attitude adjustment angle, and action stability parameters (such as speed change curve, buffer action, etc.).

[0079] After the first foreign object risk level score of each foreign object is completed, the system performs a cleaning decision matching process according to the score results and a preset cleaning strategy library. The cleaning strategy library is a structured knowledge base, usually organized in the form of a rule table, a decision tree, or a conditional mapping function, which defines the first foreign object operation requirements corresponding to different risk levels and foreign object types. The system first extracts the corresponding first foreign object risk level score for each foreign object, and combines the type identification of the foreign object (such as plastic film, branches, metal cable, etc.) to find matching rules in the strategy library. In the strategy matching process, the system uses a multi-condition logical rule matching mechanism: first, coarse classification is performed according to the risk level of the foreign object (high / medium / low), and then refined processing is performed in combination with the visual features and spatial position features of the foreign object. To improve decision accuracy, the system also introduces a fuzzy logic scoring mechanism to calculate the membership degree of the first foreign object risk level score and the risk domain in the strategy library, ensuring the flexibility of the decision under boundary conditions. Finally, each foreign object is assigned to a complete first foreign object operation requirement. In order to obtain the first current position, the first attitude angle, and the first motion state information of the inspection robot, the system uses a fusion positioning perception method, combining an inertial measurement unit (IMU), a high-precision GNSS positioning device, and an odometer to obtain preliminary motion and position data. In static areas or indoor environments where GNSS signals are not available, the system further collects laser radar point cloud data carried by the robot and the internal environment map of the substation to perform laser SLAM positioning. By matching the feature point distribution between the current laser scan frame and the constructed map, accurate estimation of the first current position is achieved. At the same time, the IMU three-axis gyroscope and accelerometer are used to record the rotational speed and linear acceleration of the robot body in real time, and the Kalman filter or extended Kalman filter algorithm is used to fuse the IMU data with the visual / radar positioning results to calculate the attitude Euler angle or quaternion information of the robot, i.e. the first attitude angle. In addition, the system extracts the linear velocity vector and acceleration vector information of the robot at the current position based on the speed and acceleration data recorded by the IMU and the wheeled encoder, forming complete first motion state information. Based on the first current position and the first three-dimensional spatial coordinates of the target foreign object, path planning algorithms such as A* (A-Star) or RRT* (Rapidly-exploring Random Tree Star) are used to generate a cleaning path for the inspection robot to reach the target foreign object. In the path generation process, not only obstacle avoidance and path smoothness are considered, but also attitude adjustment nodes are reserved to meet the end attitude accuracy requirements of the cleaning operation. Then, the system calculates the attitude deviation based on the current first attitude angle and the spatial direction of the target foreign object, and uses inverse kinematics algorithm to calculate the attitude adjustment angle required for the cleaning task, ensuring that the robot can implement the operation on the foreign object in the correct direction.On this basis, the system further generates first speed control parameters that meet dynamic constraints by using methods such as quintic polynomial trajectory interpolation or speed-acceleration constraint planning algorithms, according to the speed and acceleration data contained in the first motion state information and the speed limit and operation stability requirements in the first foreign object operation requirements, to ensure that the robot has stability and safety when approaching the target foreign object. Finally, the system generates detailed cleaning action instructions, including the motion trajectory of the end cleaner, the force, the execution duration, and other control parameters, by using strategies such as impedance control or force-position hybrid control, according to the obtained attitude adjustment angle and first speed control parameters, combined with the force control, action range, and trigger mechanism in the first foreign object operation requirements. The system packages the above path planning results, attitude adjustment sequence, and cleaning action together as first cleaning execution parameters, and sends them to the inspection robot together with the cleaning path to execute the cleaning task.

[0080] The cleaning completion information can be a state signal or data packet fed back by the inspection robot to the control system after completing the cleaning task for a certain area or specific foreign object.

[0081] The second current position can be the three-dimensional space coordinates or two-dimensional plane coordinates of the robot, indicating the specific position of the robot when it ends the cleaning task. It is usually calculated by fusing GNSS, laser radar SLAM, inertial navigation system, etc.

[0082] The second attitude angle can be the orientation or attitude of the robot at that position, usually represented by Euler angles (pitch angle, yaw angle, roll angle) or quaternions. It reflects the inclination and direction of the robot relative to the ground or equipment, which helps to evaluate the safety and feasibility of the next action.

[0083] The second motion state information can be dynamic information such as linear velocity, angular velocity, and acceleration of the robot, describing the current motion trend of the robot. It is very critical for planning the next path and action execution, ensuring smooth and safe action.

[0084] The spatial topology can be an abstract representation of the spatial relationship and connection method of various devices, facilities, channels, and obstacles in the substation. It includes the distance between devices, passable paths, connectivity between nodes, and other information, usually stored in the form of topological graphs or network structures. The spatial topology provides constraint conditions for path planning algorithms to prevent the robot from entering impassable areas or colliding.

[0085] Safety constraints can refer to safety regulations and restrictions that must be adhered to by the substation environment and robot operations. Examples include electrical safety distance requirements (minimum safe distance between the robot and high-voltage equipment), speed limits (limiting maximum speed within a specific area), load and operating force limits (preventing damage to equipment or the robot itself), restricted access areas or time windows (access to certain areas is limited to specific times), and environmental state restrictions (e.g., restricting movement in rainy or snowy weather).

[0086] An inspection path is a specific route planned for a robot within a substation, including its starting point, end point, waypoint coordinates, and movement method. This path is typically generated based on spatial topology and safety constraints, aiming to efficiently cover all electrical equipment areas requiring inspection.

[0087] Inspection actions can be specific inspection or operation tasks performed by the robot on the inspection target equipment, which may include visual photography and image acquisition, infrared thermal imaging detection, vibration, temperature, current, voltage and other sensor data acquisition, sound or gas leak detection, abnormal alarm and data upload.

[0088] Upon receiving the inspection robot's cleaning completion message, the system first utilizes the robot's integrated navigation system (including an inertial navigation unit (IMU), lidar SLAM, and visual odometry) to obtain its latest state data. This system calculates and obtains the robot's second current position (i.e., its absolute position in three-dimensional space after cleaning), second attitude angles (including pitch, yaw, and roll), and second motion state information (including linear velocity, acceleration, and angular velocity) in real time. This data constitutes the robot's kinematic state and serves as the initial constraint boundary for path planning and motion control. The system then loads and analyzes the substation's spatial topology, which is represented as a directed graph. Nodes represent electrical equipment, walls, and passageways, and edges represent traversable paths. Edges are weighted based on factors such as path length and obstacle risk level, forming a weighted graph model. Based on this, the system uses an improved heuristic A or RRT algorithm combined with a heuristic function to rapidly calculate multiple candidate paths, ensuring that path planning meets the shortest path requirements while also considering travel safety and risk minimization.

[0089] At the same time, the system extracts spatial layout information from CAD or three-dimensional BIM models, and obtains the geometric dimensions, position coordinates, spatial occupation volume, and surface normal direction of each device and obstacle in detail. Through the voxel gridding method, the space is divided into passable and impassable cells to form a three-dimensional voxel map. Combined with the current second attitude angle of the robot, the sensor field of view range and motion direction of the robot are calculated through coordinate transformation by rotation matrix, to ensure that the geometric dimensions and turning restrictions of the robot are fully considered in path planning, and to avoid collision with devices or obstacles caused by path design.

[0090] In addition, the system analyzes the current motion stage (acceleration, constant speed, or deceleration) of the robot according to the second motion state information provided by the robot. If the acceleration exceeds the preset threshold, a buffer path segment is automatically inserted during planning to reduce the risk of motion, and a quintic polynomial trajectory planning is used to smooth the speed adjustment, to ensure that the robot maintains a low-speed stable state when performing fine inspection actions, thereby improving the data collection quality.

[0091] The system further loads and applies safety constraint conditions formulated based on electrical safety regulations of the substation, including minimum safety distance in high-voltage areas, forbidden area identification (such as the area around the knife switch), heat source and electromagnetic interference area, etc. These areas are mapped as unreachable areas and high-risk areas in three-dimensional space through spatial Boolean operation, which are strictly avoided during path planning and included as hard constraints in the path search algorithm.

[0092] Based on all the above information, the system generates the final path according to the following steps: taking the second current position and the second attitude angle as the starting point, combining the spatial topological structure and safety constraint conditions, and using the heuristic A* algorithm to generate a preliminary path set that meets the safety restrictions. According to the device geometry and spatial occupation data in the spatial layout information, combined with the robot attitude adjustment field of view and motion range, the path candidates that may cause collision are removed. The dynamic constraints in the second motion state information are introduced to adjust the speed curve between path points, to ensure smooth motion of the robot and suitability for subsequent action execution. Through hierarchical risk filtering, the path with the lowest risk and meeting the operation regulations is preferentially selected, taking into account the path length and inspection efficiency. The generated final inspection path ensures that the robot reaches the target electrical device position in the safest and most compliant manner. The system then matches the corresponding inspection action template from the action database according to the device type and inspection standard, such as infrared imaging, image recognition, sound detection, and temperature and humidity collection, etc. Combined with the second attitude angle of the robot and the device orientation, the sensor adjustment angle and working distance are accurately calculated to ensure the effectiveness of action execution and data quality.

[0093] Finally, the system encodes the determined inspection path and corresponding inspection action into a structured control instruction, which is issued to the inspection robot through an industrial wireless communication network to realize closed-loop control of path navigation and action collection. The robot completes the autonomous inspection task according to the instruction and feeds back the execution state in real time to realize dynamic scheduling and safety guarantee.

[0094] In the embodiments of the present application, the system realizes clear division of labor and accurate control in the stages of foreign matter identification, spatial positioning, risk assessment, cleaning instruction generation, and post-cleaning inspection, improves the expansibility and overall response efficiency of the system, and thus guarantees efficient, safe, and stable execution of the autonomous inspection and foreign matter cleaning tasks of the transformer substation.

[0095] On the basis of the above technical solution, optionally, the cleaning execution module is further configured to:

[0096] According to the first current position and the first three-dimensional space coordinate, a path planning algorithm is used to calculate a sub-cleaning path of the inspection robot from the first current position to each foreign matter, and a cleaning path of the inspection robot is determined according to the sub-cleaning path of the inspection robot from the first current position to each foreign matter.

[0097] According to the first attitude angle and the first three-dimensional space coordinate, a first end attitude adjustment angle of the inspection robot for cleaning each foreign matter is calculated.

[0098] According to the first motion state information and the first foreign matter operation requirement, a first speed control parameter of the inspection robot in the approaching section of each foreign matter is determined.

[0099] According to the first end attitude adjustment angle, the first speed control parameter, and the first foreign matter operation requirement, a first cleaning execution parameter of each foreign matter is determined.

[0100] In the present solution, the path planning algorithm refers to an algorithm for calculating a travel path that meets the requirements (such as the shortest, safe, obstacle avoidance, and satisfaction of constraint conditions) from the current position of the inspection robot to a series of space points (such as multiple foreign matter positions). It includes: A* (A-star) algorithm: combines heuristic function and cost function to calculate the optimal path from the starting point to the target point. Dijkstra algorithm: based on graph structure, calculates the shortest path from the starting point to all points, suitable for global optimal path search without heuristic. RRT: suitable for path exploration in unstructured space in dynamic environment. TSP variant algorithm: for path optimization of multiple foreign matter points (i.e. multi-objective path), combined with constraints, genetic algorithm and ant colony algorithm can be used for optimization.

[0101] The sub-cleaning path can refer to the minimum cost path between the current pose (first current position) of the inspection robot and the three-dimensional space coordinate (first three-dimensional space coordinate) of a certain target foreign matter, which is a local cleaning path for each foreign matter.

[0102] The first end pose adjustment angle can be an angle value that needs to be adjusted by the robot execution arm or end effector before approaching the foreign matter, for accurately aligning the spatial pose of the foreign matter.

[0103] The first speed control parameter can refer to a set of speed and acceleration parameters that need to be followed by the inspection robot when approaching the foreign matter, and can include a uniform approach speed (linear speed), a maximum acceleration, a deceleration threshold (triggered according to the approach distance), and a jitter threshold (for pose stabilization).

[0104] The foreign matter approach section can refer to a small path area from the last navigation point on the path to the actual position of the foreign matter, and is usually located at the end of the cleaning path.

[0105] Based on the first current position of the inspection robot and the first three-dimensional spatial coordinates of each foreign matter, the system uses a multi-objective path optimization strategy to calculate the sub-cleaning paths from the current point to the positions of multiple foreign matters. In the path generation process, the system preferentially uses an improved A* path planning algorithm that combines the spatial topological structure and the kinematic model of the robot to calculate the shortest distance and dynamic constraints of each path section, and introduces path risk level and safety constraint information as a discriminant condition. In addition, for the order optimization of multiple foreign matters, the system combines the Traveling Salesman Problem (TSP) heuristic algorithm to combine multiple sub-cleaning paths to generate a global cleaning path that satisfies the shortest path length, minimum energy consumption, and reasonable cleaning priority, ensuring the overall efficient execution of the cleaning task.

[0106] Subsequently, the system determines the current execution mechanism state using the forward kinematics method according to the first pose angle of the inspection robot and the first three-dimensional spatial coordinates of the target foreign matter, and combines the inverse kinematics solution to calculate the first end pose adjustment angle required for the end effector or cleaning tool of the robot arm to reach each foreign matter cleaning point. This angle is expressed in Euler angles or quaternion form, and in the calculation process, the multi-degree-of-freedom redundant solution space of the robot is fully considered to select a feasible solution that is most stable in operation pose and avoids mechanical interference, ensuring normal contact and smooth operation of the cleaning tool and the foreign matter surface.

[0107] Next, the system synthesizes the current first motion state information of the inspection robot, including linear velocity, acceleration, and acceleration rate of change, and the specific job requirements of each foreign object, such as cleaning intensity, approach distance, and pose accuracy, etc., and generates first speed control parameters for each foreign object approach segment using a dynamic speed adjustment algorithm (such as a PID control algorithm for trajectory tracking or a speed planning based on model predictive control (MPC)). The parameters ensure that the robot achieves smooth deceleration when approaching the cleaning target, meets the requirement that the end speed approaches zero, while ensuring that the spatial accuracy and safety distance constraints in path planning are not destroyed. If there are specific cleaning stability or speed limitations, the system further adjusts the speed gradient using an S-shaped speed curve to reduce robot vibration and running impact.

[0108] Subsequently, the system inputs the first end pose adjustment angle calculated above into the inverse kinematics model to solve the specific pose trajectory of the end effector of the inspection robot during the cleaning action, ensuring that the end tool direction is aligned with the foreign object surface normal as much as possible, thereby improving the accuracy and stability of the cleaning job. Combined with the first speed control parameters, the system plans the velocity and acceleration curve of the end effector to ensure smooth and safe action process. The training process of the inverse kinematics model usually includes the following steps: first, collect a large amount of end effector spatial position and pose data of the robot arm under different joint angle configurations to build a high-quality mapping dataset between joint angle and end position; then, select a suitable machine learning model (such as neural network, support vector machine, or deep learning model based on gradient descent) to train the dataset, continuously adjust the model parameters to improve the prediction accuracy by minimizing the prediction error of the end position and pose; during the training process, constraint optimization is combined with physical constraints (such as joint angle limits and mechanical arm structure geometric relationships) to ensure that the output inverse solution meets the kinematic feasibility of the mechanical arm; finally, the generalization ability and stability of the model are verified through cross-validation and actual robot action testing to form an efficient prediction model that can be used for real-time inverse kinematics solving.

[0109] In addition, the system analyzes the first foreign object job requirements of each foreign object in detail, covering cleaning method types (such as mechanical brushing, air blowing, or adsorption), action distance, cleaning intensity, action trigger conditions, and action duration, etc. Through logical combination and condition matching, the system synthesizes the pose adjustment and speed control results to form complete first cleaning execution parameters. Specifically, it includes cleaning instruction sequence, end pose trajectory planning, speed and force control range, action start and stop conditions, and execution time length, ensuring that the robot performs foreign object cleaning tasks accurately, stably, and efficiently.

[0110] In the scheme, the path planning of the inspection robot and the cleaning action parameters are designed step by step, so that the system can adaptively generate an optimal cleaning path and execution parameters according to the spatial position of the foreign matter and the current state of the robot, improve the cleaning accuracy and stability in the complex substation environment, and have stronger robustness and dynamic adjustment capability.

[0111] On the basis of the above technical scheme, optionally, the inspection execution module is further used for:

[0112] determining the inspection path of the inspection robot according to the second current position and the spatial topology structure;

[0113] determining the sensing angle setting parameter of the inspection robot reaching the position of each electrical equipment in the inspection path according to the second attitude angle and the spatial layout information;

[0114] determining the path geometric feature according to the spatial topology structure and the spatial layout information, and determining the second speed control parameter of the inspection robot in the approach section of each electrical equipment according to the second motion state information, the path geometric feature and the safety constraint condition;

[0115] determining the inspection action of the inspection robot on each electrical equipment according to the sensing angle setting parameter and the second speed control parameter.

[0116] In the scheme, the sensing angle setting parameter can be a control parameter for adjusting the orientation angle (such as pitch angle, yaw angle and roll angle) of the sensor or camera of the inspection robot when performing sensing or detection tasks on a certain electrical equipment to ensure the best viewing angle and obtain effective image, thermal imaging, infrared or other sensing data.

[0117] The path geometric feature can be the geometric properties of each section of the inspection path, including but not limited to the curvature, slope, corner radius, path width, obstacle gap, turning radius, terrain elevation difference and other geometric data of the path.

[0118] The second speed control parameter can be a speed control value set in the inspection process to ensure that the robot approaches each electrical equipment stably, safely and efficiently, which usually includes forward speed, turning speed, deceleration and other dimensions.

[0119] The second current position of the inspection robot, i.e. the current three-dimensional spatial coordinates (x, y, z), is obtained, combined with the spatial topology structure of the substation, which is usually represented as a directed graph with device nodes as vertices and passable paths as edges. Based on this graph structure, the A* (A-star) algorithm or Dijkstra algorithm is used for path search, taking the second current position as the starting point and the positions of each electrical device to be inspected as target points in the path, to calculate a shortest inspection path that meets the passable constraints. The path will be represented in the form of a sequence of consecutive coordinate points, where each point represents the position reached by the robot in turn.

[0120] Subsequently, according to the second attitude angle (e.g. pitch angle, yaw angle, roll angle) and the spatial layout of each target device, an angle transformation calculation is performed to determine the relative attitude angle difference between the current attitude of the robot and the observation direction of the target device using a coordinate transformation matrix (rotation matrix and transformation matrix). On this basis, the Euler angle difference fitting or quaternion interpolation method (such as SLERP) is used to calculate the perception angle setting parameters that the robot should adopt in front of each device, i.e. the angle combination required to adjust the sensor to face the device for view collection.

[0121] Next, the geometry of each segment of the generated inspection path is analyzed, and the path geometry features of each segment are extracted based on the curvature, turning radius, channel width, obstacle distribution, and height difference of each segment on the path. This step is usually completed by curvature fitting (such as spline interpolation fitting) and sliding window method for curve feature extraction on the path point set. At the same time, according to the spatial layout information (representing the space occupation, shielding relationship, and device size of the device), the topological graph constraints are fused to determine whether there is shielding or passable conflict between devices.

[0122] Next, the above path geometry features are combined with the second motion state information (representing the current speed, acceleration, and angular velocity of the robot) and the pre-set safety constraints of the substation (such as minimum safety distance from high-voltage devices, maximum allowed speed, and safe obstacle avoidance radius), and a speed planning algorithm (such as model predictive control MPC or dynamic window method DWA) is used to calculate the optimal travel speed on each segment of the device approach path, obtaining the corresponding second speed control parameters, which are used to ensure that the robot can safely and stably approach the device without collision or observation deviation.

[0123] During the execution of the inspection path, the system calculates the orientation adjustment angle of the robot-mounted sensor (such as a camera or infrared thermal imager) based on the perception angle setting parameters corresponding to each target device position, and uses a pose solving algorithm (such as a rotation matrix based on Euler angles or quaternion interpolation) to control the rotation of the robot gimbal or mechanical arm to align its sensor with the key monitoring area (such as terminals, connection points, nameplates, etc.) of the target electrical device.

[0124] Meanwhile, according to the second speed control parameter corresponding to the path segment, the system plans the movement speed of the robot when approaching the electrical equipment to ensure that the image acquisition or infrared measurement is carried out in a stable and non-jittering state. If the equipment risk level is high or the surrounding environment is disturbed, the controller will perform deceleration, residence or dynamic slow operation according to the speed control parameter to ensure the data acquisition quality.

[0125] Finally, the system comprehensively perceives the angle setting parameter and the speed control parameter to generate a specific inspection action sequence, which includes posture adjustment actions (such as rotating the gimbal to the set angles a, b, g), sensor activation actions (such as starting image acquisition, infrared imaging, sound recording, etc.), motion control actions (such as slowing down to a distance L and stopping, keeping still for T seconds), abnormality judgment actions (such as image sharpness evaluation, temperature stability judgment), and encapsulating these action instructions into a control instruction package and sending it to the robot execution controller.

[0126] In this scheme, the path planning, posture adjustment, speed control and action decision are finely decoupled and optimized in parallel to ensure that the inspection robot completes the precise perception and data acquisition of the electrical equipment at the optimal angle and speed under the premise of safety, efficiency and stability, and improves the coverage, accuracy and robustness of the inspection.

[0127] On the basis of the above technical scheme, optionally, the system further comprises a cleaning retry module, the cleaning retry module is used for:

[0128] If the cleaning failure information sent by the inspection robot is received, the second three-dimensional space coordinates of the failure foreign matter and the abnormal type are determined;

[0129] It is determined whether the second three-dimensional space coordinates and the abnormal type meet the preset cleaning retry condition, and if the preset cleaning retry condition is met, the first posture angle and the first motion state information of the inspection robot are updated;

[0130] The first end posture adjustment angle of the inspection robot for cleaning the failure foreign matter is updated according to the updated first posture angle and the second three-dimensional space coordinates of the failure foreign matter;

[0131] The first speed control parameter of the inspection robot in the foreign matter approaching segment of the failure foreign matter is updated according to the updated first motion state information and the first foreign matter operation requirement;

[0132] The first cleaning execution parameter of the failure foreign matter is updated according to the updated first end posture adjustment angle, the updated first speed control parameter and the first foreign matter operation requirement, and the updated first cleaning execution parameter of the failure foreign matter is sent to the inspection robot;

[0133] Correspondingly, the system further comprises a cleaning success module, the cleaning success module being configured to:

[0134] If the cleaning success information sent by the inspection robot is received, a continue cleaning instruction is sent to the inspection robot, so that the inspection robot performs cleaning of the next foreign matter according to the continue cleaning instruction.

[0135] Correspondingly, the system further comprises a cleaning failure module, the cleaning failure module being configured to:

[0136] If the cleaning failure information sent by the inspection robot is received, the second three-dimensional space coordinates of the failed foreign matter are sent to the control center.

[0137] In the present solution, the cleaning failure information can refer to feedback data returned by the inspection robot during the execution of a cleaning task of a certain foreign matter due to execution failure (such as mechanical action abnormality, positioning deviation, and substandard cleaning effect).

[0138] The second three-dimensional space coordinates can refer to the three-dimensional position coordinates (x, y, z) of the failed foreign matter after cleaning failure, which are obtained by repositioning of the system or inverse calculation using the current pose of the robot.

[0139] The abnormality type can refer to the classification result of the system for the cleaning failure reason, which is used for subsequent decision support. It can include mechanical execution abnormality (such as mechanical arm jamming and grabbing failure), target positioning deviation (such as point cloud drift), environmental interference (such as occlusion and strong reflection), and foreign matter change (such as displacement and fragmentation).

[0140] The preset cleaning retry condition can refer to a set of rules defined in advance in the system that can trigger cleaning retry, including that the risk level of the foreign matter is higher than a threshold, the current failure reason is recoverable (such as angle deviation and insufficient cleaning intensity), the number of attempts is not exceeded, and the surrounding environment allows re-action (no new obstacles).

[0141] The cleaning success information can refer to the execution success feedback information returned by the inspection robot after completing the cleaning operation of a certain foreign matter. It includes a success flag, a comparison result of image / point cloud features after cleaning, cleaning time / energy consumption data, and a success timestamp.

[0142] The continue cleaning instruction can refer to a task scheduling instruction sent by the system to the inspection robot according to the current execution state (such as that a certain foreign matter has been successfully cleaned), which is used to execute the cleaning path and action of the next foreign matter.

[0143] The control center can refer to a central control system or a remote management platform in the system.

[0144] Upon receiving the cleaning failure information sent by the inspection robot, the target foreign matter identifier, cleaning action execution feedback, execution timestamp, mechanical arm end effector state, robot's current pose data, and image feedback results are extracted from the failure record. Through the combination of pose inverse calculation and point cloud coordinate projection reconstruction technology at the time of failure, the system repositions the target foreign matter in the second three-dimensional space coordinates in the current environment. Based on the operation state analysis in the abnormal feedback (such as the closing state of the gripper, the cleaning arm force sensor response, and the motor torque feedback), combined with the image / point cloud residual feature change comparison (such as whether there are still foreign matter edge features, whether the outline is complete, and whether the point cloud density rebounds, etc.), the abnormal type discrimination rule set (such as insufficient cleaning force, position deviation, and obstruction occurrence) is used to determine the current abnormal type. A set of pre-set cleaning retry conditions are introduced, and if the above cleaning retry conditions are met, the recovery strategy is executed. The system adjusts the first pose angle based on the current environmental state, the spatial relationship between the robot's end pose and the target foreign matter, and uses forward kinematics and inverse algorithm to calculate a new pose vector suitable for the current obstacle environment. Combined with the current motion state of the inspection robot (including joint angular velocity, vehicle body speed, etc.) and the first foreign matter operation requirements (such as cleaning force requirement, contact stability, and angle range), the system uses Bayesian optimization strategy or rule-based speed constraint function to recalculate the first speed control parameter of the robot in the failure foreign matter approach section, ensuring the motion stability and positioning accuracy within the obstacle avoidance range. According to the updated first end pose adjustment angle and the first speed control parameter, combined with the cleaning action trajectory, cleaning method (such as wiping, blowing, and grabbing), and execution duration set in the first foreign matter operation requirements, the first cleaning execution parameter of the target foreign matter is reconstructed to form an action package containing pose target, speed trajectory, action mode, and fault tolerance threshold, and the cleaning execution parameter is sent to the inspection robot for cleaning retry of the target foreign matter.

[0145] If the subsequent system receives cleaning success information, the next foreign matter task is taken out from the cleaning task scheduling queue, and a new continue cleaning instruction is generated according to the spatial position, operation level, and current position of the robot, which contains target number, path information, estimated execution time, etc. for the robot to complete the next foreign matter cleaning process.

[0146] If the cleaning failure information is received again, and the abnormal type or spatial state does not meet the cleaning retry conditions, the system packages the second three-dimensional space coordinates of the failure foreign matter, related abnormal reasons, image data, and failure level into an abnormal report, which is sent to the control center through TCP communication or MQTT message pushing method to trigger manual review or remote intervention process.

[0147] The scheme improves the cleaning task success rate of the robot in a complex scene, avoids task interruption caused by small range deviation or slight disturbance, ensures safe and reliable execution of the task in an automatic closed loop, and enhances the robustness and autonomous decision-making ability of the system.

[0148] Embodiment Two

[0149] Figure 2 is a structural schematic diagram of the transformer substation autonomous inspection and foreign matter cleaning cooperative control system provided by Embodiment Two of the present application. As shown in Figure 2 , it specifically comprises the following:

[0150] The system further comprises a residual identification module 106, which is configured to:

[0151] reacquire the two-dimensional image data, laser radar point cloud data and multi-modal sensor data of the transformer substation, calculate a difference map for the two-dimensional image data before and after cleaning according to a structural similarity algorithm, perform connected region analysis on the difference map, and determine an image disturbance region;

[0152] perform spatial registration processing on the laser radar point cloud data before and after cleaning, and determine a point cloud disturbance region;

[0153] verify the spatial consistency of the image disturbance region and the point cloud disturbance region, and if they are consistent, confirm the image disturbance region as a candidate residual foreign matter region;

[0154] perform a target detection operation on the candidate residual foreign matter region, and identify whether there is residual foreign matter.

[0155] In this embodiment, the structural similarity algorithm can be an image quality evaluation method for measuring the similarity of two images in terms of structure, brightness and contrast. Compared with simple pixel difference, SSIM is closer to the perception of image changes by the human eye.

[0156] The difference map can be an image difference value representation generated by comparing the two-dimensional image data before and after cleaning, for highlighting the positions in the image that have changed.

[0157] The image disturbance region can refer to a continuous change region extracted by connected region analysis in the difference map, representing the position in the image that has been significantly disturbed.

[0158] The point cloud disturbance region can be a position region where the point cloud structure has changed significantly after comparing the laser radar point cloud data before and after cleaning through spatial registration (such as ICP algorithm).

[0159] The candidate residual foreign matter region can be the intersection region of the image disturbance region and the point cloud disturbance region that is highly consistent in space, indicating that the position has changed in both the image and the physical space.

[0160] Residual foreign matter can refer to target foreign matter that is not completely removed after performing the cleaning task.

[0161] The two-dimensional image data, lidar point cloud data, and multi-modal sensor data of the substation can be reacquired, and image comparison uses a structural similarity algorithm that compares local window regions of images block by block based on three components of brightness, contrast, and structure, and outputs a similarity score map. The similarity score map is processed by pixel inversion and thresholding to generate a difference map representing locations in the image where there are large structural changes. The system uses a connected region analysis algorithm to cluster pixels in the difference map, identifying contiguous high-difference regions that are extracted as candidate image disturbance regions representing areas in the image that may not have been cleaned or have undergone physical structural changes. Registration is performed on lidar point cloud data collected before and after cleaning, using algorithms such as ICP or improved algorithms such as NDT or Go-ICP, to align the two sets of point cloud data in three-dimensional space and compare features such as local point density changes and geometric structure differences in the registration results to extract areas with significant changes as point cloud disturbance regions. Spatial consistency verification is then performed: the image disturbance region is mapped to three-dimensional space through coordinate transformation, and spatial overlap detection is performed with the point cloud disturbance region. If there is a high degree of overlap in space (e.g., center distance less than a predetermined threshold, IoU greater than a certain threshold), the region is considered a credible candidate residual foreign matter region. The system uses deep learning methods to perform object detection on the above candidate regions, common methods including YOLOv5, Mask R-CNN, or RetinaNet, with the input being the original image crop segment within the candidate region and the output being the detection result of whether there is residual foreign matter and the target box position and confidence. If the detection result indicates the presence of residual foreign matter, the information is labeled and reported to the system control center, triggering subsequent compensation cleaning or manual review operations.

[0162] In this embodiment, the accuracy and robustness of residual foreign matter detection are improved; combining structural similarity algorithms and spatial registration techniques, the system can sensitively capture minor disturbances, ensuring timely detection of cleaning omissions or environmental abnormalities, and enhancing the system's autonomous monitoring capabilities and safety assurance level.

[0163] On the basis of the above technical solutions, the system can further include a local cleaning module, which is configured to:

[0164] If there is a residual foreign matter, a third three-dimensional space coordinate of the residual foreign matter and a second foreign matter visual feature description are acquired, the third three-dimensional space coordinate, the second foreign matter visual feature description, the spatial layout information and the re-acquired multi-modal sensor data are input into a preset foreign matter risk level scoring model to obtain a second foreign matter risk level score of the residual foreign matter;

[0165] A second foreign matter operation requirement of each foreign matter is determined according to the second foreign matter risk level score and a preset cleaning strategy library;

[0166] A third current position, a third attitude angle and third motion state information of the inspection robot are acquired, and a local cleaning path of the inspection robot and a second cleaning execution parameter for the residual foreign matter are determined according to the third three-dimensional space coordinate, the third current position, the third attitude angle, the third motion state information and the second foreign matter operation requirement, and the local cleaning path and the second cleaning execution parameter for the residual foreign matter are sent to the inspection robot.

[0167] In the scheme, the third three-dimensional space coordinate can refer to a residual foreign matter re-identified after cleaning, and a three-dimensional coordinate of the residual foreign matter in the substation space is obtained through image and point cloud fusion registration positioning.

[0168] The second foreign matter visual feature description can refer to visual feature information extracted from a two-dimensional image collected after cleaning of the residual foreign matter, such as a color histogram, a texture feature (such as LBP), a depth feature (such as a CNN extracted embedding vector), etc. The description is used for comparison with historical foreign matter features, risk analysis and decision scheduling.

[0169] The second foreign matter risk level score can refer to a risk level value of the residual foreign matter predicted by using a preset foreign matter risk level scoring model based on the third three-dimensional space coordinate, the second foreign matter visual feature description, the spatial layout information and the multi-modal sensor data. It is used to measure whether it has a safety hazard or an emergency cleaning priority.

[0170] The second foreign matter operation requirement can be an operation specification required by the residual foreign matter mapped based on the second foreign matter risk level score and the preset cleaning strategy library.

[0171] The third current position can be a position of the robot in a substation coordinate system.

[0172] The third attitude angle can be, for example, Euler angles or quaternions, representing the overall direction of the robot.

[0173] The third motion state information can be, for example, a velocity vector, acceleration information, etc., reflecting the dynamic state of the robot.

[0174] The local cleaning path may be a short-distance path planned from the third current position to the third three-dimensional space coordinate, taking into account obstacle avoidance, safety margins, etc.

[0175] The second cleaning execution parameter may be a cleaning action parameter generated according to the third posture angle, the third motion state information and the second foreign object operation requirement, such as the end effector trajectory, speed, force control parameters, etc.

[0176] When the system detects that the image disturbance region and the point cloud disturbance region have consistency in space, and further confirms the existence of residual foreign matter through the target detection algorithm, the system first uses a spatial reconstruction algorithm (such as multi-view geometry-based triangulation or point cloud segmentation positioning algorithm) to locate the residual foreign matter, and combines the correspondence between the visual region center point extracted from the cleaned image and the local cluster region of the laser radar point cloud to estimate and calculate the third three-dimensional spatial coordinates of the foreign matter in the substation three-dimensional coordinate system. At the same time, the visual description information of the foreign matter region is extracted from the cleaned image, and a second foreign matter visual feature description is constructed. The feature vector can be represented by the intermediate layer embedding vector output by the pre-trained convolutional neural network (such as ResNet or EfficientNet), which captures color, edge, texture, and shape information. Using the previously constructed foreign matter risk level scoring model, the third three-dimensional spatial coordinates, the second foreign matter visual feature description, the spatial layout information of the current device region (including peripheral high-voltage device coordinates, cable routing, cleanable operation airspace, and other structural data), and the reacquired multi-modal sensor data (such as temperature and humidity, wind speed, infrared radiation, electromagnetic intensity, etc.) are integrated to form a multi-dimensional input feature vector. The vector is then input into the foreign matter risk scoring model, which can use ensemble learning (such as XGBoost) or neural network regression architecture for prediction output, obtaining the second foreign matter risk level score of the residual foreign matter, which can be normalized to the [0,1] interval. According to the mapping relationship between the risk score value and the pre-set cleaning strategy library in the system, the system automatically finds and generates the second foreign matter operation requirements corresponding to the current risk level, which includes the required cleaning mode (such as mechanical wiping, adsorption, jet), force setting, safety boundary restriction, and executor configuration, etc. The system real-time acquires the third current position (obtained by positioning system such as RTK-GPS or UWB+IMU fusion), the third attitude angle (obtained by gyroscope or visual inertial navigation to calculate Euler angle or quaternion), and the third motion state information (such as acceleration vector read by accelerometer, wheel encoder speed, etc.) of the inspection robot at this time. Combined with the third current position and the target third three-dimensional spatial coordinates, and executing a heuristic-based path planning algorithm (such as A*, RRT*, or DWA local planner) in the obstacle avoidance grid map, a local cleaning path with the shortest or optimal cost is planned. In the path planning process, the known obstacle data and safety constraints in the spatial layout are considered, such as maintaining a certain distance from high-voltage equipment, safety operation area restrictions, etc. Further combined with the planned path geometry, the third attitude angle, and the second foreign matter operation requirements, the kinematic model and velocity mapping function of the end effector in the robot controller are called to calculate the required trajectory velocity and attitude transition mode of the cleaning end when approaching the residual foreign matter, and finally the second cleaning execution parameters for controlling the robot are generated, including path segment speed, end attitude adjustment angle, execution action sequence, and execution duration, etc.The system packs the generated local cleaning path and the corresponding second cleaning execution parameters in the form of control instruction sequence, and sends them to the task control unit of the inspection robot, so as to drive the inspection robot to re-visit the residual foreign matter area and complete the secondary cleaning operation according to the path and action requirements.

[0177] In the scheme, the compensation robustness of the cleaning operation, the task closed loop and the system intelligence level are improved, and the continuous cleaning and operation safety of the substation environment are effectively ensured.

[0178] On the basis of the above technical scheme, optionally, the system further comprises an exception reporting module, the exception reporting module is used for:

[0179] If there is no residual foreign matter, edge change detection is performed on the image disturbance area to determine whether there is image structure abnormality;

[0180] If there is image structure abnormality, according to the laser radar point cloud data before and after cleaning, a density change analysis method is used to identify whether there is point cloud distribution abnormality;

[0181] If there is point cloud distribution abnormality, an exception reporting information package is generated according to the re-acquired two-dimensional image data, laser radar point cloud data and multi-modal sensor data of the substation, and the exception reporting information package is transmitted to the control center.

[0182] In the scheme, the exception reporting information package can be a data set containing abnormal information, positioning information and auxiliary analysis content, which is automatically generated in the case that no residual foreign matter is detected, but image or point cloud data has structure abnormality and spatial abnormality, and is used to report abnormal events to the control center, so as to facilitate subsequent manual review or automatic intervention.

[0183] In the case of no residual foreign matter being detected, the system will further analyze the structural abnormalities of the disturbed area of the image. First, using the two-dimensional image data of the substation collected before and after cleaning, an edge detection algorithm (such as the Canny algorithm or edge gradient detection method based on the Sobel operator) is used to extract the edges of the disturbed area. By comparing the contour shape changes of the edges of the images before and after cleaning, it is detected whether there are significant image structural abnormalities such as contour breakage, edge displacement, and closure changes. If the edge change amplitude exceeds the preset threshold, the system determines that the disturbed area has structural abnormalities. The system further performs spatial correspondence processing on the laser radar point cloud data obtained before and after cleaning, and uses the spatial registration technology based on the ICP (Iterative Closest Point) algorithm to align the point clouds before and after cleaning to the same spatial coordinate system. After registration is completed, the point cloud voxel grid (Voxel Grid) downsampling and statistical density analysis method is used to compare the point density in each voxel in space. If the point cloud density of a certain spatial region changes significantly, such as a dense area appearing a hole or a sparse area suddenly becoming dense, it is determined that the region has point cloud distribution abnormalities. In the case where both image structural abnormalities and point cloud distribution abnormalities are confirmed, the system identifies the area as an abnormal area that may have potential safety risks or structural disturbances. Subsequently, the system integrates the two-dimensional image data of the substation reacquired, the laser radar point cloud data, and the current multi-modal sensor data (such as environmental temperature and humidity, vibration, electromagnetic interference level, etc.) to automatically construct a complete abnormal report information package. The information package encapsulates the position coordinates of the image disturbed area, the image structural abnormal edge graph, the point cloud density change heat map, the current sensor data summary, the identified abnormal type label (such as occlusion enhancement, ground uplift, device tilt, etc.), and generates a risk level score in combination with the electrical equipment distribution map. Finally, it is transmitted in real time to the control center through the system communication protocol.

[0184] In this scheme, closed-loop verification of cleaning effect and risk compensation reporting are realized, thereby significantly improving the reliability and safety protection capability of the inspection system.

[0185] Embodiment Three

[0186] Figure 3 is a flowchart of the substation autonomous inspection and foreign matter cleaning cooperative control method provided by the embodiment three of the present application. As shown in Figure 3 , it specifically includes the following steps:

[0187] S301, acquiring two-dimensional image data, laser radar point cloud data and multi-modal sensor data of a substation, and using a convolutional neural network target detection algorithm to identify the two-dimensional coordinate positions and first foreign matter visual feature descriptions of each foreign matter in the two-dimensional image data;

[0188] S302, performing spatial registration processing based on the laser radar point cloud data and the two-dimensional coordinate position to obtain first three-dimensional spatial coordinates of each foreign object;

[0189] S303, obtaining spatial layout information of the electrical equipment, inputting the first three-dimensional spatial coordinates, the first foreign object visual feature description, the spatial layout information, and the multimodal sensor data into a preset foreign object risk level scoring model to obtain a first foreign object risk level score for each foreign object;

[0190] S304: Determine a first foreign object operation requirement for each foreign object based on the first foreign object risk level score and a preset cleaning strategy library, obtain a first current position, a first posture angle, and a first motion state information of the inspection robot, determine a cleaning path for the inspection robot and first cleaning execution parameters for each foreign object based on the first three-dimensional space coordinates, the first current position, the first posture angle, the first motion state information, and the first foreign object operation requirement, and send the cleaning path and the first cleaning execution parameters for each foreign object to the inspection robot;

[0191] S305. If the cleaning completion information of the inspection robot is received, the second current position, second posture angle, and second motion state information of the inspection robot are obtained, as well as the spatial topology structure and safety constraints of the substation are obtained. The inspection path of the inspection robot and the inspection actions on each electrical equipment are determined according to the second current position, second posture angle, second motion state information, spatial topology structure, spatial layout information and safety constraints, and the inspection path and the inspection actions on each electrical equipment are sent to the inspection robot.

[0192] The embodiment of the present application provides a method for coordinated control of autonomous inspection and foreign matter cleaning of substations, which corresponds to the systems provided in the above embodiments and has corresponding execution processes and beneficial effects, and will not be repeated here.

[0193] Example 4

[0194] like Figure 4 As shown, an embodiment of the present application also provides an electronic device 400, including a processor 401, a memory 402, and a program or instruction stored in the memory 402 and executable on the processor 401. When the program or instruction is executed by the processor 401, each process of the above-mentioned embodiment of the method for the coordinated control system of autonomous inspection and foreign matter cleaning of substations is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0195] It should be noted that the electronic devices in the embodiments of the present application include the mobile electronic devices and non-mobile electronic devices mentioned above.

[0196] Example 5

[0197] The embodiment of the present application also provides a readable storage medium, which stores a program or instructions, and the program or instructions are executed by a processor to implement the processes of the above-mentioned cable installation process based on tension self-adaptive control system embodiment, and achieve the same technical effects. To avoid repetition, details are not described herein.

[0198] The processor is the processor in the electronic device in the above-mentioned embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0199] It should be noted that, in this document, the terms "comprising", "including", or any other variant thereof are intended to cover non-exclusive inclusions, so that a process, method, article or system including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or includes elements inherent to such a process, method, article or system. Without more limitations, the element defined by the statement "including a" does not exclude the presence of other identical elements in the process, method, article or system including the element. In addition, it should be pointed out that the scope of the method and system in the embodiments of the present application is not limited to the order of functions shown or discussed, and can also include functions performed in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method can be performed in an order different from that described, and various steps can also be added, omitted or combined. In addition, the features described with reference to certain examples can be combined in other examples.

[0200] From the above description of the embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment method can be realized by means of software and necessary general hardware platform, of course, it can also be realized by hardware, but in many cases, the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a plurality of instructions for making a terminal (which can be a mobile phone, computer, server, or network equipment, etc.) execute the method described in each embodiment of the present application.

[0201] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of this application, ordinary technicians in this field can also make many forms without departing from the purpose of this application and the scope of protection of the claims, all of which are within the protection of this application.

[0202] The above are only preferred embodiments of the present application and the technical principles employed. The present application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions that are possible for those skilled in the art will not depart from the scope of protection of the present application. Therefore, although the present application has been described in more detail through the above embodiments, the present application is not limited to the above embodiments and may include more other equivalent embodiments without departing from the concept of the present application. The scope of the present application is determined by the scope of the claims.

Claims

1. A substation autonomous inspection and foreign matter cleaning collaborative control system, characterized in that: The system comprises: a multimodal detection module, configured to acquire two-dimensional image data, lidar point cloud data, and multimodal sensor data of the substation, and employ a convolutional neural network target detection algorithm to identify the two-dimensional coordinate position of each foreign object in the two-dimensional image data and a first foreign object visual feature description; A spatial positioning module is used to perform spatial registration processing based on the laser radar point cloud data and the two-dimensional coordinate position to obtain the first three-dimensional spatial coordinates of each foreign object; a risk decision module, configured to obtain spatial layout information of the electrical equipment, input the first three-dimensional spatial coordinates, the first foreign object visual feature description, the spatial layout information, and the multimodal sensor data into a preset foreign object risk level scoring model, and obtain a first foreign object risk level score for each foreign object; a cleaning execution module, configured to determine a first foreign object operation requirement for each foreign object based on the first foreign object risk level score and a preset cleaning strategy library, obtain a first current position, a first posture angle, and a first motion state information of the inspection robot, determine a cleaning path of the inspection robot and first cleaning execution parameters for each foreign object based on the first three-dimensional spatial coordinates, the first current position, the first posture angle, the first motion state information, and the first foreign object operation requirement, and send the cleaning path and the first cleaning execution parameters for each foreign object to the inspection robot; wherein the first cleaning execution parameters are a specific cleaning action parameter set of the inspection robot, used to control its cleaning executor to complete the cleaning task, including a cleaning mode, a cleaning trigger distance, a trigger angle, a cleaning execution force, an action duration, a posture adjustment angle, and an action stability parameter; The inspection execution module is used to obtain the second current position, second posture angle, second motion state information of the inspection robot upon receiving the cleaning completion information of the inspection robot, as well as the spatial topology structure and safety constraints of the substation, and determine the inspection path of the inspection robot and the inspection actions on each electrical equipment based on the second current position, second posture angle, second motion state information, spatial topology structure, spatial layout information and safety constraints, and send the inspection path and inspection actions on each electrical equipment to the inspection robot.

2. The substation autonomous inspection and foreign matter cleaning coordinated control system according to claim 1 is characterized in that: The cleaning execution module is further used to: Calculate, based on the first current position and the first three-dimensional spatial coordinates, a sub-cleaning path of the inspection robot from the first current position to each foreign object using a path planning algorithm, and determine a cleaning path of the inspection robot based on the sub-cleaning path of the inspection robot from the first current position to each foreign object; Calculating a first end posture adjustment angle of the inspection robot for cleaning foreign objects according to the first posture angle and the first three-dimensional space coordinate; Determining a first speed control parameter of the inspection robot in each foreign object approach section according to the first motion state information and the first foreign object operation requirement; The first cleaning execution parameter of each foreign object is determined according to the first end posture adjustment angle, the first speed control parameter and the first foreign object operation requirement.

3. The substation autonomous inspection and foreign matter cleaning coordinated control system according to claim 1 is characterized in that: The inspection execution module is further used to: Determining an inspection path of the inspection robot according to the second current position and the spatial topological structure; Determine, based on the second posture angle and the spatial layout information, the perception angle setting parameters for the inspection robot to reach the location of each electrical device in the inspection path; Determine the path geometric characteristics based on the spatial topological structure and spatial layout information, and determine the second speed control parameters of the inspection robot in the approach section of each electrical equipment based on the second motion state information, the path geometric characteristics, and the safety constraint conditions; The inspection action of the inspection robot on each electrical device is determined according to the perception angle setting parameter and the second speed control parameter.

4. The substation autonomous inspection and foreign matter cleaning coordinated control system according to claim 1 is characterized in that: The system further includes a cleaning and retrying module, which is configured to: If a cleaning failure message is received from the inspection robot, the second three-dimensional spatial coordinates of the failed foreign object and the abnormality type are determined; Determine whether the second three-dimensional space coordinates and the abnormality type meet a preset cleaning and retry condition, and if the preset cleaning and retry condition is met, update the first posture angle and the first motion state information of the inspection robot; Update the first end posture adjustment angle of the inspection robot that cleans the failed foreign object according to the updated first posture angle and the second three-dimensional space coordinates of the failed foreign object; updating a first speed control parameter of the inspection robot in a foreign object approaching section of a failed foreign object according to the updated first motion state information and the first foreign object operation requirement; updating the first cleaning execution parameter of the failed foreign object according to the updated first end posture adjustment angle, the updated first speed control parameter, and the first foreign object operation requirement, and sending the updated first cleaning execution parameter of the failed foreign object to the inspection robot; Accordingly, the system further includes a cleaning success module, which is configured to: If a cleaning success message is received from the inspection robot, a continue cleaning instruction is sent to the inspection robot, so that the inspection robot can clean the next foreign object according to the continue cleaning instruction; Accordingly, the system further includes a cleaning failure module, which is configured to: If the cleaning failure information sent by the inspection robot is received, the second three-dimensional spatial coordinates of the failed foreign object will be sent to the control center.

5. The substation autonomous inspection and foreign matter cleaning coordinated control system according to claim 1 is characterized in that: The system further includes a residue identification module, wherein the residue identification module is configured to: Reacquiring the substation's 2D image data, LiDAR point cloud data, and multimodal sensor data, calculating a difference map between the 2D image data before and after cleaning using a structural similarity algorithm, and performing connected region analysis on the difference map to determine image disturbance areas. Perform spatial registration processing on the LiDAR point cloud data before and after cleaning to determine the point cloud disturbance area; Verifying the spatial consistency of the image disturbance region and the point cloud disturbance region; if they are consistent, confirming the image disturbance region as a candidate residual foreign matter region; Perform target detection on the candidate residual foreign matter area to identify whether there is residual foreign matter.

6. The substation autonomous inspection and foreign matter cleaning coordinated control system according to claim 5 is characterized in that: The system further includes a local cleaning module, which is configured to: If there is residual foreign matter, obtaining the third three-dimensional spatial coordinates and the second foreign matter visual feature description of the residual foreign matter, inputting the third three-dimensional spatial coordinates, the second foreign matter visual feature description, the spatial layout information, and the re-acquired multimodal sensor data into a preset foreign matter risk level scoring model to obtain a second foreign matter risk level score for the residual foreign matter; Determine the second foreign object operation requirements for each foreign object based on the second foreign object risk level score and the preset cleaning strategy library; Obtain the third current position, third posture angle and third motion state information of the inspection robot, determine the local cleaning path of the inspection robot and the second cleaning execution parameters for residual foreign objects based on the third three-dimensional space coordinates, the third current position, the third posture angle, the third motion state information and the second foreign object operation requirements, and send the local cleaning path and the second cleaning execution parameters for residual foreign objects to the inspection robot.

7. The substation autonomous inspection and foreign matter cleaning coordinated control system according to claim 5 is characterized in that: The system further includes an abnormality reporting module, which is configured to: If there is no residual foreign matter, edge change detection is performed on the image disturbance area to determine whether there is any image structural abnormality; If there are abnormal image structures, density change analysis methods are used to identify whether there are abnormal point cloud distributions based on the LiDAR point cloud data before and after cleaning. If there is an abnormal point cloud distribution, an abnormality reporting information package is generated based on the re-acquired two-dimensional image data, lidar point cloud data and multimodal sensor data of the substation, and the abnormality reporting information package is transmitted to the control center.

8. A method for coordinated control of autonomous inspection and foreign matter cleaning of a substation, characterized in that: The method comprises: Acquire two-dimensional image data, lidar point cloud data, and multimodal sensor data of the substation, and use a convolutional neural network target detection algorithm to identify the two-dimensional coordinate position of each foreign object in the two-dimensional image data and a first foreign object visual feature description; Performing spatial registration processing based on the laser radar point cloud data and the two-dimensional coordinate position to obtain the first three-dimensional spatial coordinates of each foreign object; Obtaining spatial layout information of the electrical equipment, inputting the first three-dimensional spatial coordinates, the first foreign object visual feature description, the spatial layout information, and the multimodal sensor data into a preset foreign object risk level scoring model to obtain a first foreign object risk level score for each foreign object; Determine the first foreign object operation requirement for each foreign object based on the first foreign object risk level score and a preset cleaning strategy library, obtain the first current position, first posture angle, and first motion state information of the inspection robot, determine the cleaning path of the inspection robot and the first cleaning execution parameters for each foreign object based on the first three-dimensional space coordinates, the first current position, the first posture angle, the first motion state information, and the first foreign object operation requirement, and send the cleaning path and the first cleaning execution parameters for each foreign object to the inspection robot; wherein the first cleaning execution parameters are a specific cleaning action parameter set of the inspection robot, which is used to control its cleaning executor to complete the cleaning task, including cleaning mode, cleaning trigger distance, trigger angle, cleaning execution force, action duration, posture adjustment angle, and action stability parameter; If the cleaning completion information of the inspection robot is received, the second current position, second posture angle, and second motion state information of the inspection robot are obtained, as well as the spatial topology structure and safety constraints of the substation are obtained, and the inspection path of the inspection robot and the inspection actions on each electrical equipment are determined according to the second current position, second posture angle, second motion state information, spatial topology structure, spatial layout information and safety constraints, and the inspection path and inspection actions on each electrical equipment are sent to the inspection robot.

9. An electronic device, characterized in that: It includes a processor, a memory, and a program or instruction stored in the memory and executable on the processor. When the program or instruction is executed by the processor, the steps of the coordinated control method for autonomous inspection and foreign matter cleaning of substations as described in claim 8 are implemented.

10. A readable storage medium, characterized in that: The readable storage medium stores a program or instruction, and when the program or instruction is executed by the processor, the steps of the substation autonomous inspection and foreign matter cleaning coordinated control method as claimed in claim 8 are implemented.

Images