Image-based control method and system for general-purpose transfer vehicles in substations
By using image perception and multi-source positioning fusion technology, the problem of manual dependence in the transfer process of 10kV handcart switchgear in substations has been solved, realizing automated equipment positioning and mechanical docking, and improving operational efficiency and safety.
Patent Information
- Application Number
- CN202611618396.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-20
- Publication Date
- 2026-11-17
AI Technical Summary
During the transfer of existing 10kV handcart-type switchgear in substations, the operation tasks rely on manual identification of operation tickets, adjustment of pallet height and bayonet width, and lack of unified data association and continuous control relationship, resulting in increased operation time.
Image perception technology is used to establish an interval information model, operation tickets are identified by OCR and structured tasks are generated, and multi-source positioning fusion is performed by combining RFID and LiDAR to realize automatic positioning and visual docking of general-purpose transfer vehicles. The IEC 61850 standard is used for manual and electric control of the vehicle to ensure the continuity and safety of operation.
It enables automatic parsing of operation tickets, accurate positioning of target equipment, and mechanical docking, reducing manual intervention and improving the efficiency and safety of substation switching operations.
Smart Images

Figure CN123122029A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of image recognition technology, specifically to a control method and system for a general-purpose transfer vehicle in a substation based on image perception. Background Technology
[0002] Currently, substation switching operations involve tasks such as adjusting operating modes, switching equipment status, and coordinating maintenance. 10kV trolley-type switchgear typically adopts a centrally located structure, with the trolley operating position approximately 0.8–1m above the ground. When the trolley is removed from the switchgear or pushed into the switchgear from the maintenance position, a transfer vehicle is required for receiving and moving it. With the increasing number of substations and the growing number of equipment operation tasks, 10kV trolley transfer has become a routine part of switching operations. The compatibility, positioning, and operational coordination between the transfer vehicle and different equipment bays directly affect the trolley operation process.
[0003] For the switching operations of the aforementioned 10kV trolley-type switchgear, the current procedures typically involve maintenance personnel reading the operation ticket and confirming the target bay. Based on the switchgear and trolley model, a corresponding transfer vehicle is selected, and the transfer vehicle is moved to the front of the target switchgear. The height and support position of the receiving tray are then manually adjusted to align the receiving tray with the switchgear track. After docking, the operator either withdraws or pushes the trolley in, and uses the transfer vehicle to move the withdrawn trolley away from the switchgear. When operating multiple bays consecutively, the target bay confirmation, transfer vehicle adjustment, and trolley transfer are completed sequentially.
[0004] However, existing transfer methods primarily rely on manual processing for identifying operation tickets, confirming target intervals, and adjusting transfer vehicle positions. There is a lack of unified data correlation between operational tasks, equipment interval positions, pallet heights, and bayonet widths. Some transfer vehicles have fixed bayonet widths, requiring manual pallet height adjustments, and different switchgear models require corresponding transfer vehicles. Furthermore, after the transfer vehicle arrives at the target interval, its position confirmation and mechanical docking also rely on manual observation. Simultaneously, there is a lack of continuous control between the electric operation of the handcart and the positioning and docking status of the transfer vehicle. In multi-interval maintenance operations, the handcart must be withdrawn, transferred, and reused sequentially, increasing the time required for continuous switching operations and maintenance work.
[0005] Therefore, the present invention provides a control method and system for a general-purpose transfer vehicle in substations based on image perception, in order to overcome the shortcomings of the prior art. Summary of the Invention
[0006] To address the shortcomings of existing technologies, this invention provides a general-purpose substation transfer vehicle control method and system based on image perception. This solves the problems of manual reading of operation tickets, manual adjustment of pallet height and bayonet width between different equipment intervals, manual observation of target switchgear docking, and lack of unified control relationship between electric operation of the transfer vehicle and the transfer process in existing substation transfer processes.
[0007] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, the present invention provides a control method for a general-purpose transfer vehicle in a substation based on image perception, employing the following technical solution: A control method for a general-purpose transfer vehicle in a substation based on image perception includes the following steps: Establish an interval information model to associate the dual serial number of the equipment with the interval spatial coordinates, pallet height, and bayonet width; The operation ticket image is collected, preprocessed, OCR recognized, and semantically parsed to generate a structured operation task containing the target device's dual number, initial position, target position, and handcart operation direction. Based on the target device's dual number, the target interval spatial coordinates, target pallet height, and target gate width are determined in the interval information model. The general-purpose transfer vehicle is controlled to move according to the target interval spatial coordinates and multi-source positioning fusion is performed. Based on the fusion position result, the target pallet height and the target bayonet width, visual docking with the target switch cabinet socket is completed, and a mechanical docking completion status is generated. Based on the structured operation task and the mechanical docking completion status, a handcart operation execution permission is generated, and the handcart electric control module is controlled to execute the handcart's exit or push-in operation. When the structured operation task is an exit operation, the exited handcart is transferred to the target maintenance station on the maintenance platform and its positioning and locking are completed.
[0008] By adopting the above technical solution, the dual equipment number is used as the basis for linking the operation ticket image recognition result with the interval information model, enabling the target equipment in the operation ticket to correspond to the target interval spatial coordinates, target pallet height, and target bayonet width. After the target interval is located, visual docking is further completed using the target switch cabinet socket image, and the mechanical docking completion status is used to participate in the handcart operation execution permission judgment. Therefore, the operation ticket task parsing, target equipment determination, universal transfer car movement, visual docking, and handcart electric operation form a continuous control relationship.
[0009] The image perception of this invention includes task information perception of the operation ticket image and docking position perception of the target switchgear socket image. The former converts the dual equipment number, initial position, target position, and trolley operation direction in the operation ticket into a structured operation task, while the latter determines the positional deviation between the general-purpose transfer vehicle and the target switchgear socket based on the center of the target switchgear socket. The two types of image perception results are correlated through the target equipment dual number, target interval spatial coordinates, and structured operation task, enabling the operation object recognition results to directly participate in subsequent position control and trolley electric operation control.
[0010] Preferably, an equipment interval record is established, and the interval spatial coordinates, pallet height, and bayonet width are associated with the dual equipment number; mechanical zero-point calibration is performed on the pallet lifting actuator and bayonet width adjustment actuator of the transfer vehicle to obtain height calibration compensation data and width calibration compensation data; the equipment interval record is retrieved according to the dual number of the target equipment, and the target pallet height is determined based on the retrieved pallet height and the height calibration compensation data, and the target bayonet width is determined based on the retrieved bayonet width and the width calibration compensation data.
[0011] By adopting the above technical solution, a correspondence is established between the dual serial number of the equipment and the mechanical adaptation parameters of the general-purpose transfer vehicle. The mechanical zero-point error and installation deviation are corrected by the height calibration compensation data and the width calibration compensation data, so that different equipment intervals can complete the pre-adjustment of the receiving pallet by using the corresponding target pallet height and target bayonet width.
[0012] Preferably, the operation ticket image is acquired by an industrial camera, and noise suppression, grayscale normalization, and binarization are performed sequentially; text lines or table borders are detected by Hough transform and tilt correction is performed; the corrected operation ticket image is subjected to OCR recognition to obtain the dual equipment number, operation object, initial position, and target position; semantic parsing is performed on the OCR text, and the operation direction of the handcart is determined according to the operation object, the initial position, and the target position.
[0013] By adopting the above technical solution, the imaging noise, local illumination changes and image tilt of the operation ticket image are preprocessed, and then text fields related to the equipment object and operation direction are extracted, so that the subsequent structured operation tasks have clear equipment objects, position change relationships and handcart operation directions.
[0014] Preferably, the confidence scores of OCR text recognition, semantic matching of operation statements, and device dual ID matching are obtained, and then weighted and summed according to their corresponding fusion weights to obtain the comprehensive confidence score of the operation task. The preset task confidence threshold is between 0.90 and 1.00. When the comprehensive confidence score of the operation task reaches the task confidence threshold, and the target device dual ID corresponds to only one valid device interval record, a structured operation task containing the target device dual ID, initial position, target position, and handcart operation direction is generated.
[0015] By adopting the above technical solution, the text recognition result, operation semantic matching result and device dual number matching result are used together to judge the conditions for generating structured operation tasks, avoiding the formation of subsequent automatic control tasks based solely on a single OCR recognition result, while limiting the actual operation object through the unique correspondence of the target device dual number.
[0016] Preferably, a movement path is generated based on the current position of the general-purpose transfer vehicle, the spatial coordinates of the target interval, the passable area of the switch room, and the position of fixed obstacles; RFID identification results are obtained through RFID devices, industrial camera positioning data is obtained through industrial cameras, and lidar ranging data is obtained through lidar; the industrial camera positioning data and the lidar ranging data are time-aligned, Kalman filtering is used to obtain the fused position result, and position compensation is performed based on historical positioning correction data.
[0017] By adopting the above technical solution, RFID identification results are used to confirm the identity of the target device, industrial camera positioning data and lidar ranging data are used to form continuous position observation, and fused position results are formed through time alignment, Kalman filtering and historical positioning correction, so that the target device is confirmed to be in the same positioning process as the target in spatial positioning.
[0018] Preferably, based on the fusion position result, the general-purpose transfer vehicle is controlled to enter the visual docking area, acquire images of the target switch cabinet socket, identify the socket edge and hole mask, and determine the center of the target switch cabinet socket; based on the calibration relationship between the industrial camera imaging parameters and the target switch cabinet socket reference plane, the lateral spatial deviation and vertical spatial deviation are determined; the moving chassis of the general-purpose transfer vehicle is adjusted according to the lateral spatial deviation, the pallet lifting actuator is adjusted according to the target pallet height and the vertical spatial deviation, and the bayonet width adjustment actuator is adjusted so that the effective bayonet width reaches the target bayonet width; when the absolute values of the lateral spatial deviation and the vertical spatial deviation are not greater than the corresponding allowable values, and the receiving pallet height reaches the target pallet height and the effective bayonet width reaches the target bayonet width, the mechanical docking completion state is generated.
[0019] By adopting the above technical solution, the fused position results are used to determine the initial position of the general-purpose transfer vehicle after entering the visual docking area. Then, the horizontal and vertical spatial deviations are generated using the target switch cabinet socket image, and the mobile chassis and pallet lifting actuator are controlled to perform closed-loop adjustment respectively. At the same time, the effective bayonet width is adjusted according to the target bayonet width, so that the mechanical docking completion state is determined by the spatial position, the height of the receiving pallet, and the effective bayonet width.
[0020] The lateral and vertical spatial deviations are determined based on the lateral and vertical pixel position differences between the center of the target switchgear socket and the preset docking reference center, respectively, the distance from the optical center of the industrial camera to the reference plane of the target switchgear socket, and the equivalent focal length of the industrial camera in the corresponding direction. The target pallet height is determined based on the pallet height and height calibration compensation data, and the target bayonet width is determined based on the bayonet width and width calibration compensation data. This allows changes in pixel position in the visual image to be converted into corresponding spatial adjustment amounts for the moving chassis and pallet lifting actuator.
[0021] Preferably, the operation object consistency status is determined based on the target device's dual identification number and RFID identification result in the structured operation task; the target interval positioning validity status is determined based on the fused location result; the communication link validity status is determined based on the IEC 61850 communication connection and status message reception status; the equipment safety status is determined based on the equipment status and operation interlocking conditions returned by the handcart electric control module; when the operation object consistency status, the target interval positioning validity status, the mechanical docking completion status, the communication link validity status, and the equipment safety status are all verified as passed, the handcart operation execution permit is generated.
[0022] By adopting the above technical solution, the target equipment identity, target interval position, mechanical docking, communication link and equipment safety status are used as the conditions for generating the handcart operation execution permission. This establishes a state constraint relationship between the visual docking result of the target switch cabinet socket and the handcart electric operation control. If any necessary state is not verified, the actual handcart operation process will not begin.
[0023] Preferably, a trolley exit control request or a trolley push-in control request is generated based on the initial position and target position in the structured operation task; after generating the trolley operation execution permission, a trolley action command is sent via MMS, and the current position status, action execution status, and equipment abnormal status of the trolley are received via GOOSE; if no valid status message is received for three consecutive communication cycles, the sending of the trolley action command is stopped, and when the equipment abnormal status is received, stop control data is output; the operation ends when the trolley reaches the target position and the action execution status is completed.
[0024] By adopting the above technical solution, the positional change relationship in the structured operation task is used to determine the actual operation direction of the handcart, and the handcart action command is transmitted through MMS. The status feedback during the handcart action process is obtained through GOOSE, so that the handcart action command sending, action status feedback, communication anomaly handling and operation completion judgment form a closed loop.
[0025] Preferably, when the structured operation task is an exit operation, after confirming that the exiting handcart has entered the receiving pallet and the mechanical limiting structure is in place, the mechanical docking state between the general-purpose transfer car and the target switchgear is released; candidate transfer paths are generated for the idle handcart maintenance workstations, and the estimated travel time of each candidate transfer path is determined according to the path length, allowed travel speed and additional waiting time, and the candidate transfer path with the shortest estimated travel time is determined as the target transfer path; the exiting handcart is sent into the target maintenance workstation, and positioning and locking are completed by positioning pins, mechanical lock holes, locking hooks and braking mechanisms, and the target maintenance workstation is updated to an occupied state.
[0026] By adopting the above technical solution, the mechanical connection between the withdrawing trolley and the target switchgear is released only after it is confirmed that the withdrawing trolley has been stably received by the receiving pallet. The target maintenance station and target transfer path are determined by using the occupancy status of the trolley maintenance station and the estimated travel time of the candidate transfer path, so that the withdrawal operation and the subsequent trolley transfer and placement are executed continuously.
[0027] The estimated travel time for a candidate transfer path is determined by the path length and permissible speed of each segment within the candidate transfer path, and additional waiting time due to turning, obstacle avoidance, passage confirmation, or speed switching between adjacent segments is added. The target transfer path is determined by comparing the estimated travel times of each candidate transfer path, enabling the maintenance platform to select workstations and paths according to unified path evaluation conditions when multiple idle handcart maintenance positions exist.
[0028] Secondly, the present invention provides a general-purpose substation transfer vehicle control system based on image perception, which adopts the following technical solution: A substation general-purpose transfer vehicle control system based on image perception, used to execute the above-mentioned substation general-purpose transfer vehicle control method based on image perception, includes an bay modeling module, an image perception module, a positioning and docking module, a linkage control module, and a transfer and placement module; The interval modeling module is used to build an interval information model, which associates the dual serial number of the equipment with the interval spatial coordinates, pallet height and bayonet width; The image perception module is used to acquire operation ticket images for preprocessing, OCR recognition and semantic parsing, generate a structured operation task containing the target device's dual number, initial position, target position and handcart operation direction, and determine the target interval spatial coordinates, target pallet height and target gate width in the interval information model based on the target device's dual number. The positioning and docking module is used to control the movement of the general-purpose transfer vehicle according to the target interval spatial coordinates and to perform multi-source positioning fusion. Based on the fusion position result, the target pallet height and the target bayonet width, it completes the visual docking with the target switch cabinet socket and generates a mechanical docking completion status. The linkage control module is used to generate a handcart operation execution permission based on the structured operation task and the mechanical docking completion status, and control the handcart electric operation control module to execute the handcart's exit or push-in operation. The transfer and placement module is used to transfer the exited handcart to the target maintenance station of the maintenance platform and complete the positioning and locking when the structured operation task is an exit operation. The interval modeling module, the image perception module, the positioning and docking module, the linkage control module, and the transfer and resettlement module interact with each other through the central processing unit.
[0029] By adopting the above technical solution, the interval modeling module provides equipment interval parameters, the image perception module forms a structured operation task and completes the image recognition of the target switchgear socket, the positioning and docking module forms the fused position result and the mechanical docking completion status, the linkage control module controls the handcart electric operation control module according to the structured operation task and the mechanical docking completion status, and the transfer and placement module completes the transfer and positioning locking of the withdrawn handcart. Each module transmits the data and status formed in the previous processing stage through the central processing unit, thereby realizing continuous control between operation ticket image perception, target interval positioning, target switchgear socket visual docking, handcart electric operation linkage, and handcart transfer and placement.
[0030] The technical solution of this invention uses dual numbering of target equipment and structured operation tasks as the basis for operation object association, integrates location results and target switchgear socket image recognition results as the basis for mechanical docking control, and uses the mechanical docking completion status as the necessary input for trolley operation execution permission. After the trolley is withdrawn, the structured operation tasks continue to be used to complete the transfer and placement of the maintenance platform. Thus, the operation object and location change information in the operation ticket can be continuously transmitted to the movement of the general-purpose transfer vehicle, visual docking, electric operation of the trolley, and the placement of the target maintenance workstation, forming an image-perception-based substation general-purpose transfer vehicle control technology solution.
[0031] This invention provides a control method and system for a general-purpose transfer vehicle in substations based on image perception. It has the following beneficial effects: 1. This invention establishes an interval information model, which associates the dual serial number of the equipment with the interval spatial coordinates, pallet height, and bayonet width. It also combines height calibration compensation data and width calibration compensation data to determine the target pallet height and target bayonet width, enabling the general-purpose transfer vehicle to call the corresponding interval position and mechanical adaptation parameters according to the dual serial number of the target equipment, providing a data basis for the automatic movement of different equipment intervals and the adjustment of receiving pallets.
[0032] 2. This invention utilizes an industrial camera to acquire images of operation tickets and target switchgear sockets. It then uses OCR recognition and semantic analysis to form structured operation tasks. Furthermore, it combines RFID identification results, industrial camera positioning data, and LiDAR ranging data for multi-source positioning fusion. Upon reaching the target interval, it determines the lateral and vertical spatial deviations based on the center of the target switchgear socket. This allows for the control of the mobile chassis, pallet lifting actuator, and bayonet width adjustment actuator to complete visual docking, thus forming a continuous control process that integrates operation task identification, target equipment confirmation, and mechanical docking.
[0033] 3. This invention realizes data interaction between a universal transfer vehicle and a handcart electric control module based on IEC 61850. It generates a handcart operation execution permission based on the consistent status of the operation object, the effective status of the target interval positioning, the mechanical docking completion status, the effective status of the communication link, and the equipment safety status. It sends handcart action commands through MMS and receives handcart status information through GOOSE. After the handcart exits, it continues to perform maintenance platform transfer and target maintenance station positioning and locking, thereby realizing the linkage control between the handcart exit or push-in operation and the subsequent transfer and placement process. Attached Figure Description
[0034] Figure 1 This is a framework diagram of the image-sensing-based substation universal transfer vehicle control system of the present invention; Figure 2 This is a flowchart illustrating the overall process of the image-sensing-based control method for a general-purpose transfer vehicle in substations according to the present invention. Figure 3 This is a flowchart illustrating the process of establishing the interval information model of the present invention. Figure 4 This is a flowchart of the operation ticket image perception and task parsing process of the present invention; Figure 5 This is a flowchart of the target interval positioning and visual docking process of the present invention; Figure 6 This is a flowchart of the handcart electric operation linkage control of the present invention; Figure 7 This is a flowchart illustrating the transfer and placement process of the handcart according to the present invention; Figure 8 This is a graph showing the changes in the lateral and vertical absolute docking deviations of the universal transfer vehicle of the present invention as a function of the visual docking control cycle. Figure 9 This is a graph comparing the single-interval complete operation time of the traditional manual transfer method and the control method of the present invention. Detailed Implementation
[0035] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0036] The universal transfer vehicle in this embodiment includes a mobile chassis, a pallet receiving unit, a pallet lifting actuator, a bayonet width adjustment actuator, an industrial camera, a positioning sensing device, a central processing unit, and a communication interface. The mobile chassis carries the universal transfer vehicle and moves it within the substation switch room. The pallet receiving unit receives the handcart. The pallet lifting actuator adjusts the vertical position of the pallet receiving unit, and the bayonet width adjustment actuator adjusts the effective bayonet width of the pallet receiving unit. The pallet lifting actuator uses an electro-hydraulic scissor lift mechanism, and the bayonet width adjustment actuator uses a linear adjustment mechanism formed by a servo drive mechanism and a T-screw.
[0037] Industrial cameras are used to acquire images of operation tickets, target switchgear areas, and target switchgear sockets. Positioning and sensing devices include Radio Frequency Identification (RFID) devices and LiDAR. The central processing unit establishes data connections with the industrial cameras, RFID devices, LiDAR, mobile chassis, pallet lifting actuator, bayonet width adjustment actuator, and communication interfaces, and outputs corresponding control data based on the acquired image data, equipment identification data, ranging data, and actuator status data.
[0038] See attached document Figure 1 This invention provides a general-purpose transfer vehicle control system for substations based on image perception, including a bay modeling module, an image perception module, a positioning and docking module, a linkage control module, and a transfer and placement module. The bay modeling module, image perception module, positioning and docking module, linkage control module, and transfer and placement module interact with each other through a central processing unit, exchanging structured operation tasks, equipment bay information, positioning information, image recognition results, mechanical docking completion status, vehicle status, and vehicle maintenance station status.
[0039] The interval modeling module is used to establish an interval information model, associating the equipment dual ID with interval spatial coordinates, pallet height, and bayonet width. The module treats equipment intervals as independent data objects, with each interval corresponding to a single equipment interval record. Each equipment interval record includes at least the equipment dual ID, interval spatial coordinates, pallet height, bayonet width, and equipment type. The equipment dual ID serves as a unique equipment index for the interval record, establishing the correspondence between the equipment object in the operation ticket and the actual equipment interval; the interval spatial coordinates determine the target interval location; and the pallet height and bayonet width form the initial adaptation parameters for receiving the pallet.
[0040] A unified spatial coordinate benchmark is established in the switch room. Equipment intervals, accessible areas, maintenance platforms, and fixed obstacles are all stored according to this benchmark. The central processing unit retrieves the corresponding equipment interval record based on the target equipment's dual identification number, obtaining the target interval's spatial coordinates, pallet height, and bayonet width. The interval modeling module also stores height and width calibration compensation data to correct for mechanical zero-point errors and installation deviations in the pallet lifting actuator and bayonet width adjustment actuator, thereby determining the target pallet height and target bayonet width.
[0041] The image perception module is used to acquire operation ticket images for preprocessing, optical character recognition (OCR), and semantic parsing. It generates a structured operation task containing the target device's dual serial number, initial position, target position, and trolley operation direction. Based on the target device's dual serial number, it determines the target interval spatial coordinates, target pallet height, and target bayonet width in the interval information model. The image perception module includes an image acquisition unit, an image preprocessing unit, a character recognition unit, a task semantic parsing unit, and a cabinet visual recognition unit.
[0042] After the industrial camera captures the operation ticket image, the image preprocessing unit sequentially performs noise suppression, grayscale normalization, binarization, and tilt correction. Noise suppression reduces the impact of paper texture and isolated pixel noise on character edges; grayscale normalization reduces local illumination differences; and binarization separates text areas from background areas. Tilt correction detects text lines or table borders, uses Hough transform to determine the main line direction, and then performs rotation correction on the operation ticket image based on the main line direction.
[0043] The text recognition unit uses OCR to convert the pre-processed operation ticket image into text data. The OCR recognition model performs text region detection and text sequence recognition, and obtains the dual device number, the operation object, the initial position, and the target position. The task semantic parsing unit performs semantic parsing on the OCR text, determines the handcart operation direction based on the operation object, the initial position, and the target position, and matches the dual device number with the interval information model. When the target device dual number corresponds to only one valid device interval record, a structured operation task is formed.
[0044] The image perception module is also used for visual recognition of the target switchgear sockets. After the general-purpose transport vehicle enters the area adjacent to the target bay, the industrial camera acquires images of the target switchgear area and the target switchgear sockets. The cabinet visual recognition unit uses a pre-trained neural network to identify the target switchgear socket area, socket edges, and hole position masks, and determines the center of the target switchgear socket based on the socket edges and hole position masks. The center of the target switchgear socket is used to determine the lateral and vertical spatial deviations.
[0045] The positioning and docking module controls the movement of the general-purpose transfer vehicle based on the spatial coordinates of the target interval and performs multi-source positioning fusion. Based on the fused position result, the target pallet height, and the target bayonet width, it completes visual docking with the target switchgear socket and generates a mechanical docking completion status. The central processing unit generates a movement path based on the spatial coordinates of the target interval, the current position of the general-purpose transfer vehicle, the passable area of the switch room, and the position of fixed obstacles. The mobile chassis travels according to the movement path, and the LiDAR continuously acquires distance information of the surrounding environment.
[0046] After the general-purpose transfer vehicle enters the vicinity of the target interval, it obtains RFID identification results through RFID devices, industrial camera positioning data through industrial cameras, and lidar ranging data through lidar. The central processing unit compares the dual device identification number from the RFID identification results with the dual identification number of the target device in the structured operation task. The RFID identification results are used to constrain the identity of the target device. The industrial camera positioning data is used to indicate the position of the target switch cabinet relative to the general-purpose transfer vehicle, and the lidar ranging data is used to indicate the distance between the general-purpose transfer vehicle and the plane where the target switch cabinet is located.
[0047] The central processing unit performs time alignment on the industrial camera positioning data and LiDAR ranging data, uses Kalman filtering to obtain the fused position result, and performs position compensation based on historical positioning correction data associated with the target equipment's dual identification number. When the observation residuals corresponding to the industrial camera positioning data or LiDAR ranging data increase, the corresponding observation noise parameters are adjusted. The fused position result after position compensation is used to control the general-purpose transport vehicle to enter the visual docking area.
[0048] After the general-purpose transfer vehicle enters the visual docking area, the industrial camera captures images of the target switch cabinet sockets, and the cabinet visual recognition unit determines the center of the target switch cabinet sockets. The central processing unit determines the lateral and vertical spatial deviations based on the calibration relationship between the industrial camera's imaging parameters and the target switch cabinet socket reference plane. Based on the lateral spatial deviation, the general-purpose transfer vehicle's chassis is adjusted; based on the target pallet height and vertical spatial deviation, the pallet lifting actuator is adjusted; and the bayonet width adjustment actuator is adjusted to ensure the effective bayonet width reaches the target bayonet width. After each position adjustment, the target switch cabinet socket image is recaptured, and the spatial deviation is re-determined.
[0049] The linkage control module generates trolley operation execution permissions based on structured operation tasks and mechanical docking completion status, controlling the trolley electric operation control module to execute trolley retraction or insertion operations. Data exchange between the central processing unit and the trolley electric operation control module is conducted using the International Electrotechnical Commission 61850 (IEC 61850) standard. The linkage control module sends trolley action commands using the Manufacturing Message Specification (MMS) and receives the trolley's current position status, action execution status, and equipment abnormal status using Generic Object Oriented Substation Event (GOOSE).
[0050] Before generating a trolley operation execution permit, the linkage control module determines the consistency status of the operation object based on the dual identification of the target equipment in the structured operation task and the RFID identification result; determines the effective status of the target interval positioning based on the fused location result; determines the effective status of the communication link based on the IEC 61850 communication connection and status message reception; and determines the equipment safety status based on the equipment status and operation interlocking conditions returned by the trolley electric operation control module. A trolley operation execution permit is generated when the operation object consistency status, target interval positioning effective status, mechanical docking completion status, communication link effective status, and equipment safety status are all verified as passed.
[0051] The transfer and placement module is used to transfer the exiting handcart to the target maintenance station on the maintenance platform and lock it in place when the structured operation task is an exit operation. The maintenance platform has one or more handcart maintenance stations. Each handcart maintenance station includes an adjustable recessed groove, a sliding rail, a mechanical lock hole, a positioning post, a locking hook, and a braking mechanism. The adjustable recessed groove accommodates handcarts of different sizes, the sliding rail adjusts the position of the handcart maintenance station, the mechanical lock hole is adapted to the handcart's position locking structure, the positioning post and locking hook restrict the handcart's position, and the braking mechanism restricts further movement of the handcart. The central processing unit stores the occupancy status of each handcart maintenance station and selects an available handcart maintenance station when executing a handcart transfer task.
[0052] See attached document Figure 2 This invention provides a control method for a general-purpose transfer vehicle in a substation based on image perception. The control method sequentially executes the following steps: establishing an interval information model, processing operation ticket images and generating structured operation tasks, moving the general-purpose transfer vehicle and fusing multi-source positioning, visual docking with the target switchgear interface, electronic operation linkage of the transfer vehicle, and transferring and placing the withdrawn transfer vehicle. Step S100: Establish an interval information model and associate the dual device number with the interval spatial coordinates, pallet height, and bayonet width.
[0053] Step S200: Collect operation ticket images for preprocessing, OCR recognition and semantic parsing to generate a structured operation task containing the target device's dual number, initial position, target position and handcart operation direction.
[0054] Step S300: Control the movement of the general-purpose transfer vehicle according to the target interval spatial coordinates and perform multi-source positioning fusion. Based on the fusion position result, target pallet height and target bayonet width, complete the visual docking with the target switch cabinet socket and generate the mechanical docking completion status.
[0055] Step S400: Generate a handcart operation execution permission based on the structured operation task and the mechanical docking completion status, and control the handcart electric control module to execute the handcart's exit or push-in operation.
[0056] Step S500: When the structured operation task is an exit operation, the exited handcart is transferred to the target maintenance station on the maintenance platform and the positioning and locking are completed.
[0057] See attached document Figure 3 Step S100: Establish an interval information model and associate the dual device number with the interval spatial coordinates, pallet height and bayonet width.
[0058] This is performed via the interval modeling module. It includes the following sub-steps.
[0059] Sub-step S101: Collect the dual serial number of the device, the spatial coordinates of the interval, the tray height, and the bayonet width.
[0060] First, a unified spatial coordinate reference is determined for the switch room. Under the unified spatial coordinate reference, the position of each equipment bay is recorded, and an independent equipment bay record is established for each equipment bay.
[0061] The equipment interval record shall include at least the equipment dual number, interval spatial coordinates, pallet height, bayonet width, and equipment type. The equipment dual number shall be obtained from the corresponding switchgear equipment identification information; the interval spatial coordinates shall be determined according to a unified spatial coordinate benchmark; the pallet height shall be determined according to the vertical position of the foundation required for the receiving pallet to be adapted to the corresponding handcart; and the bayonet width shall be determined according to the effective bayonet width of the foundation required for the receiving pallet to be adapted to the corresponding handcart.
[0062] After data collection is completed, integrity and uniqueness checks are performed. When the same device has two serial numbers corresponding to multiple device interval records, the corresponding records will not be used for automatic transfer tasks; if the interval spatial coordinates, pallet height, or bayonet width is missing, the corresponding device interval record will be set to an incomplete state.
[0063] Sub-step S102: Establish the association between the dual equipment number and the interval space coordinates, pallet height and bayonet width, and verify the validity of the parameters.
[0064] The dual device identification number is used as the device index field in the interval information model, while the interval spatial coordinates, pallet height, and bayonet width are used as the corresponding device parameter fields. After obtaining the dual device identification number of the target device, the central processing unit retrieves the corresponding device interval record by searching the interval information model using the dual device identification number.
[0065] The central processing unit checks whether the target bay's spatial coordinates are located within the switch room equipment area, and also checks whether there is a continuous passable area between the current position of the general-purpose transfer vehicle and the target bay's spatial coordinates. If no continuous passable area exists, no movement task is generated.
[0066] The pallet height is compared with the effective adjustment range of the pallet lifting actuator, and the bayonet width is compared with the effective adjustment range of the bayonet width adjustment actuator. When both are within their respective effective adjustment ranges, initial pallet fitting data is generated.
[0067] Sub-step S103 involves performing mechanical zero-point calibration on the pallet lifting actuator and the bayonet width adjustment actuator, and determining the target pallet height and the target bayonet width.
[0068] Mechanical zero-point calibration was performed on the pallet lifting actuator and the bayonet width adjustment actuator of the general-purpose transfer vehicle. After the mechanical zero-point calibration of the pallet lifting actuator was completed, height calibration compensation data was obtained between the actual height of the received pallet and the height recorded by the control. After the mechanical zero-point calibration of the bayonet width adjustment actuator was completed, width calibration compensation data was obtained between the actual effective bayonet width and the width recorded by the control.
[0069] Based on the dual-numbered equipment records, the target pallet height is determined using the retrieved pallet height and height calibration compensation data. The target bayonet width is determined using the retrieved bayonet width and width calibration compensation data. The calculation formula for the transfer vehicle adaptation settings is as follows: ; ; in, For the first The target pallet height corresponding to each equipment interval represents the vertical position that the general-purpose transfer vehicle needs to reach when receiving the pallet from the handcart, and the unit is mm; For the first The pallet height corresponding to each equipment interval represents the basic vertical position of the receiving pallet and the corresponding handcart when they are adapted to each other, as recorded in the interval information model, and the unit is mm. For the first The height calibration compensation value corresponding to each equipment interval represents the height correction amount formed by the zero point error of the pallet lifting actuator, installation deviation, and on-site calibration results, and the unit is mm; For the first The target bayonet width corresponding to each equipment interval represents the effective bayonet width required when the receiving pallet and the corresponding handcart are adapted to each other, and the unit is mm. For the first The bayonet width corresponding to each equipment interval represents the basic effective bayonet width when the receiving pallet and the corresponding handcart are adapted to each other as recorded in the interval information model, and the unit is mm. For the first The width calibration compensation value corresponding to each equipment interval represents the width correction amount formed by the zero point error of the bayonet width adjustment actuator, the mechanical installation clearance, and the field calibration results, and the unit is mm; This is the device interval sequence number, which represents the device record index in the interval information model, and is a positive integer.
[0070] Once the target pallet height and target bayonet width are determined, the pallet lifting actuator and the bayonet width adjustment actuator complete the pre-adjustment of receiving the pallet, and then continue to adjust the position according to the visual docking result of the target switch cabinet socket.
[0071] See attached document Figure 4 In step S200, the operation ticket image is collected, preprocessed, OCR recognized, and semantically parsed to generate a structured operation task containing the target device's dual number, initial position, target position, and handcart operation direction.
[0072] This is performed via the image perception module. It includes the following sub-steps.
[0073] Sub-step S201: Acquire the operation ticket image and perform noise suppression, grayscale normalization, binarization and tilt correction.
[0074] The industrial camera captures operation ticket images containing the equipment's dual serial number, the object being operated, the initial position, and the target position, and sends the operation ticket images to the central processing unit.
[0075] The central processing unit sequentially performs noise suppression, grayscale normalization, and binarization. Noise suppression reduces the impact of paper texture and isolated pixel noise on the character area, grayscale normalization reduces local illumination differences, and binarization separates the text area from the background area.
[0076] Subsequently, text lines or table borders in the binary image of the operation ticket are detected, the main line direction is obtained through Hough transform, and tilt correction is performed on the operation ticket image according to the offset angle of the main line direction relative to the horizontal direction.
[0077] Sub-step S202 involves performing OCR recognition on the corrected operation ticket image to obtain the dual device number, operation object, initial position, and target position.
[0078] The corrected operation ticket image is input into the OCR recognition model. The OCR recognition model first detects the text region, and then recognizes the character sequence according to the text arrangement order in the operation ticket. For tabular operation tickets, the corresponding cell region is determined according to the table borders and text region positions, and the text content is read sequentially.
[0079] OCR identification results include at least the dual device identification number, the object being operated on, the initial position, and the target position. The dual device identification number is used to retrieve the interval information model, and the initial and target positions are used to determine the direction of the trolley's position change.
[0080] After OCR recognition is completed, the integrity of key fields is checked. If the device's dual serial number, the operation object, the initial position, or the target position is missing, a structured operation task is not generated, and the operation ticket image is re-acquired or the process enters manual review mode.
[0081] Sub-step S203 involves semantic parsing of the OCR text and generating a structured operation task based on the recognition credibility.
[0082] The central processing unit performs semantic parsing on the OCR text, determines the trolley operation direction based on the operation object, initial position, and target position, and matches the identified dual device serial numbers with the dual device serial numbers in the interval information model to determine the target device dual serial number.
[0083] The central processing unit acquires the confidence scores of OCR text recognition, semantic matching of the operation statement, and dual device ID matching. These scores are then weighted and summed according to their respective fusion weights to obtain the overall confidence score of the operation task. The formula for calculating the overall confidence score of the operation task is as follows: ; ; in, The overall confidence score for the operation task represents the degree of confidence that the current operation ticket recognition result can be used to generate an automatically executed task, and its value ranges from 0 to 1. The confidence score for OCR text recognition represents the reliability of the recognition results of the text character sequence on the ticket, and its value ranges from 0 to 1. The confidence score for semantic matching of operation statements represents the degree of matching between the OCR text and the preset semantic type of vehicle operation, and its value ranges from 0 to 1. The confidence score for the dual device ID matching represents the degree of matching between the dual device ID identified by OCR and the device record in the interval information model, with a value ranging from 0 to 1. The fusion weight corresponding to the confidence score of OCR text recognition represents the proportion of data in the comprehensive judgment of the operation task of OCR text recognition, and the value ranges from 0 to 1. The fusion weight is the confidence score of the semantic matching of the operation statement. It represents the proportion of the semantic matching result of the operation statement in the comprehensive judgment of the operation task. The value range is 0 to 1. The fusion weight is the confidence level of the dual device ID matching. It represents the proportion of the dual device ID matching result in the comprehensive judgment of the operation task, and its value ranges from 0 to 1.
[0084] The preset task confidence threshold is between 0.90 and 1.00. When the overall confidence level of the operation task reaches the task confidence threshold, and the target device's dual ID corresponds to only one valid device interval record, a structured operation task containing the target device's dual ID, initial position, target position, and trolley operation direction is generated.
[0085] The central processing unit retrieves the corresponding equipment interval record in the interval information model based on the dual number of the target equipment, determines the spatial coordinates of the target interval, and determines the target pallet height and target bayonet width based on the pallet height, bayonet width, height calibration compensation data and width calibration compensation data in the equipment interval record.
[0086] See attached document Figure 5 In step S300, the general-purpose transfer vehicle is moved according to the target interval spatial coordinates and multi-source positioning fusion is performed. Based on the fusion position result, target pallet height and target bayonet width, visual docking with the target switch cabinet socket is completed, and mechanical docking completion status is generated.
[0087] This is executed via the positioning and docking module. It includes the following sub-steps.
[0088] Sub-step S301: Generate a movement path based on the current position of the general-purpose transfer vehicle, the spatial coordinates of the target interval, the passable area of the switch room, and the position of fixed obstacles.
[0089] The central processing unit obtains the spatial coordinates of the target interval based on the dual number of the target equipment in the structured operation task, and reads the passable area of the switch room and the location of fixed obstacles.
[0090] The central processing unit generates a movement path based on the current position of the general-purpose transfer vehicle, the spatial coordinates of the target interval, the passable area of the switch room, and the location of fixed obstacles. The mobile chassis travels along the movement path. During the movement, the lidar continuously acquires lidar ranging data to determine the actual passability of the movement path.
[0091] After the general-purpose transfer vehicle enters the area adjacent to the target interval, the RFID device reads the RFID tag on the device to generate an RFID identification result. The central processing unit compares the dual ID of the device corresponding to the RFID identification result with the dual ID of the target device in the structured operation task. If they match, multi-source positioning fusion is then performed.
[0092] Sub-step S302 involves time-aligning the industrial camera positioning data and lidar ranging data and using Kalman filtering to obtain the fused position result.
[0093] Industrial cameras acquire positioning data to indicate the position of the target switchgear relative to the universal transport vehicle; LiDAR acquires ranging data to indicate the distance between the universal transport vehicle and the plane containing the target switchgear. RFID identification results are used to form the target equipment's identity constraint.
[0094] The central processing unit performs time alignment on the industrial camera positioning data and lidar ranging data, ensuring that data within the same control cycle corresponds to the same position state. The time-aligned industrial camera positioning data and lidar ranging data are then input into the Kalman filter process as continuous position observation data. Based on the position state of the previous control cycle and the observation data of the current control cycle, a fused position result is obtained.
[0095] For equipment intervals that have completed positioning calibration, the central processing unit reads historical positioning correction data associated with the dual serial numbers of the target equipment, and performs position compensation on the fused position result based on the historical positioning correction data. The fused position result after position compensation is used to control the general-purpose transfer vehicle to enter the visual docking area.
[0096] Sub-step S303: Based on the fusion position result, target tray height and target bayonet width, complete the visual docking with the target switch cabinet socket and generate a mechanical docking completion status.
[0097] Based on the fusion location results, a general-purpose transfer vehicle is controlled to enter the vision docking area, and an industrial camera captures images of the target switchgear sockets. The cabinet vision recognition unit identifies the socket area, socket edges, and hole mask, and determines the center of the target switchgear socket based on the socket edges and hole mask.
[0098] Before converting pixel position to spatial position, a calibration relationship is established between the industrial camera imaging parameters and the reference plane of the target switch cabinet socket. When perspective changes occur in the image of the target switch cabinet socket, perspective correction is performed based on the target switch cabinet socket reference plane.
[0099] The lateral and vertical spatial deviations are determined based on the calibration relationship between the industrial camera imaging parameters and the reference plane of the target switch cabinet socket. The calculation formulas are as follows: ; ; in, For the first The horizontal spatial deviation corresponding to the frame target switch cabinet socket image represents the distance that needs to be corrected in the horizontal direction between the receiving tray docking reference center and the center of the target switch cabinet socket, in mm; For the first The vertical spatial deviation corresponding to the frame target switch cabinet socket image represents the distance that needs to be corrected in the vertical direction between the receiving tray docking reference center and the center of the target switch cabinet socket, in mm; To collect the first The distance from the optical center of the industrial camera to the reference plane of the target switch cabinet socket when the frame image of the target switch cabinet socket is displayed. It represents the target distance used when converting the pixel position to the spatial position. The unit is mm, and the value is greater than 0. For the first The horizontal pixel coordinates of the socket center identified in the frame target switch cabinet socket image represent the position of the socket center in the horizontal direction of the target switch cabinet socket image, in pixels, and the value is within the effective horizontal pixel range of the target switch cabinet socket image. The horizontal pixel coordinates of the docking reference center, determined by the industrial camera calibration, represent the horizontal reference position of the image when the horizontal spatial deviation is 0. The unit is pixels, and the value is within the effective horizontal pixel range of the target switch cabinet socket image. For the first The vertical pixel coordinates of the socket center identified in the frame target switch cabinet socket image represent the position of the socket center in the vertical direction of the target switch cabinet socket image, in pixels, and the value is within the effective vertical pixel range of the target switch cabinet socket image; The vertical pixel coordinates of the docking reference center, determined by the industrial camera calibration, represent the vertical reference position of the image when the vertical spatial deviation is 0. The unit is pixels, and the value is within the effective vertical pixel range of the target switch cabinet socket image. is the equivalent focal length in the horizontal direction of the industrial camera. It represents the conversion parameter between the horizontal pixel position of the target switch cabinet socket image and the spatial position of the target switch cabinet socket reference plane. The unit is pixel and the value is greater than 0. The vertical equivalent focal length of the industrial camera represents the conversion parameter between the vertical pixel position of the target switch cabinet socket image and the spatial position of the target switch cabinet socket reference plane. The unit is pixels and the value is greater than 0. The frame number of the target switchgear socket image represents the sequence of consecutive acquisitions of the target switchgear socket images during the visual docking phase, and is a positive integer.
[0100] The central processing unit adjusts the mobile chassis of the general-purpose transfer vehicle based on the lateral spatial deviation, adjusts the pallet lifting actuator based on the target pallet height and vertical spatial deviation, and adjusts the bayonet width adjustment actuator to ensure the effective bayonet width reaches the target bayonet width. After each adjustment, the target switchgear socket image is re-acquired, and the lateral and vertical spatial deviations are re-determined.
[0101] The allowable values for lateral and vertical spatial deviations are set within the range of 0.5mm to 3.0mm. When the absolute values of the lateral and vertical spatial deviations are not greater than their respective allowable values, and the receiving pallet height reaches the target pallet height and the effective bayonet width reaches the target bayonet width, the mechanical docking is considered complete.
[0102] See attached document Figure 6 In step S400, a handcart operation execution permission is generated based on the structured operation task and the mechanical docking completion status, and the handcart electric control module is controlled to execute the handcart's exit or push-in operation.
[0103] This is executed through the linkage control module. It includes the following sub-steps.
[0104] Sub-step S401 generates a handcart exit control request or a handcart push-in control request based on the initial position and target position in the structured operation task.
[0105] The central processing unit reads the target device's dual number, initial position, target position, and trolley operation direction from the structured operation task, and reads the trolley's current status.
[0106] When the initial position is the test position and the target position is the maintenance position, a trolley exit control request is generated; when the initial position is the maintenance position and the target position is the test position, a trolley push-in control request is generated. If the current state of the trolley is inconsistent with the initial position in the structured operation task, no trolley action command is sent.
[0107] Sub-step S402 generates a handcart operation execution permit based on the structured operation task and the mechanical docking completion status.
[0108] The central processing unit determines the consistency status of the operation object based on the dual numbering of the target equipment in the structured operation task and the RFID identification result; determines the effective status of the target interval positioning based on the fused location result; determines the effective status of the communication link based on the IEC 61850 communication connection and status message reception; determines the equipment safety status based on the equipment status and operation interlocking conditions returned by the handcart electric control module, and reads the mechanical docking completion status generated in step S300.
[0109] The operation object consistency status, target interval positioning validity status, mechanical docking completion status, communication link validity status, and equipment safety status are represented by verification pass status and verification fail status, respectively. Within the central processing unit, the verification pass status is coded as 1, and the verification fail status is coded as 0. The formula for calculating the trolley operation execution permission status is as follows: ; in, This represents the handcart operation execution permission status, indicating whether it is currently allowed to send handcart action commands to the handcart electric control module. The value is 0 or 1. A value of 1 generates a handcart operation execution permission, while a value of 0 does not generate a handcart operation execution permission. The operation object consistency status indicates whether the dual device ID of the currently identified device is consistent with the dual device ID of the target device in the structured operation task, and the value is 0 or 1. The effective status of the target interval is determined, which represents whether the fused position result corresponds to the target interval of the structured operation task. The value is 0 or 1. This indicates the mechanical docking completion status, representing whether the receiving pallet height, effective bayonet width, and spatial deviation meet the visual docking completion conditions, with a value of 0 or 1. This indicates the effective status of the communication link, representing whether the IEC 61850 communication connection and status message reception meet the control requirements. The value is 0 or 1. The equipment safety status indicates whether the equipment status and operation interlocking conditions returned by the handcart electric control module meet the current handcart operation requirements, and the value is 0 or 1.
[0110] When the operation object consistency status, target interval positioning valid status, mechanical docking completed status, communication link valid status, and equipment safety status are all verified, the handcart operation execution permit status is set to 1, and a handcart operation execution permit is generated; if any of the statuses are verified as unsuccessful, no handcart operation execution permit is generated.
[0111] In sub-step S403, the handcart action command is sent via MMS and status information is received via GOOSE.
[0112] After generating the handcart operation execution permission, the central processing unit sends the handcart action command through the MMS. The handcart electric operation control module executes the handcart exit or push-in operation according to the corresponding handcart exit control request or handcart push-in control request.
[0113] During the operation of the handcart, the handcart's electric control module sends the handcart's current position status, action execution status, and equipment malfunction status via GOOSE. The central processing unit receives this status information via GOOSE and compares the handcart's current position status with the target position in the structured operation task.
[0114] If no valid status message is received for three consecutive communication cycles, the transmission of action commands will stop; if an abnormal device status is received, stop control data will be output. The operation ends when the trolley reaches the target position and the action execution status is completed. When the structured operation task is an exit operation, proceed to step S500; when the structured operation task is a push-in operation, exit the current control process after the trolley completes the push-in operation.
[0115] See attached document Figure 7 In step S500, when the structured operation task is an exit operation, the exited handcart is transferred to the target maintenance station of the maintenance platform and the positioning and locking are completed.
[0116] This is performed via the transfer and resettlement module. It includes the following sub-steps.
[0117] Sub-step S501: Confirm that the exiting handcart has entered the receiving pallet and the mechanical limit structure is in place, and release the mechanical docking state.
[0118] When the structured operation task is an exit operation, the trolley exits from the target switchgear and enters the receiving pallet. The central processing unit reads the trolley position status and the receiving pallet status, and confirms that the exiting trolley has entered the receiving pallet.
[0119] The receiving pallet is equipped with a mechanical limit structure to restrict the continued movement of the exiting trolley relative to the receiving pallet. Once it is confirmed that the exiting trolley has entered the receiving pallet and the mechanical limit structure is in place, the mechanical docking between the general-purpose transfer vehicle and the target switchgear is released.
[0120] The general-purpose transfer vehicle moves in the opposite direction of docking and obtains lidar ranging data through lidar. Once the general-purpose transfer vehicle reaches the specified disengagement distance from the target switchgear, it enters the path planning process of the maintenance platform.
[0121] Sub-step S502: Generate candidate transfer paths and determine the target transfer path for idle handcart maintenance workstations.
[0122] The central processing unit reads the location of the maintenance platform and the occupancy status of each handcart maintenance bay, and generates candidate transfer paths for vacant handcart maintenance bays. Each candidate transfer path consists of multiple continuous path segments. Each path segment records the path length and allowed driving speed, and records the additional waiting time caused by turning, obstacle avoidance, passage confirmation, or speed switching between adjacent path segments.
[0123] The estimated travel time for each candidate diversion route is determined based on route length, permitted speed, and additional waiting time. The formula for calculating the estimated travel time of a candidate diversion route is as follows: ; in, The estimated travel time for the candidate transfer path represents the path evaluation time required for a general-purpose transfer vehicle to move from its current location to the target maintenance station on the maintenance platform, in seconds. The number of path segments contained in the candidate transfer path, which represents the total number of continuous path segments obtained by dividing the complete candidate transfer path, and is an integer greater than or equal to 1. For the first The length of each path segment represents the length of the general-purpose transfer vehicle in the first path segment. The actual distance that each path segment needs to move, in meters, and the value is greater than or equal to 0; For the first The permissible driving speed for each path segment represents the operating speed set for the general-purpose transfer vehicle for the corresponding path segment, in m / s, and the value is greater than 0. For the first The additional waiting time between a path segment and the next path segment represents the additional time caused by turning, obstacle avoidance, passage confirmation or speed switching, and is in seconds, with a value greater than or equal to 0. is the path segment number, which represents the path segment index in the candidate transfer path, and is a positive integer.
[0124] The central processing unit calculates the estimated travel time for each candidate transfer path and determines the candidate transfer path with the shortest estimated travel time as the target transfer path. The general-purpose transfer vehicle moves according to the target transfer path and continuously monitors the actual traffic status of the target transfer path using lidar.
[0125] Sub-step S503: Send the withdrawn handcart into the target maintenance station and complete the positioning and locking.
[0126] After the general-purpose transfer vehicle arrives at the target maintenance station on the maintenance platform, it aligns the receiving pallet with the target maintenance station and adjusts the position of the target maintenance station by moving the slide rail.
[0127] After the trolley is moved into the target maintenance station, the positioning post restricts the trolley's lateral position, the mechanical lock hole connects with the trolley's position locking structure, the locking hook prevents the trolley from leaving the target maintenance station, and the braking mechanism enters the locked state. After the positioning and locking are completed through the positioning post, mechanical lock hole, locking hook, and braking mechanism, the central processing unit updates the target maintenance station to an occupied state.
[0128] When there are subsequent structured operation tasks in the operation task queue, the central processing unit reads the next structured operation task and enters the subsequent control process; when there are no subsequent structured operation tasks, the general-purpose transfer vehicle moves to the designated standby position.
[0129] Specific application examples: See attached document Figure 8 This embodiment is applied to a 10kV switch room in a substation, performing the operation of moving a 10kV handcart from the test position to the maintenance position and transferring it to the target maintenance work position on the maintenance platform.
[0130] Pallet height corresponding to the target equipment interval Take 900mm as the height calibration compensation value. Take 2mm as the bayonet width parameter. Take 520mm as the width calibration compensation value. Take 4mm. Substitute the above parameters into the calculation formula for the transfer vehicle adaptation setting value in step S100, and the target pallet height is calculated. The sum of 900mm and 2mm, i.e., 902mm; target bayonet width. The sum of 520mm and 4mm is 524mm. The central processing unit controls the pallet lifting actuator and the bayonet width adjustment actuator to complete the pre-adjustment of the receiving pallet.
[0131] After the industrial camera captures the image of the operation ticket, it performs preprocessing, OCR recognition, and semantic parsing. In this embodiment, the confidence level of OCR text recognition is... The confidence score for semantic matching of the operation statement is set to 0.98. Take 0.96, confidence level for dual device number matching Set to 1.00; fusion weights corresponding to OCR text recognition confidence scores. The fusion weight is set to 0.45, corresponding to the semantic matching confidence score of the operation statement. The fusion weight is set to 0.35, corresponding to the confidence level of the device's dual serial number matching. Take 0.20.
[0132] Substitute the above parameters into the formula for calculating the overall confidence level of the operation task in step S200, and the overall confidence level of the operation task is calculated. The sum of 0.45×0.98, 0.35×0.96, and 0.20×1.00 equals 0.977. In this embodiment, the task confidence threshold is set to 0.95. The overall confidence level of the operation task is 0.977, which meets the task confidence threshold. Since the dual ID of the target device corresponds to only one valid device interval record, a structured operation task containing the dual ID of the target device, the initial position, the target position, and the direction of the handcart operation is generated.
[0133] The central processing unit determines the target interval spatial coordinates, target pallet height, and target bayonet width in the interval information model based on the target equipment's dual identification number. Subsequently, it controls the movement of the general-purpose transfer vehicle according to the target interval spatial coordinates, and completes target equipment confirmation and multi-source positioning fusion using RFID identification results, industrial camera positioning data, and LiDAR ranging data to obtain the fused position result. Based on the fused position result, it controls the general-purpose transfer vehicle to enter the visual docking area.
[0134] During a vision docking control cycle, the distance from the optical center of the industrial camera to the reference plane of the target switchgear socket is... Take 1200mm, horizontal pixel coordinates of the socket center. Take 964 pixels and align them with the horizontal pixel coordinates of the reference center. Take 960 pixels, vertical pixel coordinates of the center of the slot. Take 538 pixels and align them with the vertical pixel coordinates of the reference center. Take 540 pixels as the equivalent focal length in the horizontal direction of an industrial camera. Take 1600 pixels as the equivalent focal length in the vertical direction of the industrial camera. Get 1600 pixels.
[0135] Substitute the above parameters into the lateral space deviation calculation formula in step S300. Multiply 1200mm by the difference between 964 pixels and 960 pixels, then divide by 1600 pixels to get 3.0mm; substitute the above parameters into the vertical spatial deviation calculation formula, Multiplying 1200mm by the difference between 538 and 540 pixels, and then dividing by 1600 pixels, yields -1.5mm. The central processing unit adjusts the mobile chassis of the general-purpose transfer vehicle based on the lateral spatial deviation, adjusts the pallet lifting actuator based on the target pallet height and vertical spatial deviation, and adjusts the bayonet width adjustment actuator to ensure the effective bayonet width reaches the target bayonet width.
[0136] In this embodiment, the allowable values for both lateral and vertical spatial deviations are set to 1.5 mm. (See attached image) Figure 8 During the control process shown, the initial lateral absolute docking deviation was 22.0 mm, and the lateral absolute docking deviations for the first to sixth control cycles were 14.2 mm, 8.1 mm, 4.3 mm, 2.4 mm, 1.3 mm, and 0.9 mm, respectively; the initial vertical absolute docking deviation was 18.0 mm, and the vertical absolute docking deviations for the first to sixth control cycles were 11.3 mm, 6.4 mm, 3.2 mm, 1.8 mm, 1.1 mm, and 0.7 mm, respectively.
[0137] From the appendix Figure 8 As can be seen, in the 5th control cycle, the lateral absolute docking deviation was 1.3mm and the vertical absolute docking deviation was 1.1mm, both of which were not greater than their respective allowable values. The central processing unit further confirmed that the receiving pallet height reached the target pallet height of 902mm and the effective bayonet width reached the target bayonet width of 524mm before generating a mechanical docking completion status.
[0138] After visual docking is completed, in this embodiment, the operation object consistency status, target interval positioning valid status, mechanical docking completion status, communication link valid status, and equipment safety status are all verified as passed, and are all encoded as 1 within the central processing unit. Substituting the above status values into the handcart operation execution permission status calculation formula in step S400, we obtain... The value is 1×1×1×1×1, which is 1, therefore a handcart operation execution permit is generated.
[0139] In this embodiment's structured operation task, the initial position is the test position, and the target position is the maintenance position; therefore, a trolley exit control request is generated. After generating the trolley operation execution permission, the central processing unit sends the trolley action command via MMS. The trolley electric control module executes the trolley exit operation and sends the trolley's current position status, action execution status, and equipment abnormality status via GOOSE. The trolley exit operation ends when the trolley reaches the target position and the action execution status is completed.
[0140] When the structured operation task is an exit operation, after confirming that the exiting trolley has entered the receiving pallet and the mechanical limit structure is in place, the mechanical docking state between the general-purpose transfer vehicle and the target switchgear is released. This embodiment selects a candidate transfer path consisting of three path segments. Take 3. The first path segment Take 6.0m Take 0.5 m / s; the second path segment Take 4.0m Take 0.4 m / s; the third path segment Take 3.0m, Take 0.3 m / s. The distance between two adjacent path segments. Take 1.5s and 1.0s respectively.
[0141] Substituting the above parameters into the formula for calculating the estimated travel time of the candidate transfer path in step S500, the first path segment corresponds to 12s, the second path segment to 10s, and the third path segment to 10s, with a total additional waiting time of 2.5s. Therefore... The time is the sum of 12s, 10s, 10s, and 2.5s, which equals 34.5s. The central processing unit compares this candidate transfer path with other candidate transfer paths and determines the candidate transfer path with the shortest expected travel time as the target transfer path.
[0142] After the general-purpose transfer vehicle arrives at the target maintenance station on the maintenance platform, it sends the decommissioned handcart into the target maintenance station. Positioning and locking are achieved through positioning posts, mechanical lock holes, locking hooks, and a braking mechanism, and the target maintenance station is updated to an occupied state. This completes the following processes: operation ticket image processing, structured operation task generation, general-purpose transfer vehicle movement, multi-source positioning fusion, visual docking with the target switchgear interface, handcart electric operation linkage, and the transfer and placement of the decommissioned handcart.
[0143] Experimental verification and effect comparison: See attached document Figure 9 During the verification process, under the same target interval, the same handcart operation direction, and the same starting position, 10 identical handcart transfer tasks were completed using both the traditional manual transfer method and the control method of this invention.
[0144] The single-interval complete operation times for the 1st to 10th operations using the traditional manual transfer method are 224s, 231s, 219s, 226s, 235s, 228s, 221s, 230s, 223s, and 229s, respectively, with an average time of 226.6s for 10 operations. The single-interval complete operation times for the 1st to 10th operations using the control method of this invention are 108s, 104s, 111s, 106s, 102s, 109s, 105s, 107s, 103s, and 106s, respectively, with an average time of 10 operations of 10 times. The control method of this invention reduces the average complete operation time by 120.5s compared to the traditional manual transfer method, representing a reduction of 53.2%.
[0145] From the appendix Figure 9 It can be seen that the single-interval complete operation time of the traditional manual transfer method is distributed in the range of 219s to 235s, while the single-interval complete operation time of the control method of the present invention is distributed in the range of 102s to 111s. The maximum single-interval complete operation time of the traditional manual transfer method is 235s, and the minimum single-interval complete operation time is 219s; the maximum single-interval complete operation time of the control method of the present invention is 111s, and the minimum single-interval complete operation time is 102s.
[0146] In the same 10 tasks, the final spatial deviation between the reference center of the receiving pallet and the center of the target switchgear socket was recorded when the mechanical docking was completed. The final spatial deviations corresponding to the traditional manual transfer method were 7.1mm, 6.8mm, 7.5mm, 6.9mm, 7.4mm, 7.2mm, 6.7mm, 7.3mm, 6.9mm and 7.1mm, respectively, with an average final spatial deviation of 7.09mm. The final spatial deviations corresponding to the control method of this invention were 1.4mm, 1.1mm, 1.3mm, 1.2mm, 1.5mm, 1.0mm, 1.2mm, 1.1mm, 1.3mm and 1.2mm, respectively, with an average final spatial deviation of 1.23mm, representing a reduction of 5.86mm in the average final spatial deviation.
[0147] When verifying the image processing of the operation ticket, the image acquisition position and local illumination are changed while keeping the text content of the operation ticket unchanged. Preprocessing, OCR recognition, and semantic parsing are performed sequentially. If the overall confidence level of the operation task does not reach the task confidence threshold, the central processing unit will not generate a structured operation task and will not send automatic execution instructions to the mobile chassis and the handcart electric control module.
[0148] When verifying the multi-source positioning fusion process, partial field-of-view occlusion of industrial cameras, single-read anomalies of RFID, and partial ranging anomalies of LiDAR are set up respectively. The central processing unit constrains the target device identity based on the RFID identification result and performs time alignment and Kalman filtering on the valid industrial camera positioning data and LiDAR ranging data. When a fused position result that meets the requirements of target device identity and target interval position can be obtained, the visual docking with the target switch cabinet interface continues; if a valid fused position result cannot be obtained, the subsequent visual docking process is not initiated.
[0149] When verifying the linkage control process, the following states are set to "verification failed": operation object consistency status, target interval positioning validity status, mechanical docking completion status, communication link validity status, or equipment safety status. If any of these states is "verification failed," no trolley operation execution permission is generated, and the central processing unit does not send trolley action commands. A trolley operation execution permission is generated only when all four states—operation object consistency status, target interval positioning validity status, mechanical docking completion status, communication link validity status, and equipment safety status—are all "verification passed."
[0150] When verifying an abnormal communication status, status message transmission is stopped during the operation of the handcart. If no valid status message is received for three consecutive communication cycles, the communication link status changes to a verification failure status, and the central processing unit stops sending action commands. After communication is restored, the current device status is reread, and a new determination is made as to whether to generate a handcart operation execution permission.
[0151] When verifying the transfer and placement process of multiple used vehicles, multiple vacant used vehicle maintenance bays are set up on the maintenance platform. After the first used vehicle completes positioning and locking, the corresponding target maintenance bay is updated to an occupied state. When the subsequent structured operation task is an exit operation, the central processing unit generates candidate transfer paths only for the remaining vacant used vehicle maintenance bays, and determines the estimated passage time based on the path length, allowed travel speed, and additional waiting time. The candidate transfer path with the shortest estimated passage time is determined as the target transfer path, thereby sending the different exited used vehicles to their corresponding target maintenance bays.
[0152] In this embodiment, the target equipment's dual number, target interval spatial coordinates, target pallet height, and target bayonet width are transmitted from the interval information model to the subsequent control process; the structured operation task is transmitted to the general-purpose transfer vehicle movement, multi-source positioning fusion, and handcart-electric operation linkage process; the fused position result is transmitted to the visual docking and target interval positioning validity status judgment process; the mechanical docking completion status and the structured operation task are used together to generate the handcart operation execution permission; when the structured operation task is an exit operation, the exited handcart continues to enter the maintenance platform transfer and target maintenance workstation positioning and locking process. This forms a continuous control process from operation ticket image processing, structured operation task generation, target interval positioning, target switchgear socket visual docking, handcart-electric operation linkage to handcart transfer and placement.
Claims
1. A control method for a general-purpose transfer vehicle in a substation based on image perception, characterized in that, Includes the following steps: Establish an interval information model to associate the dual serial number of the equipment with the interval spatial coordinates, pallet height, and bayonet width; The operation ticket image is collected, preprocessed, OCR recognized, and semantically parsed to generate a structured operation task containing the target device's dual number, initial position, target position, and handcart operation direction. Based on the target device's dual number, the target interval spatial coordinates, target pallet height, and target gate width are determined in the interval information model. The general-purpose transfer vehicle is controlled to move according to the target interval spatial coordinates and multi-source positioning fusion is performed. Based on the fusion position result, the target pallet height and the target bayonet width, visual docking with the target switch cabinet socket is completed, and a mechanical docking completion status is generated. Based on the structured operation task and the mechanical docking completion status, a handcart operation execution permission is generated, and the handcart electric control module is controlled to execute the handcart's exit or push-in operation. When the structured operation task is an exit operation, the exited handcart is transferred to the target maintenance station on the maintenance platform and its positioning and locking are completed.
2. The image-sensing-based control method for a general-purpose transfer vehicle in a substation, as described in claim 1, is characterized in that... The specific steps for establishing the interval information model and associating the dual device serial number with the interval spatial coordinates, pallet height, and bayonet width are as follows: Establish equipment interval records, linking the interval spatial coordinates, pallet height, and bayonet width with the equipment's dual serial number; Mechanical zero-point calibration was performed on the pallet lifting actuator and the bayonet width adjustment actuator of the transfer vehicle to obtain height calibration compensation data and width calibration compensation data; The device interval record is retrieved based on the dual serial number of the target device. The target pallet height is determined based on the retrieved pallet height and the height calibration compensation data. The target bayonet width is determined based on the retrieved bayonet width and the width calibration compensation data.
3. The image-sensing-based control method for a general-purpose transfer vehicle in a substation, as described in claim 1, is characterized in that... The specific steps for preprocessing, OCR recognition, and semantic parsing of the acquired ticket image are as follows: Operation ticket images are captured using an industrial camera, and noise suppression, grayscale normalization, and binarization are performed sequentially. Detect text lines or table borders using Hough transform and perform skew correction; The corrected operation ticket image is subjected to OCR recognition to obtain the dual device number, operation object, initial position and target position; Semantic parsing is performed on the OCR text to determine the handcart operation direction based on the operation object, the initial position, and the target position.
4. The image-sensing-based control method for a general-purpose transfer vehicle in a substation, as described in claim 1, is characterized in that... The specific steps for generating a structured operation task that includes the target device's dual identification number, initial position, target position, and trolley operation direction are as follows: The confidence scores of OCR text recognition, semantic matching of operation statements, and dual device ID matching are obtained. They are then weighted and summed according to their respective fusion weights to obtain the comprehensive confidence score of the operation task. The preset task confidence threshold is between 0.90 and 1.00; When the overall confidence level of the operation task reaches the task confidence threshold, and the dual ID of the target device corresponds to only one valid device interval record, a structured operation task containing the dual ID of the target device, the initial position, the target position, and the trolley operation direction is generated.
5. The image-sensing-based control method for a general-purpose transfer vehicle in a substation, as described in claim 1, is characterized in that... The specific steps for controlling the movement of the general-purpose transport vehicle based on the target interval spatial coordinates and performing multi-source positioning fusion are as follows: A movement path is generated based on the current position of the general-purpose transfer vehicle, the spatial coordinates of the target interval, the passable area of the switch room, and the position of fixed obstacles. RFID identification results are obtained through RFID devices, industrial camera positioning data is obtained through industrial cameras, and lidar ranging data is obtained through lidar. The industrial camera positioning data and the lidar ranging data are time-aligned, Kalman filtering is used to obtain the fused position result, and position compensation is performed based on historical positioning correction data.
6. The image-sensing-based control method for a general-purpose transfer vehicle in a substation, as described in claim 1, is characterized in that... The specific steps for visually aligning with the target switch cabinet socket based on the fusion location result, the target tray height, and the target bayonet width, and generating a mechanical alignment completion status, are as follows: Based on the fusion position result, the general-purpose transfer vehicle is controlled to enter the visual docking area, acquire the target switch cabinet socket image, identify the socket edge and hole mask, and determine the center of the target switch cabinet socket. Based on the calibration relationship between the industrial camera imaging parameters and the target switch cabinet socket reference plane, determine the lateral spatial deviation and the vertical spatial deviation; Adjust the mobile chassis of the general-purpose transfer vehicle according to the lateral spatial deviation, adjust the pallet lifting actuator according to the target pallet height and the vertical spatial deviation, and adjust the bayonet width adjustment actuator so that the effective bayonet width reaches the target bayonet width. When the absolute values of the lateral spatial deviation and the vertical spatial deviation are not greater than their respective allowable values, and the receiving pallet height reaches the target pallet height and the effective bayonet width reaches the target bayonet width, the mechanical docking completion state is generated.
7. The image-sensing-based control method for a general-purpose transfer vehicle in a substation, as described in claim 1, is characterized in that... The specific steps for generating a handcart operation execution permit based on the structured operation task and the mechanical docking completion status are as follows: The consistency status of the operation object is determined based on the dual identification number of the target device in the structured operation task and the RFID identification result. Determine the effective status of target interval positioning based on the fusion location results; The effective status of the communication link is determined based on the communication connection and status message reception information in IEC 61850. The safety status of the equipment is determined based on the equipment status and operation interlocking conditions returned by the handcart's electric control module. When the operation object is in a consistent state, the target interval positioning is in a valid state, the mechanical docking is in a completed state, the communication link is in a valid state, and the equipment security state are all in a verified state, the handcart operation execution permit is generated.
8. The image-sensing-based control method for a general-purpose transfer vehicle in a substation, as described in claim 1, is characterized in that... The specific steps for the control module to perform the handcart's exit or push-in operation are as follows: Based on the initial position and target position in the structured operation task, generate a handcart exit control request or a handcart push-in control request; After generating the handcart operation execution permission, the handcart action command is sent via MMS, and the current position status of the handcart, action execution status and equipment abnormal status are received via GOOSE; If no valid status message is received for three consecutive communication cycles, the sending of the handcart action command will be stopped; if the abnormal status of the device is received, stop control data will be output. The operation ends when the handcart reaches the target position and the action execution status is completed.
9. The image-sensing-based control method for a general-purpose transfer vehicle in a substation, as described in claim 1, is characterized in that... When the structured operation task is an exit operation, the specific steps for transferring the exited handcart to the target maintenance station on the maintenance platform and completing the positioning and locking are as follows: Once the withdrawing trolley has entered the receiving pallet and the mechanical limit structure is in place, the mechanical docking between the general-purpose transfer vehicle and the target switchgear is released. For idle handcart maintenance workstations, candidate transfer routes are generated. The estimated travel time of each candidate transfer route is determined based on the route length, allowed travel speed, and additional waiting time. The candidate transfer route with the shortest estimated travel time is determined as the target transfer route. The withdrawn handcart is sent into the target maintenance station, and the positioning and locking are completed by the positioning column, mechanical lock hole, locking hook and braking mechanism, and the target maintenance station is updated to the occupied state.
10. A general-purpose transfer vehicle control system for substations based on image perception, characterized in that, The method for controlling a general-purpose substation transport vehicle based on image perception as described in any one of claims 1-9 includes: The interval modeling module is used to establish an interval information model, which associates the dual serial number of the equipment with the interval spatial coordinates, pallet height and bayonet width; The image perception module is used to acquire operation ticket images for preprocessing, OCR recognition and semantic parsing, generate a structured operation task containing the target device's dual number, initial position, target position and handcart operation direction, and determine the target interval spatial coordinates, target pallet height and target gate width in the interval information model based on the target device's dual number. The positioning and docking module is used to control the movement of the general-purpose transfer vehicle according to the target interval spatial coordinates and to perform multi-source positioning fusion. Based on the fusion position result, the target pallet height and the target bayonet width, it completes the visual docking with the target switch cabinet socket and generates a mechanical docking completion status. The linkage control module is used to generate a handcart operation execution permission based on the structured operation task and the mechanical docking completion status, and control the handcart electric operation control module to execute the handcart's exit or push-in operation. The transfer and placement module is used to transfer the exited handcart to the target maintenance station of the maintenance platform and complete the positioning and locking when the structured operation task is an exit operation. The interval modeling module, the image perception module, the positioning and docking module, the linkage control module, and the transfer and resettlement module interact with each other through the central processing unit.
