Intelligent navigation automatic transportation system and method for 10kV switch trolley

Abstract

The invention discloses a 10kV switch trolley intelligent navigation automatic transportation system and method. A five-layer modular architecture of a sensing layer, a transmission layer, a decision-making layer, an execution layer and an interaction layer is adopted. The sensing layer realizes high-precision environment modeling and positioning through four-direction stereoscopic vision, a hybrid positioning system (UWB + IMU + RFID) and a pressure sensor array in combination with SGM, ICP and ESKF algorithms; the decision-making layer operates an improved A * path planning algorithm and a DWA obstacle avoidance strategy, electrical equipment safety spacing constraints are embedded, and a crab-walking mode is started in a narrow channel; the execution layer drives omnidirectional movement through servo three-ring cascade control, and the transmission layer adopts a double-link redundancy design to ensure reliable data transmission; the interaction layer integrates two-factor authentication and digital twinborn visualization functions. The system solves the problems that traditional manual transportation is low in efficiency and poor in positioning precision, and existing automatic equipment is insufficient in adaptability, millimeter-level precise transportation and whole-process remote monitoring are achieved, manual or crane loading and unloading are matched, and the system is suitable for the high-precision operation and maintenance requirements of the 10kV switch trolley of the electric power system.

Description

Technical Field

[0001] This invention belongs to the field of power equipment automation technology, specifically relating to a 10kV switch trolley intelligent navigation automated transportation system and method. Background Technology

[0002] In the operation and maintenance of power systems, the transportation of 10kV switchgear is a crucial link. Existing 10kV switchgear relies on manual operation, which presents two major technical problems: First, manual transportation is inefficient and lacks positioning accuracy, easily leading to equipment damage or safety hazards due to operational errors, failing to meet the high-precision operation and maintenance requirements of power equipment; second, existing automated transportation equipment mostly uses a single navigation method, lacking deep interaction with power equipment signals, making it difficult to cope with dynamic obstacles in complex power environments, and unable to achieve dynamic path planning and autonomous repositioning, resulting in insufficient adaptability and reliability. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent navigation automated transportation system and method for 10kV switch trolleys, addressing the problems of low efficiency, insufficient positioning accuracy, safety hazards, and difficulty in dealing with dynamic obstacles in complex power environments associated with existing manual transportation of 10kV switch trolleys.

[0004] To solve the above problems, the technical solution of the present invention is as follows: The 10kV switchgear intelligent navigation automated transportation system comprises a five-layer modular architecture consisting of a perception layer, a transmission layer, a decision-making layer, an execution layer, and an interaction layer. Each layer achieves fully automated closed-loop control through signal interaction. The perception layer and the decision-making layer establish a bidirectional data connection through the transmission layer, and the decision-making layer and the execution layer are connected through control signal lines. The interaction layer establishes communication connections with both the perception layer and the decision-making layer. The perception layer is used for high-precision environmental modeling and positioning. The decision-making layer completes dynamic path planning and heading correction based on the data collected by the perception layer. The execution layer realizes movement according to the instructions of the decision-making layer. The transmission layer is used for data transmission between layers, and the interaction layer is used to provide human-machine interaction and remote monitoring.

[0005] Furthermore, the perception layer includes a four-way stereo vision unit, a hybrid positioning system, and a pressure sensor array. The four-way stereo vision unit is deployed at the four corners of the switch trolley and includes a binocular camera, a ToF sensor, and an infrared fill light. The outputs of the binocular camera, ToF sensor, and infrared fill light are all connected to the signal processing module of the perception layer. The signal processing module constructs a 3D environment map by registering the SGM algorithm with the ICP point cloud. The hybrid positioning system includes a UWB base station network, an IMU, and an RFID verification tag. The outputs of the UWB base station network and the IMU are both connected to the fusion processing module of the perception layer. The RFID verification tag is communicatively connected to the fusion processing module, which achieves positioning through the ESKF algorithm. The pressure sensor array is deployed at the four corners of the switch trolley's carrying platform. The output of each pressure sensor is connected to the signal acquisition module of the perception layer for detecting the switch loading status and off-center load alarms. The signal processing module, fusion processing module, and signal acquisition module of the perception layer are all connected to the input of the transmission layer, enabling data interaction with the decision layer through the transmission layer.

[0006] Furthermore, the transmission layer includes a dual-link module and a multi-protocol processing module. The dual-link module consists of a wireless communication module and a wired communication module. The multi-protocol processing module includes a control command protocol unit, a sensor data protocol unit, and a cloud communication protocol unit. The input end of the multi-protocol processing module is electrically connected to the signal output end of the perception layer, and the output end of the multi-protocol processing module is electrically connected to the signal input end of the decision layer. The multi-protocol processing module is bidirectionally connected to both the wireless communication module and the wired communication module. The wireless communication module and the wired communication module are deployed in parallel to achieve redundancy backup. The multi-protocol processing module connects to the transmission requirements of different types of data through corresponding protocol units, forming a closed-loop link for data transmission between layers.

[0007] Furthermore, the decision-making layer includes a main controller and a real-time controller. The main controller is used for environment modeling and path planning, while the real-time controller is used for real-time issuance of motion control commands. The decision-making layer runs an improved A... Path planning algorithms and DWA local obstacle avoidance strategy, the improved A* algorithm Embedding safety distance constraints for power equipment, the DWA obstacle avoidance strategy optimizes motion parameters through an evaluation function, which is S=aH+bO+cP, where H is the heading deviation, O is the obstacle distance, P is the path fit, and a, b, and c are weighting coefficients with a+b+c=1.

[0008] Furthermore, the execution layer includes an omnidirectional drive module and a servo control module. The omnidirectional drive module consists of a brushless motor, a reducer, and multiple independent Mecanum wheels. The servo control module includes a current loop control unit, a speed loop control unit, and a position loop control unit. The output shaft of the brushless motor is drivenly connected to the input end of the reducer, and the output end of the reducer is drivenly connected to the shaft of the Mecanum wheel. The input end of the servo control module is electrically connected to the output end of the real-time controller of the decision layer, and the output end of the servo control module is electrically connected to the brushless motor. The current loop control unit, speed loop control unit, and position loop control unit are cascaded sequentially to form a three-loop collaborative control link, driving the omnidirectional drive module to achieve motion.

[0009] Furthermore, the interaction layer includes an identity authentication module, a visualization module, an instruction input module, and a communication interface module. The identity authentication module adopts the GMM-UBM model and includes a voiceprint acquisition unit and a two-factor authentication unit. The visualization module is a digital twin platform, and the instruction input module is an industrial touchscreen. The output of the voiceprint acquisition unit is connected to the input of the two-factor authentication unit, and the output of the two-factor authentication unit is electrically connected to the decision layer through the communication interface module. The industrial touchscreen and the digital twin platform have bidirectional data communication, and the signal terminal of the industrial touchscreen is connected to the perception layer and the decision layer through the communication interface module. The digital twin platform is used for 3D visualization of equipment status, and the industrial touchscreen is used for natural language instruction input and monitoring data display.

[0010] Furthermore, the fusion processing module of the hybrid positioning system incorporates a dynamic weight adjustment unit and a position reset unit. The dynamic weight adjustment unit is connected to the signal output terminals of the UWB base station network and the IMU, respectively, while the position reset unit is communicatively connected to the RFID verification tag. The fusion processing module achieves optimal estimation through the ESKF algorithm unit, which engages in bidirectional data interaction with the dynamic weight adjustment unit and the position reset unit. The signal processing module of the four-way stereo vision unit incorporates an iterative solution unit, which is connected to the signal output terminals of the binocular camera and the ToF sensor to achieve point cloud registration. The signal processing module communicates with the environmental modeling unit of the decision layer through the transmission layer.

[0011] Furthermore, each pressure sensor in the pressure sensor array is electrically connected to the signal input terminal of the signal acquisition module of the sensing layer, and the signal acquisition module has a built-in differential pressure determination unit; the output terminal of the differential pressure determination unit is connected to the motion control command output unit of the decision layer through the transmission layer, and is used to trigger attitude adjustment signals; the fusion processing module of the hybrid positioning system has a built-in heading correction model unit, which is connected to the path planning algorithm unit of the decision layer. The heading correction model unit obtains data through the positioning reference point detection unit and the obstacle distance detection unit, so that the distance between the positioning reference point and the obstacle is equal to the average value of the preset standard distance range.

[0012] Furthermore, the main controller of the decision layer incorporates a crab-walking mode control unit and an obstacle avoidance algorithm unit; the crab-walking mode control unit is connected to the four-wheel synchronous steering control module, and the output of the four-wheel synchronous steering control module is electrically connected to the servo control module of the execution layer through a real-time controller; the obstacle avoidance algorithm unit runs the DWA algorithm and communicates data with the improved A* algorithm of the path planning algorithm unit; the decision layer incorporates a heading correction calculation unit, which is connected to the heading correction model unit of the fusion processing module, and the heading correction calculation unit uses formulas... or The corrected travel time is calculated, and its output is connected to the motion control command output unit of the real-time controller; the cost function calculation unit of the improved A* algorithm is connected to the dangerous area weight configuration unit, which presets different values ​​according to the equipment type.

[0013] A method for intelligent navigation and automated transportation of a 10kV switch trolley includes the following steps: S1: System initialization, the perception layer completes device calibration and adjustment, the transmission layer establishes a dual-link connection, and the decision layer loads the environment map and security boundary parameters; S2: Task triggered, inputting transportation instructions and completing identity authentication through the interaction layer, and detecting the switch loading status and initial positioning through the perception layer; S3: Environmental perception, four-way stereo vision unit builds real-time 3D map and tracks obstacles, hybrid positioning system outputs high-precision pose data; S4: Path planning and heading correction. The decision-making layer generates a global path by improving the A* algorithm, dynamically adjusts the local path by combining the DWA algorithm, corrects driving deviations by the heading correction model, activates "crab mode" in narrow passages, and meets safety distance constraints during the path planning process. S5: Motion control. The execution layer achieves precise motion based on decision commands, and the pressure sensor array monitors the loading status in real time, triggering adjustments when abnormalities occur. S6: Task completed. After completion, the system automatically plans the return route and uses a hybrid positioning system to achieve precise positioning of the storage area. The device then enters a hibernation preparation state.

[0014] The beneficial effects of this invention are as follows: 1. This invention solves the problems of low efficiency and poor positioning accuracy in traditional manual transportation by using a five-layer modular architecture and multimodal data fusion. The transportation positioning accuracy reaches a certain level, which greatly reduces labor costs and the risk of operational errors.

[0015] 2. By adopting an improved A* algorithm and DWA obstacle avoidance strategy, combined with a hybrid positioning system and four-way stereo vision perception, the problem of lack of power equipment signal interaction and inability to dynamically plan paths in a single navigation method is solved. It can adapt to complex power environments and narrow passage scenarios, maintain a safe distance between the path and live parts, and improve the safety and reliability of the transportation process.

[0016] 3. It integrates functions such as voiceprint authentication, digital twin and dual-link transmission to realize full-process automation and remote monitoring of transportation. It is easy to operate and efficient in operation and maintenance, and is suitable for the intelligent transportation needs of 10kV switch trolleys in various power systems. Attached Figure Description

[0017] The invention will be further described below with reference to the accompanying drawings: Figure 1 This is a three-dimensional structural diagram of the present invention. Figure 2 This is a schematic diagram of the overall hierarchical connection relationship of the system of the present invention. Figure 3 This is a schematic diagram showing the connection relationship of the internal components of the sensing layer of the present invention. Figure 4 This is a schematic diagram showing the connection relationship between the decision-making layer and other components at other levels in this invention. Figure 5 This is a schematic diagram illustrating the connection relationship between the execution layer and the decision layer of this invention. Figure 6 This is a schematic diagram illustrating the internal connections of the interaction layer and other layers in this invention.

[0018] In the diagram: 1. Mecanum wheel; 2. Brushless motor; 3. Infrared fill light; 4. ToF sensor; 5. Binocular camera; 6. Switch trolley; 7. Pressure sensor; 8. Cargo box. Detailed Implementation

[0019] The technical solutions of 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.

[0020] The 10kV switchgear intelligent navigation automated transportation system comprises a five-layer modular architecture consisting of a perception layer, a transmission layer, a decision-making layer, an execution layer, and an interaction layer. Each layer forms a closed-loop control through clearly defined signal connections, and achieves full-process automated closed-loop control through signal interaction. The perception layer and the decision-making layer establish a bidirectional data connection through the transmission layer, and the decision-making layer and the execution layer are connected through control signal lines. The interaction layer establishes communication connections with both the perception layer and the decision-making layer. The perception layer is responsible for high-precision environmental modeling and positioning data acquisition. The decision-making layer completes dynamic path planning and heading correction based on the acquired data. The execution layer executes movement according to decision-making instructions. The transmission layer undertakes data transmission tasks between layers. The interaction layer provides human-machine interaction interface and remote monitoring functions. This led to the construction of a modular system framework with a clear structure and well-defined functions, solving the problems of fragmented structure and poor coordination of existing 10kV switchgear transportation equipment, and providing a stable foundation for the precise collaboration of subsequent functional modules.

[0021] Furthermore, the perception layer includes a four-way stereo vision unit, a hybrid positioning system, and an array of pressure sensors 7. The four-way stereo vision unit is deployed at the four corners of the switch cart 6 and includes a binocular camera 5 (Basler acA2440-20gc), a ToF sensor 4 (TI OPT8241), and an infrared fill light 3 (wavelength 850nm). The outputs of the binocular camera 5, the ToF sensor 4, and the infrared fill light 3 are all connected to the signal processing module (STM32H743) of the perception layer. The signal processing module processes the images from the binocular camera 5 using the SGM algorithm to generate a disparity map, and combines it with ICP point cloud registration technology to fuse the depth data from the ToF sensor 4 to construct a 3D environment map. The signal processing module constructs the 3D environment map by registering the SGM algorithm with the ICP point cloud. The hybrid positioning system includes a UWB base station network (Decawave DW1000), an IMU (ADIS16470), and an RFID verification tag (ISO). The UWB base station network and IMU outputs are connected to the fusion processing module (NXP i.MX8MPlus) of the sensing layer, and the RFID verification tag is communicatively connected to the fusion processing module. The fusion processing module uses the ESKF algorithm to dynamically fuse UWB positioning data and IMU inertial data to achieve high-precision positioning. The RFID verification tag is used for position reset calibration. The pressure sensor array (HBM U9B) is deployed at the four corners of the switch trolley 6 carrying platform. The output of each pressure sensor 7 is connected to the signal acquisition module (ADS1256) of the sensing layer to detect the switch loading status and off-center load in real time and trigger alarms. The signal processing module, fusion processing module and signal acquisition module of the sensing layer are all connected to the input of the transmission layer. Data interaction is performed between the transmission layer and the decision layer, and various sensing data are transmitted to the decision layer through the transmission layer. This enables multi-dimensional and high-precision environmental perception and status detection. The 3D environmental map construction can accurately capture information on dynamic and static obstacles. The hybrid positioning system achieves positioning accuracy through data fusion. The pressure sensor array of 7 can monitor loading stability in real time. This solves the problems of single perception method, insufficient data reliability and low positioning accuracy in traditional methods. It provides comprehensive and accurate basic data support for decision-making and ensures the accuracy of subsequent path planning and motion control.

[0022] Furthermore, the transmission layer includes a dual-link module and a multi-protocol processing module. The dual-link module consists of a wireless communication module (Huawei MH5000-315GCPE) and a wired communication module (Moxa EDS-405A switch). The multi-protocol processing module (TIAM62A7) includes a control command protocol unit, a sensor data protocol unit, and a cloud communication protocol unit. The input end of the multi-protocol processing module is electrically connected to the signal output end of the sensing layer, and the output end of the multi-protocol processing module is electrically connected to the signal input end of the decision layer. The multi-protocol processing module is bidirectionally connected to both the wireless communication module and the wired communication module. The wireless communication module and the wired communication module are deployed in parallel to achieve redundancy backup. The control command protocol unit addresses the control signal transmission requirements, the sensor data protocol unit adapts to high-throughput sensor data transmission, and the cloud communication protocol unit is responsible for command interaction and log uploading with the cloud. Through the collaboration of multiple protocols and dual links, a closed-loop link for data transmission between layers is formed. The multi-protocol processing module addresses the transmission requirements of different types of data through corresponding protocol units, forming a closed-loop link for data transmission between layers. The dual-link redundancy design ensures the continuity of data transmission. The multi-protocol processing module enables the efficient transmission of different types of data, such as control commands, sensor data, and cloud interactions, ensuring low latency and high reliability of signal interaction between layers, avoiding transmission interruptions due to data transmission failures, and improving the overall stability of the system.

[0023] Furthermore, the decision-making layer includes a main controller (NVIDIA Jetson AGX Orin) and a real-time controller (Siemens S7-1500), which work together to make decisions and issue commands. The main controller is used for environment modeling and path planning, while the real-time controller is used for real-time issuance of motion control commands. The decision-making layer runs an improved A* path planning algorithm and a DWA local obstacle avoidance strategy. * The algorithm embeds safety distance constraints for power equipment, ensuring a safe distance between the planned path and the equipment through algorithmic logic. The DWA obstacle avoidance strategy optimizes motion parameters through an evaluation function to achieve flexible avoidance of dynamic obstacles. The evaluation function is S=aH+bO+cP, where H is the heading deviation, O is the obstacle distance, P is the path fit, and a, b, and c are weight coefficients, with a+b+c=1. After the main controller completes environmental modeling and global path planning based on the data transmitted from the perception layer, it transmits the planning results to the real-time controller, which then converts them into specific motion control commands and issues them to the execution layer. Improved A * The algorithm ensures the safety and optimality of the path. The DWA algorithm realizes dynamic obstacle avoidance in complex environments. The collaboration between the main controller and the real-time controller improves the efficiency of decision-making and command issuance, ensuring that the system can accurately plan the path and flexibly deal with obstacles in the power environment, thereby improving the safety and adaptability of the transportation process.

[0024] Furthermore, the execution layer includes an omnidirectional drive module and a servo control module (Maxon ESCON50 / 5). The omnidirectional drive module consists of a brushless motor (Maxon EC-4pole30), a reducer (Harmonic Drive CSF-17-50-2A-GR), and multiple independent Mecanum wheels 1. The servo control module includes a current loop control unit, a speed loop control unit, and a position loop control unit. The output shaft of the brushless motor is drivenly connected to the input end of the reducer, and the output end of the reducer is drivenly connected to the shaft of the Mecanum wheel 1. The input end of the servo control module is electrically connected to the output end of the real-time controller of the decision layer, and the output end of the servo control module is electrically connected to the brushless motor. The current loop control unit, speed loop control unit, and position loop control unit are cascaded sequentially to form a three-loop collaborative control link. The servo control module drives the omnidirectional drive module to achieve multi-directional movement through the three-loop collaborative control, thereby driving the omnidirectional drive module to achieve movement.

[0025] Furthermore, the interaction layer includes an identity authentication module, a visualization module, an instruction input module, and a communication interface module (W5500 Ethernet chip). The identity authentication module adopts the GMM-UBM model and includes a voiceprint acquisition unit (XMOSXVF3610 microphone array) and a two-factor authentication unit. The visualization module is a digital twin platform (developed using Unity 3D), and the instruction input module is an industrial touchscreen (Advantech TPC-7150). The output of the voiceprint acquisition unit is connected to the input of the two-factor authentication unit, and the output of the two-factor authentication unit is electrically connected to the decision layer through the communication interface module to achieve secure identity authentication. The industrial touchscreen and the digital twin platform have bidirectional data communication, and the signal terminal of the industrial touchscreen is connected to the perception layer and the decision layer through the communication interface module. The digital twin platform is used for 3D visualization of equipment operating status, path information, etc., and the industrial touchscreen supports natural language instruction input and monitoring data display to achieve human-computer interaction and remote monitoring. The digital twin platform is used for 3D visualization of equipment status, and the industrial touchscreen is used for natural language instruction input and monitoring data display. The GMM-UBM model and two-factor validation enhance the security of operation permissions, the digital twin platform enables visualized monitoring of equipment status, and the industrial touch screen provides a convenient way to input commands, making it easy for operators to keep track of the system's operating status and issue transportation instructions in real time, thus improving operation and maintenance efficiency and ease of operation.

[0026] Furthermore, the fusion processing module of the hybrid positioning system incorporates a dynamic weight adjustment unit and a position reset unit. The dynamic weight adjustment unit is connected to the signal output terminals of the UWB base station network and the IMU, respectively, and adjusts their weights in real time according to signal quality. The position reset unit is connected to the RFID verification tag for coordinate reset when positioning deviation occurs. The fusion processing module achieves optimal estimation through the ESKF algorithm unit, which interacts bidirectionally with the dynamic weight adjustment unit and the position reset unit. The signal processing module of the four-way stereo vision unit incorporates an iterative solution unit, which is connected to the signal output terminals of the binocular camera 5 and the ToF sensor 4. It optimizes the point cloud registration effect through iterative calculations to achieve point cloud registration. The signal processing module communicates with the environment modeling unit of the transmission layer and the decision layer to achieve dynamic updates of the environment model. The system improves positioning accuracy and real-time environmental modeling. The dynamic weight adjustment unit solves the problem of single positioning source being easily interfered with, the position reset unit ensures positioning accuracy, and the iterative solution unit optimizes point cloud registration effect, making 3D environmental maps more accurate. It solves the problems of insufficient positioning stability and visual data processing delay in complex power environments, providing decision-makers with more reliable positioning data and environmental models, and improving the overall adaptability of the system.

[0027] Furthermore, each pressure sensor 7 in the pressure sensor array is electrically connected to the signal input terminal of the signal acquisition module of the sensing layer. The signal acquisition module has a built-in differential pressure determination unit. The differential pressure determination unit calculates the pressure difference at the four corners in real time. When the difference meets the off-center load condition, its output terminal is connected to the motion control command output unit of the decision layer through the transmission layer to trigger the attitude adjustment signal. The output terminal of the differential pressure determination unit is connected to the motion control command output unit of the decision layer through the transmission layer to trigger the attitude adjustment signal. The fusion processing module of the hybrid positioning system has a built-in heading correction model unit. The heading correction model unit is connected to the path planning algorithm unit of the decision layer. It obtains the coordinates of the vehicle's positioning reference point through the positioning reference point detection unit and collects obstacle distance data through the obstacle distance detection unit to construct a heading correction model so that the distance between the positioning reference point and the obstacle is equal to the average value of the preset standard distance range, thereby realizing heading deviation correction. This solves the problems of equipment tilting caused by uneven loading and path deviation caused by heading deviation during transportation. The differential pressure judgment unit can trigger attitude adjustment in real time to prevent the switch from being damaged by uneven loading. The heading correction model unit ensures that the trolley travels along the preset path, improving the stability and safety of the transportation process and ensuring the accuracy of switch transportation.

[0028] Furthermore, the main controller of the decision layer incorporates a crab-walking mode control unit and an obstacle avoidance algorithm unit. The crab-walking mode control unit is connected to a four-wheel synchronous steering control module (TIDRV8301 driver chip). The output of the four-wheel synchronous steering control module is electrically connected to the servo control module of the execution layer via a real-time controller, achieving lateral translation of the vehicle by controlling the four-wheel synchronous steering. The obstacle avoidance algorithm unit runs the DWA algorithm and communicates with the improved A* algorithm of the path planning algorithm unit to achieve coordination between global path and local obstacle avoidance. The decision layer incorporates a heading correction calculation unit, which is connected to the heading correction model unit of the fusion processing module, and uses formulas... or The system calculates and corrects travel time, and its output is connected to the motion control command output unit of the real-time controller to achieve inertial navigation correction. The cost function calculation unit of the improved A* algorithm is connected to the danger zone weight configuration unit, which presets different values ​​according to the equipment type to increase the priority of detouring through danger zones. The crab mode allows the vehicle to move flexibly laterally in narrow passages, the heading correction calculation unit realizes accurate correction of heading deviation, and the danger zone weight configuration ensures safe avoidance of dangerous areas such as power equipment, further improving the system's adaptability to special scenarios and the safety and accuracy of transportation routes.

[0029] formula or middle: The correction time (unit: s) corresponds to the adjustment time for two different heading deviation scenarios, and is used to control the specific time for the car to correct its heading by turning. The initial standard heading angle (unit: rad or °) is the target heading angle preset by the decision layer. It is determined by the path planning algorithm unit based on the global path and serves as the benchmark for heading correction. The current heading angle (unit: rad or °) is obtained in real time by the fusion of IMU and UWB data from the hybrid positioning system, reflecting the actual heading status of the vehicle; w is the heading angular velocity (unit: rad / s or ° / s), which is the angular velocity parameter when the vehicle turns, preset and dynamically adjusted by the servo control module of the execution layer to ensure smooth and accurate heading correction. When the current heading angle... When the car veers to the left, use the formula. Calculate the right turn correction time; when When the car veers to the right, use the formula. The left turn correction time is calculated, and finally, the command is issued through the real-time controller to achieve precise correction of the heading deviation.

[0030] A method for intelligent navigation automated transportation of a 10kV switch trolley includes the following steps: S1. System Initialization: Perception layer calibration and adjustment: The binocular camera 5 adopts the checkerboard calibration method (OpenCV standard procedure) to acquire 20 sets of checkerboard images from different angles, calculate intrinsic and extrinsic parameters, and the reprojection error is <0.5 pixels; the ToF sensor 4 performs zero-point calibration to compensate for background noise; the IMU performs 30-second static calibration to correct the gyroscope zero bias; the pressure sensor 7 performs zero-point calibration to eliminate initial error.

[0031] Transmission layer connection establishment: The 5G module detects signal strength, automatically connects to the base station and establishes a communication link; the industrial Ethernet module starts the STP protocol, generates a ring network topology, and completes the establishment of redundant links; the multi-protocol processing module initializes the DDS / ZeroMQ / MQTT protocol stack and establishes communication interfaces with each layer.

[0032] Decision layer parameter loading: The main controller downloads the switch room point cloud map and equipment safety boundary parameters from the cloud; the real-time controller initializes motion control parameters and sets the motor rated current, speed limit, etc.; the algorithm unit loads the initial parameters (such as weight coefficients, search step size, etc.) of the improved A* algorithm and DWA algorithm.

[0033] S2, Task Triggered: Interactive layer command input and identity authentication: Operators input transportation commands (such as "transport cabinet #3 switch to the maintenance area") or voice commands through the industrial touch screen; the voiceprint acquisition unit collects the operator's voiceprint signal, compares it with the GMM-UBM model, and at the same time reads the operator's RFID badge and generates a dynamic verification code to complete two-factor authentication.

[0034] Sensing layer status detection: The pressure sensor array 7 collects pressure data at the four corners of the carrying platform. If the sum of the pressure at the four points is within ±5% of the rated value and the variance is <10%, it is determined to be a stable load; if it is unloaded, the sum of the pressure at the four points is <10kg and the variance is <5kg; the RFID reader reads the electronic tag on the switch to obtain information such as the switch model and weight; the hybrid positioning system obtains the initial positioning data of the vehicle. If the coordinate error read by the RFID verification tag is >5mm, UWB recalibration is triggered.

[0035] S3, Environmental Perception: 3D environment modeling: The binocular camera 5 of the four-way stereo vision unit acquires environmental images, generates a disparity map through the SGM algorithm, and converts it into a sparse point cloud; the ToF sensor 4 acquires depth data, which is then filtered and denoised to generate a dense point cloud; the signal processing module registers the two types of point clouds through the ICP algorithm, dynamically assigns weights (ToF point cloud weight 0.6, binocular point cloud weight 0.4), and constructs a real-time 3D map.

[0036] Obstacle tracking: Dynamic obstacles are detected using the Lucas-Kanade pyramid optical flow method, and the motion state of obstacles is calculated by ESKF filtering combined with IMU data; static obstacles are constructed by extracting ORB feature points to build a sparse map, and a surface model after voxel filtering is generated by combining dense point cloud. Map reconstruction is triggered when the matching rate between the new keyframe and the historical map is less than 70%.

[0037] High-precision positioning: The UWB base station network of the hybrid positioning system calculates the absolute position of the vehicle, and the IMU collects motion data and calculates the relative displacement and attitude angle through the fourth-order Runge-Kutta method integration; the fusion processing module dynamically fuses UWB and IMU data through the ESKF algorithm. When the vehicle approaches the RFID verification tag, it reads the tag coordinates to reset the position and finally outputs the pose data.

[0038] S4. Path planning and course correction: Global Path Planning: Improvements to the Path Planning Algorithm Unit of the Decision-Making Main Controller (A) * The algorithm, with the cost function as follows: The weight of dangerous areas of high-voltage equipment is set to 10, and that of ordinary obstacles is set to 1. The algorithm embeds safety distance constraints of power equipment and ensures that the distance between the path and the live body is ≥200mm by expanding the radius of the obstacle (200mm). A bidirectional heuristic search is adopted, and the step size in narrow areas is adaptively set to 50mm. The planning time for a 20m×20m scene is <80ms, generating the globally optimal path.

[0039] Local obstacle avoidance adjustment: The DWA obstacle avoidance algorithm unit sets the linear velocity range [0, 1.2] m / s and the angular velocity range [-30°, 30°] / s based on the real-time 3D map and obstacle status; it optimizes the motion parameters through the evaluation function S=0.6H+0.3O+0.1P, where H is the heading deviation, O is the obstacle distance, and P is the path fit, and dynamically adjusts the vehicle's trajectory to achieve local obstacle avoidance.

[0040] Crab mode activated: When the detected channel width is ≥1.5m and <2m, the decision layer automatically activates the crab mode. The crab mode control unit drives the wheels to steer synchronously (steering angle ±45°) through the four-wheel synchronous steering control module to achieve lateral translation (speed 0.5m / s). The wheel speed is controlled by the kinematic model to ensure that the deviation between the vehicle body and the center line of the channel is <2cm.

[0041] Heading Correction: The heading correction calculation unit obtains the current heading angle. Compared with the initial standard heading angle ,when At that time, through the formula Calculate the corrected travel time for a left turn; when At that time, through the formula Calculate the right turn correction travel time and send the correction command to the real-time controller to achieve heading deviation correction (correction error <0.1°).

[0042] S5, Motion Control: Precise motion control: The servo control module of the execution layer receives motion commands from the real-time controller, the current loop control unit stabilizes the motor current, the speed loop control unit adjusts the motor speed, and the position loop control unit precisely controls the displacement of the vehicle; the omnidirectional drive module realizes omnidirectional motion such as forward, backward, lateral translation, and rotation through Mecanum wheel 1, and adjusts the motion posture according to the path planning results.

[0043] Off-center load adjustment: The pressure sensor array 7 monitors the loading status in real time. When the pressure difference between any two points is greater than 20% of the rated value, the pressure difference judgment unit of the signal acquisition module triggers the attitude adjustment signal. The decision layer generates an adjustment command, and the execution layer fine-tunes the trolley attitude through the servo control module until the pressure difference is ≤20%.

[0044] S6, Task Return: Return route planning: After manual or crane loading and unloading is completed, the decision-making level automatically plans the return route, adopts the "storage area priority" strategy, dynamically adjusts the route weight, and prioritizes straight and open routes, with a planning time of <50ms.

[0045] Precise positioning of the storage area: When the vehicle travels along the return path and approaches the storage area, the hybrid positioning system combines the UWB base station network and the RFID tag of the storage area for dual positioning. At the same time, the four-way stereo vision unit identifies the positioning mark of the storage area to achieve precise positioning of the storage area (positioning error <1mm).

[0046] System hibernation preparation: After the vehicle is parked in the designated position, the motors in the execution layer enter the zero-position holding mode; the sensors in the perception layer reduce the sampling frequency and enter a low-power state; the 5G module in the transmission layer switches to a low-power mode, while the Ethernet module maintains the communication connection; the interaction layer displays "Task completed, returned to position", and the system enters the hibernation preparation state, waiting for the next task to be triggered.

[0047] The embodiments described in this specification are merely examples of implementations of the inventive concept. The scope of protection of this invention should not be considered as limited to the specific forms stated in the embodiments. The scope of protection of this invention also extends to equivalent technical means that can be conceived by those skilled in the art based on the inventive concept.

Claims

A 1.10kV switch trolley intelligent navigation automated transportation system, characterized in that, The system comprises a five-layer modular architecture consisting of a perception layer, a transmission layer, a decision-making layer, an execution layer, and an interaction layer. Each layer achieves fully automated closed-loop control through signal interaction. The perception layer and the decision-making layer establish a bidirectional data connection through the transmission layer, and the decision-making layer and the execution layer are connected through control signal lines. The interaction layer establishes communication connections with both the perception layer and the decision-making layer. The perception layer is used for high-precision environmental modeling and positioning. The decision-making layer completes dynamic path planning and heading correction based on the data collected by the perception layer. The execution layer is located in the switch trolley and moves according to the instructions of the decision-making layer. The transmission layer is used for data transmission between the layers, and the interaction layer provides human-machine interaction and remote monitoring.

2. The 10kV switch trolley intelligent navigation automated transportation system according to claim 1, characterized in that, The perception layer includes a four-way stereo vision unit, a hybrid positioning system, and a pressure sensor array. The four-way stereo vision unit, deployed at the four corners of the switch trolley, includes a binocular camera, a ToF sensor, and an infrared fill light. The outputs of the binocular camera, ToF sensor, and infrared fill light are all connected to the signal processing module of the perception layer. The signal processing module constructs a 3D environmental map through SGM algorithm and ICP point cloud registration. The hybrid positioning system includes a UWB base station network, an IMU, and an RFID verification tag. The outputs of the UWB base station network and IMU are connected to the fusion processing module of the perception layer. The RFID verification tag communicates with the fusion processing module, which achieves positioning through the ESKF algorithm. The pressure sensor array is deployed at the four corners of the switch trolley's carrying platform. The output of each pressure sensor is connected to the signal acquisition module of the perception layer for detecting the switch loading status and off-center load alarms. The signal processing module, fusion processing module, and signal acquisition module of the perception layer are all connected to the input of the transmission layer, enabling data interaction with the decision layer through the transmission layer.

3. The 10kV switch trolley intelligent navigation automated transportation system according to claim 1, characterized in that, The transmission layer includes a dual-link module and a multi-protocol processing module. The dual-link module consists of a wireless communication module and a wired communication module. The multi-protocol processing module includes a control command protocol unit, a sensor data protocol unit, and a cloud communication protocol unit. The input terminal of the multi-protocol processing module is electrically connected to the signal output terminal of the sensing layer, and the output terminal of the multi-protocol processing module is electrically connected to the signal input terminal of the decision layer. The multi-protocol processing module is bidirectionally connected to both the wireless communication module and the wired communication module. The wireless communication module and the wired communication module are deployed in parallel to achieve redundancy backup. The multi-protocol processing module connects to the transmission requirements of different types of data through corresponding protocol units, forming a closed-loop link for data transmission between layers.

4. The 10kV switch trolley intelligent navigation automated transportation system according to claim 1, characterized in that, The decision-making layer includes a main controller and a real-time controller. The main controller is used for environment modeling and path planning, while the real-time controller is used for real-time issuance of motion control commands. The decision-making layer runs an improved A... * Path planning algorithm and DWA local obstacle avoidance strategy, the improved A * The algorithm incorporates safety distance constraints for power equipment. The DWA obstacle avoidance strategy optimizes motion parameters through an evaluation function, which is S=aH+bO+cP, where H is the heading deviation, O is the obstacle distance, P is the path fit, and a, b, and c are weight coefficients with a+b+c=1.

5. The 10kV switch trolley intelligent navigation automated transportation system according to claim 1, characterized in that, The execution layer includes an omnidirectional drive module and a servo control module. The omnidirectional drive module consists of a brushless motor, a reducer, and multiple independent Mecanum wheels. The servo control module includes a current loop control unit, a speed loop control unit, and a position loop control unit. The output shaft of the brushless motor is drivenly connected to the input end of the reducer, and the output end of the reducer is drivenly connected to the shaft of the Mecanum wheels. The input end of the servo control module is electrically connected to the output end of the real-time controller of the decision layer, and the output end of the servo control module is electrically connected to the control end of the brushless motor. The current loop control unit, speed loop control unit, and position loop control unit are cascaded sequentially to form a three-loop collaborative control link, driving the omnidirectional drive module to achieve motion.

6. The 10kV switch trolley intelligent navigation automated transportation system according to claim 1, characterized in that, The interaction layer includes an identity authentication module, a visualization module, an instruction input module, and a communication interface module. The identity authentication module adopts the GMM-UBM model and includes a voiceprint acquisition unit and a two-factor authentication unit. The visualization module is a digital twin platform, and the instruction input module is an industrial touchscreen. The output of the voiceprint acquisition unit is connected to the input of the two-factor authentication unit, and the output of the two-factor authentication unit is electrically connected to the decision layer through the communication interface module. The industrial touchscreen and the digital twin platform have bidirectional data communication, and the signal terminal of the industrial touchscreen is connected to the perception layer and the decision layer through the communication interface module. The digital twin platform is used for 3D visualization of equipment status, and the industrial touchscreen is used for natural language instruction input and monitoring data display.

7. The 10kV switch trolley intelligent navigation automated transportation system according to claim 2, characterized in that, The fusion processing module of the hybrid positioning system incorporates a dynamic weight adjustment unit and a position reset unit. The dynamic weight adjustment unit is connected to the signal output terminals of the UWB base station network and the IMU, respectively, while the position reset unit is communicatively connected to the RFID verification tag. The fusion processing module achieves optimal estimation through the ESKF algorithm unit, which interacts bidirectionally with the dynamic weight adjustment unit and the position reset unit. The signal processing module of the four-way stereo vision unit incorporates an iterative solution unit, which is connected to the signal output terminals of the binocular camera and the ToF sensor to achieve point cloud registration. The signal processing module communicates with the environmental modeling unit of the decision layer through the transmission layer.

8. The 10kV switch trolley intelligent navigation automated transportation system according to claim 2, characterized in that, Each pressure sensor in the pressure sensor array is electrically connected to the signal input terminal of the signal acquisition module in the perception layer. The signal acquisition module has a built-in differential pressure determination unit. The output terminal of the differential pressure determination unit is connected to the motion control command output unit of the decision layer through the transmission layer, and is used to trigger attitude adjustment signals. The fusion processing module of the hybrid positioning system has a built-in heading correction model unit, which is connected to the path planning algorithm unit of the decision layer. The heading correction model unit obtains data through the positioning reference point detection unit and the obstacle distance detection unit, so that the distance between the positioning reference point and the obstacle is equal to the average value of the preset standard distance range.

9. The 10kV switch trolley intelligent navigation automated transportation system according to claim 4, characterized in that, The main controller of the decision layer incorporates a crab-walking mode control unit and an obstacle avoidance algorithm unit; the crab-walking mode control unit is connected to the four-wheel synchronous steering control module, and the output of the four-wheel synchronous steering control module is electrically connected to the servo control module of the execution layer through a real-time controller; the obstacle avoidance algorithm unit runs the DWA algorithm, and is an improved version of the path planning algorithm unit. * Algorithm data interoperability; the decision layer has a built-in heading correction calculation unit, which is connected to the heading correction model unit of the fusion processing module. The heading correction calculation unit uses formulas... or The corrected travel time is calculated, and its output is connected to the motion control command output unit of the real-time controller; the cost function calculation unit of the improved A* algorithm is connected to the dangerous area weight configuration unit, which presets different values ​​according to the equipment type.

10. A method for using the 10kV switch trolley intelligent navigation automated transportation system according to any one of claims 1 to 9, comprising the following steps: S1: System initialization, the perception layer completes device calibration and adjustment, the transmission layer establishes a dual-link connection, and the decision layer loads the environment map and security boundary parameters; S2: Task triggered, inputting transportation instructions and completing identity authentication through the interaction layer, and detecting the switch loading status and initial positioning through the perception layer; S3: Environmental perception, four-way stereo vision unit builds real-time 3D map and tracks obstacles, hybrid positioning system outputs high-precision pose data; S4: Path planning and heading correction. The decision-making layer generates a global path by improving the A* algorithm, dynamically adjusts the local path by combining the DWA algorithm, corrects driving deviations by the heading correction model, activates "crab mode" in narrow passages, and meets safety distance constraints during the path planning process. S5: Motion control. The execution layer achieves precise motion based on decision commands, and the pressure sensor array monitors the loading status in real time, triggering adjustments when abnormalities occur. S6: Task completed. After completion, the system automatically plans the return route and uses a hybrid positioning system to achieve precise positioning of the storage area. The device then enters a hibernation preparation state.

Images