Crane automatic obstacle avoidance path planning system based on machine vision and frequency converter

Abstract

The application discloses a tower crane automatic obstacle avoidance path planning system based on machine vision and a frequency converter, belongs to the field of construction engineering machinery, and comprises a perception module, a path planning module, a motion control module, a safety protection module, a communication module and an intelligent load wind deflection prediction and trajectory compensation module; the perception module is used for collecting environment and tower crane posture data, realizing obstacle identification, three-dimensional modeling and dynamic electronic fence; the path planning module is used for executing dynamic obstacle avoidance path planning and hazard level determination according to perception information and constraint conditions; and the motion control module is used for realizing precise speed regulation, sensorless anti-sway and zero-speed hovering through a PLC and a vector closed-loop frequency converter. The application realizes full-automatic, high-precision, high-reliability and high-safety operation of a tower crane, solves collision risks in group tower operation, wind disturbance environment and complex construction site scenes, is suitable for intelligent reconstruction of new tower cranes and in-service tower cranes, and has wide engineering application value.

Description

Technical Field

[0001] This invention belongs to the field of construction machinery, specifically, it relates to an automatic obstacle avoidance path planning system for tower cranes based on machine vision and frequency converters. Background Technology

[0002] Tower cranes are core lifting equipment in building construction and are widely used in construction scenarios such as high-rise residential buildings, rail transit hubs, and industrial plants.

[0003] Traditional tower cranes rely on manual operation and limited perception methods, depending on the operator's visual judgment. Obstacle recognition accuracy is low at night, in foggy conditions, and in dusty environments. They lack 3D spatial modeling and dynamic electronic fences, making them prone to jib interference and collisions with personnel and buildings during multi-tower operations. Motion control uses analog terminal speed regulation, resulting in low accuracy, poor anti-interference capabilities, and significant load swaying without effective anti-sway measures. Safety protection mechanisms are simplistic, lacking tiered risk avoidance and failure protection; brake failures can easily lead to hook slippage and falling objects. Communication links are limited, and obstructions and interference at the construction site can easily cause signal interruptions and control failures. Under wind disturbances, load swaying and drift cannot be predicted or compensated for, significantly reducing lifting accuracy and safety. Intelligent retrofitting of in-service tower cranes is costly, has poor compatibility, and lacks unified data management and health assessment methods. In view of these problems, this paper proposes an automatic obstacle avoidance path planning system for tower cranes based on machine vision and frequency converters. Summary of the Invention

[0004] To address the aforementioned problems and technical deficiencies, this invention adopts the following technical solution: a tower crane automatic obstacle avoidance path planning system based on machine vision and frequency converters, comprising a perception module, a path planning module, a motion control module, a safety protection module, a communication module, and an intelligent load wind deflection prediction and trajectory compensation module.

[0005] The sensing module is used to collect environmental data, tower crane attitude data, positioning data and wind field data at the construction site. It uses deep learning algorithms to classify and identify obstacles, integrates multi-source data to construct a three-dimensional spatial model and a dynamic electronic fence, and outputs information on obstacle location, hazard level and intrusion trend.

[0006] The path planning module is used to perform dynamic obstacle avoidance path planning based on perception information and tower crane operation constraints, trigger corresponding actions according to multi-level distance thresholds, support multiple automatic operation modes, and adapt to group tower crane collaboration and fixed working conditions.

[0007] The motion control module is based on a PLC and controls a vector closed-loop frequency converter through digital communication to achieve continuous and precise speed regulation of multiple mechanisms of the tower crane. It adopts an input shaping algorithm to achieve sensorless anti-swaying of heavy loads and supports zero-speed hovering of heavy objects.

[0008] The safety protection module is used to execute a multi-level linkage safety obstacle avoidance mechanism, and is equipped with multiple protections such as soft limit, anti-slip hook, anti-overrun, and anti-overload, to achieve all-dimensional safety protection.

[0009] The communication module adopts a primary and backup dual-link architecture. The primary link is responsible for high-speed data transmission, while the backup link has strong diffraction and anti-blocking capabilities, supports seamless link switching and data encryption verification, and ensures reliable communication.

[0010] The intelligent load wind deflection prediction and trajectory compensation module has a built-in load feature database. It integrates load attitude and wind field data to estimate aerodynamic loads, generates trajectory compensation commands, and realizes adaptive correction of hoisting trajectory under wind disturbance.

[0011] Preferably, the sensing module includes a multi-channel visual acquisition unit, a lidar ranging unit, a centimeter-level positioning unit, an attitude sensing unit, and a wind field monitoring unit; the hook visual acquisition unit integrates automatic target tracking, automatic focusing, and wireless charging functions.

[0012] Preferably, the perception module classifies and identifies buildings, scaffolding, edge openings, tower crane booms, personnel, and vehicles at the construction site based on deep learning algorithms, establishes a three-dimensional coordinate system with the tower crane's rotation center as the origin, and updates the three-dimensional electronic fence in real time.

[0013] Preferably, the path planning module triggers full-speed operation, automatic deceleration, path replanning, parking, and emergency braking zero-speed hovering actions in stages according to four threshold levels: safe distance, warning distance, danger distance, and critical distance. It plans the path by comprehensively considering torque, wind speed, angle, amplitude, height, and the constraints of the tower interference zone.

[0014] Preferably, the path planning module supports fixed-point positioning, fixed-distance operation, one-click return, and multi-point sequential operation modes, enabling automatic hoisting, continuous operation, and multi-point cyclic execution.

[0015] Preferably, the motion control module uses digital communication to control the frequency converter to achieve vector closed-loop continuous speed regulation; the PLC uses an input shaping algorithm to calculate the anti-sway parameters in real time to achieve sensorless anti-sway under load.

[0016] Preferably, the safety protection module includes a five-level safety obstacle avoidance mechanism, namely, audible and visual warning and slight deceleration, speed limit and safety distance tightening, automatic detour trajectory generation, stop mechanism braking, brake failure protection and zero-speed hovering of heavy objects; it also includes soft limit virtual protection and anti-hook slippage, anti-overrun, and anti-overload protection.

[0017] Preferably, the communication module uses high-speed mobile communication as the main link and long-distance radio as the backup link, integrating data encryption, verification and packet loss retransmission mechanisms, and the backup link seamlessly takes over control when the main link is interrupted.

[0018] Preferably, the system also includes a cloud data management unit for uploading operational data, obstacle avoidance events, and operation videos in real time, supporting historical data tracing, tower crane health assessment, and intelligent operation and maintenance; the system is compatible with newly built tower cranes and the retrofitting of existing tower cranes, and supports multiple communication deployment methods.

[0019] Preferably, the intelligent load wind deviation prediction and trajectory compensation module has a built-in load feature database, integrates load three-dimensional attitude and wind field data to estimate aerodynamic load in real time, predicts drift and rotation deviation, and synchronously corrects path and anti-sway parameters.

[0020] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention achieves closed-loop intelligent control of the entire tower crane operation process through the collaborative operation of six modules: multi-source sensing fusion, intelligent path planning, high-precision frequency conversion control, multi-level safety protection, dual-link redundant communication, and intelligent load wind deflection compensation. It can accurately identify obstacles on the construction site, dynamically plan the optimal obstacle avoidance path, smoothly control the tower crane operation, trigger safety avoidance actions in stages, ensure uninterrupted communication, and adaptively correct the hoisting trajectory under wind disturbance. It effectively solves the problems of traditional tower cranes, such as single sensing, delayed obstacle avoidance, coarse control, insufficient safety protection, unreliable communication, and large impact of wind deflection. It significantly improves the safety, accuracy, and stability of tower crane operations, reduces the intensity of manual operation and the risk of safety accidents, and is compatible with the intelligent transformation of newly built tower cranes and existing tower cranes. It supports cloud data management and intelligent operation and maintenance and has good engineering applicability. Attached Figure Description

[0021] In the attached diagram: Figure 1 This is a diagram of the overall system architecture of the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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 some embodiments of the present invention, but not all embodiments. Generally, the components of the embodiments of the present invention described and shown in the accompanying drawings can be arranged and designed in various different configurations.

[0023] Example: Refer to Figure 1 The tower crane automatic obstacle avoidance path planning system based on machine vision and frequency converter includes a perception module, a path planning module, a motion control module, a safety protection module, a communication module, and an intelligent load wind deviation prediction and trajectory compensation module. The perception module collects real-time construction site environmental data, tower crane real-time attitude data, obstacle spatial coordinate data and wind field environmental data through various sensors and visual acquisition devices. After data preprocessing and feature extraction, it completes obstacle identification, 3D spatial modeling, dynamic electronic fence construction and dangerous area marking. The processed effective data is output to the path planning module and intelligent load wind deviation prediction and trajectory compensation module in real time. The path planning module receives various data output by the sensing module, combines the tower crane's own operational constraints (torque, wind speed, limit switches, etc.) and obstacle hazard levels, executes real-time path optimization and dynamic obstacle avoidance decisions through preset algorithms, triggers corresponding action commands according to hazard threshold levels, and simultaneously sends path commands to the motion control module. The motion control module uses a PLC as the control core and establishes linkage with the vector closed-loop frequency converter through digital communication. It receives action commands from the path planning module, calculates the operating parameters of the three major mechanisms of the tower crane: hoisting, luffing, and slewing, and controls the frequency converter to output the corresponding frequency to achieve precise speed regulation of multiple mechanisms. At the same time, it uses a dedicated algorithm to achieve sensorless anti-sway control of the suspended load and zero-speed hovering control of the heavy object, ensuring smooth operation. The safety protection module receives status data from the sensing module, path planning module, and motion control module in real time, monitors the tower crane's operating status and working environment in real time, and automatically executes multi-level linkage safety avoidance, forced speed limit, and fault protection actions based on the hazard level and fault status to prevent safety accidents from occurring. The communication module establishes a primary and backup dual-link communication channel. The primary link is responsible for the high-speed transmission of daily sensing data, control commands, and status information. The backup link is in a standby state in real time and can seamlessly take over the communication task when the primary link is interrupted, ensuring highly reliable transmission of various data and commands and preventing control failures caused by transmission interruptions. The intelligent load wind deflection prediction and trajectory compensation module has a built-in load feature database. It receives load three-dimensional attitude and wind field data collected by the sensing module, calculates wind load and load drift based on load aerodynamic characteristics, generates trajectory compensation commands and sends them to the path planning module and motion control module simultaneously, dynamically corrects the tower crane's operating trajectory, realizes adaptive trajectory correction in wind disturbance environment, and avoids collision risks caused by wind deflection.

[0024] This system adopts a centralized hardware topology, with a PLC as the main controller. It connects to the frequency converter, sensing module, safety protection module, and communication module via RS485 bus and Ethernet respectively. The system uses a three-phase five-wire power supply, and the sensors and actuators employ isolated signal wiring to resist electromagnetic interference at the construction site. It includes a multi-channel vision acquisition unit, LiDAR, and RTK. All GNSS units use an integrated PoE power supply and data transmission interface to access the system, simplifying the wiring structure and facilitating the retrofitting and installation of in-service tower cranes.

[0025] In summary, this system, through the collaborative design of six major modules, achieves a closed-loop process for tower crane obstacle avoidance, control, protection, and communication. It solves the problems of traditional tower cranes, such as reliance on manual operation, low obstacle avoidance accuracy, significant wind deflection impact, unreliable communication, and limited safety protection. The six modules have clear division of labor and deep linkage, which not only ensures the real-time and accurate nature of obstacle avoidance decisions but also improves the safety and stability of tower crane operations, while taking into account operational efficiency and adapting to various complex construction sites.

[0026] The perception module includes a multi-channel visual acquisition unit, a lidar ranging unit, and an RTK (Real-Time Kinematics) sensor. The GNSS centimeter-level positioning unit, tower crane attitude sensing unit, and wind field monitoring unit work together as follows: The multi-channel vision acquisition unit includes high-definition image acquisition equipment arranged at the front end of the tower crane boom, the top of the tower, and the hook position. The hook vision acquisition unit integrates AI target automatic tracking, automatic focusing, night vision supplementary lighting, and wireless automatic charging functions. During operation, it acquires real-time image information of the construction site. The AI ​​target automatic tracking function can lock the hook and obstacles, automatic focusing ensures image clarity, night vision supplementary lighting function is suitable for nighttime operation scenarios, and wireless automatic charging function ensures continuous operation of the equipment. The acquired image data is transmitted to the data processing unit of the perception module in real time. The lidar ranging unit is deployed at the front end of the tower crane boom. It scans the working area by emitting laser pulses, receives reflected signals, calculates the distance, angle and height data between obstacles and various parts of the tower crane, and generates three-dimensional point cloud data to make up for the shortcomings of visual acquisition in complex environments (fog, dust). RTK The GNSS centimeter-level positioning unit is installed on the top of the tower crane. It receives satellite signals and combines them with reference station data to achieve centimeter-level positioning of the tower crane itself. At the same time, it helps to calibrate the spatial coordinates of obstacles to ensure that the positioning accuracy meets the obstacle avoidance requirements. The tower crane attitude sensing unit includes a tower crane angle encoder, a body tilt sensor, a height sensor, and a luffing sensor. The angle encoder collects the tower crane's slewing angle, the body tilt sensor collects the tower's tilt angle, the height sensor collects the hook lifting height, and the luffing sensor collects the luffing radius. Each sensor collects data in real time and performs analog-to-digital conversion to output standardized tower crane attitude data. The wind field monitoring unit includes a wind speed sensor and a wind direction sensor, which are deployed at an unobstructed location on the top of the tower crane to collect wind speed and wind direction data in the working area in real time. The sampling frequency is no less than 10Hz to ensure the real-time nature of the wind field data and provide data support for wind deflection compensation. The data processing unit of the perception module performs synchronous alignment, noise reduction filtering, and feature extraction on the heterogeneous data collected by each unit, removes invalid data, integrates it into standardized data, and outputs it to subsequent modules.

[0027] In summary, this perception module adopts a multi-unit collaborative perception design, solving the problems of traditional tower cranes' single perception, low data accuracy, and limited adaptability to various scenarios; the fusion of vision and LiDAR acquisition overcomes the shortcomings of single vision acquisition in inaccurate identification in complex environments (fog, dust, night), achieving accurate obstacle identification in all scenarios; RTK The combination of GNSS centimeter-level positioning and multi-attitude sensors ensures the positioning accuracy of the tower crane's own attitude and obstacle coordinates, solving the problems of large positioning errors and delayed obstacle avoidance decisions in traditional systems. The high-frequency sampling design of the wind field monitoring unit provides accurate data support for wind deflection compensation, solving the problem of untimely trajectory correction caused by lag in wind field data. The data from each unit undergoes unified preprocessing to ensure data standardization and effectiveness, avoiding invalid data from interfering with subsequent decisions and improving the decision-making efficiency and accuracy of the entire system. At the same time, the integrated function of the hook vision unit ensures continuous and stable operation of the equipment, adapting to all-weather operation scenarios.

[0028] The perception module uses deep learning object detection algorithms to classify, identify, and track the location of buildings, scaffolding, openings, tower cranes, workers, and engineering vehicles at the construction site. The specific working process is as follows: First, the image data collected by the multi-channel vision acquisition unit is input into the deep learning target detection algorithm (preferably YOLOv8 algorithm). The algorithm classifies and identifies targets in the image through a pre-trained model, accurately distinguishing different types of obstacles such as buildings, scaffolding, edge openings, multi-tower booms, workers, and engineering vehicles, and outputs the two-dimensional coordinate information of each obstacle. Secondly, it integrates the 3D point cloud data generated by the lidar ranging unit with RTK. The positioning data output by the GNSS centimeter-level positioning unit and the attitude data such as the slewing angle and luffing radius collected by the tower crane attitude sensing unit are used to perform coordinate transformation and establish a three-dimensional polar coordinate space model with the tower crane's slewing center as the origin. Then, based on the three-dimensional polar coordinate space model, the spatial coordinates, volume and relative position of each obstacle with respect to the tower crane are calibrated in real time, the three-dimensional electronic fence of the construction site is dynamically updated, the safety threshold of different types of obstacles is set, and the danger level (safe, warning, danger, critical) is determined according to the distance between the obstacle and the tower crane and the relative movement trend, and the intrusion trend of the obstacle is calculated. Finally, information such as the distance, angle, height, volume, hazard level, and intrusion trend of obstacles is simultaneously output to the path planning module, providing accurate data support for path planning and obstacle avoidance decisions, and ensuring the timeliness and accuracy of obstacle avoidance actions.

[0029] The deep learning algorithm takes multiple real-time image data as input. After target detection, classification, confidence determination, and spatial positioning, it outputs the target category ID, confidence level, two-dimensional pixel coordinates, three-dimensional actual distance, and hazard level. The output adopts a standard structure format, which can be directly parsed by the PLC and path planning module without the need for format conversion, thus improving the system response speed.

[0030] In summary, the YOLOv8 deep learning algorithm addresses the problems of low accuracy, inability to classify and identify obstacles, and slow recognition speed in traditional obstacle recognition, enabling rapid and accurate classification of various obstacles, especially high-risk targets such as workers, thus reducing the risk of collisions. The multi-source data fusion and the construction of a 3D polar coordinate space model solve the problems of traditional 2D recognition's inability to accurately locate obstacle spatial positions and its tendency to cause obstacle avoidance deviations, achieving precise spatial positioning of obstacles. The design of dynamic electronic fences and hazard level determination solves the problems of traditional obstacle avoidance lacking clear hazard classification and untimely warnings, allowing for early warning and obstacle avoidance actions triggered in advance based on obstacle intrusion trends, improving the initiative and timeliness of obstacle avoidance. The complete output of various data provides comprehensive and accurate data support for the path planning module, avoiding obstacle avoidance decision errors due to data gaps, and further improving the reliability and safety of the system's obstacle avoidance.

[0031] The path planning module presets four threshold levels: safe distance, warning distance, danger distance, and critical distance. These thresholds trigger corresponding actions such as full-speed operation, automatic deceleration, path replanning, stopping, and emergency braking to zero-speed hovering. The specific operating mode is as follows: First, the path planning module pre-enters the tower crane's rated parameters, operating range, torque limit, wind speed limit, slewing angle limit, luffing range limit, height limit, and interference zone of multiple towers, and also presets four levels of distance thresholds to clarify the trigger actions corresponding to each threshold. During operation, the path planning module receives obstacle information, tower crane attitude data, and wind field data output by the perception module in real time. Combined with preset constraints, it uses A-Star or RRT optimization algorithms to solve for the collision-free optimal path and plans the optimal hoisting path from the current position to the target position. During the path planning process, it avoids obstacles and dangerous areas in real time. When the tower crane is within a safe distance from an obstacle, full-speed operation is triggered, and the tower crane operates normally according to the planned path. When the distance reaches the warning distance, automatic deceleration is triggered, reducing the tower crane's operating speed and issuing a warning signal to remind the operator to pay attention. When the distance reaches the danger distance, path replanning is immediately triggered, and the algorithm quickly re-solves the optimal path to guide the tower crane around the obstacle. When the distance reaches the critical distance, a stop action is triggered. If there is a risk of collision, emergency braking is immediately executed to control the tower crane to hover at zero speed to avoid a collision accident. The entire path planning process is updated dynamically in real time. If the sensing module detects changes in the position of obstacles, sudden changes in the wind field, or abnormal tower crane posture, the path planning module immediately recalculates the path to ensure that the path is always in a safe and optimal state.

[0032] The path planning algorithm takes the current 3D coordinates, target 3D coordinates, obstacle point cloud data, and tower crane motion constraints as input. After A-Star / RRT path search, collision detection, trajectory smoothing, and graded threshold judgment, it outputs a 3D trajectory point sequence, target velocity, acceleration, deceleration, and braking point. The output binds the trajectory points with anti-sway parameters and sends them down, realizing the integrated collaboration of path planning and motion control.

[0033] In summary, the design of the four-level distance threshold and graded actions solves the problems of traditional path planning, such as the lack of clear hazard classification, single obstacle avoidance actions, and the tendency to over-braking or delayed braking. It achieves graded control of obstacle avoidance actions, balancing operational efficiency and safety. The application of A-Star or RRT optimization algorithms solves the problems of low efficiency and inability to quickly solve for the optimal path in traditional path planning, ensuring the real-time and optimal nature of path planning and adapting to dynamically changing construction sites. Path planning combined with various constraints of the tower crane solves the problems of traditional path planning ignoring the tower crane's own performance limitations and being prone to safety hazards such as overload and exceeding limits. The design of real-time dynamic path updates solves the problem of path failure caused by dynamic changes in obstacles, wind fields, and tower crane attitude at the construction site, ensuring that the path is always safe and adaptable, effectively avoiding collision accidents caused by fixed paths, and improving the flexibility and safety of tower crane operations.

[0034] The path planning module supports four operation modes: fixed-point positioning, fixed-distance operation, one-click call to return to position, and multi-point sequence operation. The specific working methods of each mode are as follows: Fixed-point positioning operation mode: The user presets the three-dimensional coordinates of the hoisting target point through the human-machine interface. After receiving the target point instruction, the path planning module combines the real-time data of the sensing module to plan the optimal path from the current position to the target point, and controls the tower crane to automatically run to the target point. After completing the hoisting, it automatically stops. It is suitable for hoisting operations at a single fixed point. Fixed distance operation mode: The user presets the tower crane's operating distance (luffing distance, slewing angle, or lifting height). The path planning module plans the operating path based on the preset distance and constraints, controlling the tower crane to operate precisely at the preset distance. It automatically stops after reaching the target distance, which is suitable for short-distance, precise lifting scenarios. One-click call to return mode: When the tower crane completes the hoisting operation or an abnormal situation occurs, the user triggers the one-click call to return mode. The path planning module immediately plans the path from the current position to the preset return point (such as the initial position of the tower crane or the safe parking position), controls the tower crane to return automatically, and automatically avoids obstacles during the return process, improving work efficiency and safety. Multi-point sequence operation mode: Users preset the sequence and parameters of multiple hoisting target points. The path planning module plans the optimal path between each target point in the preset order, controlling the tower crane to achieve automatic hoisting, continuous operation and multi-point cyclic execution without frequent manual intervention. It is suitable for the needs of group tower collaborative operation, standard floor repetitive operation and fixed working conditions, and greatly improves operation efficiency.

[0035] In summary, the multi-operation mode design solves the problems of traditional tower cranes, such as single operation mode, limited adaptability to various scenarios, excessive manual intervention, and low operation efficiency. The fixed-point positioning and fixed-distance operation modes achieve precise lifting, solving the problems of large deviations and easy material damage or collisions in traditional manual lifting operations, and are suitable for high-precision operation scenarios. The one-click call-to-return mode solves the problems of cumbersome return to position after tower crane operation and the risk of collisions during the return process, improving operational convenience and safety. The multi-point sequential operation mode achieves continuous automatic operation, solving the problems of frequent manual operation, low operation efficiency, and poor adaptability to multi-tower collaboration and repetitive operation scenarios in traditional tower cranes, significantly reducing the intensity of manual operation, improving operation efficiency, and adapting to the diverse operational needs of different construction sites, enhancing the system's versatility and practicality.

[0036] The motion control module directly controls the frequency converter using RS485 or Ethernet digital communication, replacing traditional analog terminal speed regulation. This achieves continuously adjustable output frequency and high-precision vector closed-loop operation. The specific working process is as follows: The motion control module uses a PLC as the core control unit. The PLC receives path instructions and action instructions from the path planning module in real time. Combined with the tower crane posture data and load data output by the sensing module, it calculates the operating speed curves, acceleration and deceleration rates and braking points of the three major mechanisms of the tower crane: hoisting, luffing and slewing, in real time to ensure that the mechanism operates smoothly and accurately. The PLC establishes bidirectional communication with the vector closed-loop frequency converter via RS485 or Ethernet digital communication, and sends the calculated operating parameters (frequency, speed, etc.) to the frequency converter in the form of digital signals, replacing the traditional analog signal speed regulation of analog terminals, avoiding interference in the analog signal transmission process, realizing continuous adjustment of the frequency converter output frequency with an accuracy of up to 0.1Hz, and meeting the speed requirements of different operating scenarios. After receiving the control signal from the PLC, the vector closed-loop frequency converter adjusts the output voltage and current in real time to drive the motors of each mechanism of the tower crane. At the same time, the encoder built into the frequency converter feeds back the motor speed and position information to form a vector closed-loop control, ensuring the accuracy and stability of motor operation. To address the issue of load swaying, the PLC employs a ZV / ZVD input shaping algorithm. Based on the load mass, lifting height, and operating speed, it calculates anti-sway control parameters in real time and sends them to the frequency converter. By adjusting the motor's operating speed and acceleration, the load swaying is counteracted, achieving sensorless anti-sway control. Simultaneously, it can control the load to achieve zero-speed hovering according to operational requirements, ensuring a safe and stable lifting process. The algorithm (ZV / ZVD input shaping) takes the load mass m, lifting height L, operating speed v, and gravitational acceleration g as inputs. It first calculates the load swaying period T = 2π√(L / g), then calculates the optimal acceleration and deceleration time sequence based on the period to counteract residual swaying. Finally, it outputs the frequency converter's operating frequency, acceleration time, deceleration time, and zero-speed hovering command. The motion control module transmits the speed, current, torque, and load status back to the safety protection module in real time, forming a full-state feedback closed loop.

[0037] In summary, the adoption of digital communication to control the frequency converter, replacing traditional analog terminal speed regulation, solves the problems of analog signal transmission being susceptible to interference, low speed regulation accuracy, and discontinuous frequency adjustment, achieving high-precision, continuously adjustable speed control to adapt to the speed requirements of different operating scenarios. The vector closed-loop control design solves the problems of low operating accuracy, poor stability, and easy speed deviation of traditional motors, ensuring the smooth and precise operation of all mechanisms of the tower crane. The application of the ZV / ZVD input shaping algorithm solves the problems of large swaying of traditional tower cranes when lifting loads, requiring manual intervention to prevent swaying, and the easy occurrence of material falling or collisions, achieving sensorless precise anti-swaying and reducing the intensity of manual operation. The zero-speed hovering function design solves the problems of poor stability and easy swaying when the tower crane needs to temporarily stop during hoisting, improving the safety and convenience of hoisting. At the same time, the overall design improves the response speed and accuracy of motion control, ensuring the accurate implementation of path planning commands.

[0038] The safety protection module includes a five-level obstacle avoidance mechanism, namely: Level 1: audible and visual warning and slight deceleration; Level 2: speed limit and safety distance tightening; Level 3: automatic detour trajectory generation; Level 4: braking; and Level 5: brake failure protection and zero-speed hovering of heavy objects. The specific working methods of each mechanism and additional protection are as follows: Level 1 audible and visual warning and slight deceleration: When the sensing module detects that the distance between the tower crane and an obstacle has reached the warning distance, or when parameters such as wind speed and torque are close to the limit, the safety protection module immediately triggers the audible and visual warning and sends a slight deceleration command to the motion control module to reduce the tower crane's operating speed and remind the operator to pay attention. At this time, the tower crane can still operate normally, only the warning and deceleration are performed. Level 2 speed limit and safety distance tightening: When an obstacle enters the warning area, or the parameters approach the limit, the safety protection module triggers level 2 protection, further reducing the tower crane's operating speed (to below 50% of the rated speed), while tightening the safety distance threshold to improve obstacle avoidance safety and prevent risk escalation; Level 3 Automatic Detour Trajectory Generation: When an obstacle continues to approach and the distance reaches a dangerous distance, and the path planning module fails to complete path replanning in time, the safety protection module actively triggers Level 3 protection, which in turn triggers the path planning module to quickly generate an automatic detour trajectory, controlling the tower crane to detour around the obstacle and avoid collision. Level 4 shutdown action: When the distance between the tower crane and an obstacle reaches the critical distance, or when a minor fault occurs, the safety protection module immediately triggers the shutdown action command, controlling all mechanisms of the tower crane to stop operating, ensuring that the tower crane maintains a safe distance from the obstacle. Operation can only be resumed after the fault is eliminated or the obstacle is removed. Five-level brake failure protection and zero-speed suspension of heavy objects: When a serious fault such as brake failure or motor failure is detected in the tower machine, or when there is an immediate risk of collision, the safety protection module triggers five-level emergency protection, starts the backup braking device, and controls the frequency converter to output the corresponding signal to achieve zero-speed suspension of heavy objects, preventing heavy objects from falling or colliding, and minimizing the loss of safety accidents. In addition, the safety protection module also includes software-defined soft limit virtual protection. By preset the extreme positions of the tower crane's slewing, luffing, and hoisting, the tower crane will automatically decelerate and stop when it approaches the extreme position, replacing some of the functions of traditional hardware limiters. At the same time, it integrates anti-hook slippage protection (immediately brakes when abnormal sliding of the hoisting mechanism is detected), anti-overhead collision protection (stops hoisting when the hook approaches the top limit), and anti-overload protection (in conjunction with torque sensor data, immediately stops operation and alarms when overload occurs), forming multiple safety protections to comprehensively ensure the safety of tower crane operations.

[0039] The safety protection module and motion control module adopt a hard interrupt linkage method. Danger signals can directly trigger the inverter's emergency stop and zero-speed hovering commands. The system response time is less than 100ms, ensuring that heavy objects can still be reliably hovered even under extreme conditions such as brake failure, thus eliminating the risk of falling objects.

[0040] In summary, the five-level safety obstacle avoidance mechanism addresses the problems of traditional safety protection mechanisms being singular, lacking tiered protection, and having untimely early warning and braking. It achieves full-process tiered control from early warning to emergency protection, enabling proactive risk avoidance and minimizing accident losses in emergencies. The coordinated design of each level of the mechanism solves the problem of disconnect between traditional safety protection and path planning and motion control, ensuring the coordination and timeliness of protective actions. The integration of multiple additional protections solves the problems of narrow coverage of traditional tower crane safety protection and the susceptibility to safety hazards such as hook slippage, overrun, and overload, achieving comprehensive safety protection. The design of soft limit virtual protection replaces some functions of traditional hardware limiters, solving the problems of easy damage, high maintenance costs, and low limit accuracy of hardware limiters, while improving the flexibility and reliability of safety protection, thus enhancing the overall safety of tower crane operations and reducing the accident rate.

[0041] The communication module uses 5G as the primary communication link and 230MHz or 433MHz radio as a backup link, forming a dual-link communication architecture. The specific operation mode is as follows: The communication module has a built-in 5G communication unit and a 230MHz / 433MHz radio communication unit. When working, the main link (5G) is started first, which is responsible for transmitting real-time data from the sensing module, instruction data from the path planning module, status data from the motion control module, and operation video data. The 5G link has the advantages of high-speed transmission and low latency (end-to-end latency ≤10ms), which meets the real-time control requirements. The backup link (230MHz / 433MHz radio) is in standby mode in real time, continuously monitoring the communication status of the main link. This backup link has strong diffraction capability and anti-blocking capability, and can adapt to the complex environment of the construction site (building blockage, electromagnetic interference), avoiding communication failure caused by the main link being blocked or signal interruption. The communication link integrates AES-256 data encryption algorithm, CRC data verification mechanism and packet loss retransmission mechanism. The encryption algorithm ensures that the transmitted data is not stolen or tampered with, the verification mechanism is used to detect errors in the data transmission process, and the packet loss retransmission mechanism ensures that lost data is resent in a timely manner, thus ensuring the integrity and security of data transmission. When the main link fails to function properly due to signal interruption, fault, or other reasons, the communication module automatically detects the link abnormality and completes a seamless switchover to the backup link within 500ms. The backup link takes over all communication tasks, ensuring the normal transmission of sensing data and control commands without affecting the tower crane's obstacle avoidance control and operation. Once the main link returns to normal, it automatically switches back to the main link, achieving dual-link redundancy backup and improving communication reliability.

[0042] The system and the cloud data management unit use the MQTT communication protocol for data interaction, and the messages are encapsulated in JSON format; AES-256 encryption algorithm and two-way authentication are used to ensure transmission security; the terminal maintains the link status with a 500ms heartbeat packet, and automatically switches to the backup link in case of an anomaly to ensure uninterrupted control.

[0043] In summary, the dual-link communication architecture solves the problems of traditional tower cranes' single communication link being susceptible to obstruction and signal interruption, leading to data transmission failure and control loss. It achieves redundant backup of the communication link, ensuring high reliability of communication. The application of the 5G main link solves the problems of slow transmission speed and high latency of traditional communication links, which cannot meet the requirements of real-time control, ensuring high-speed and low-latency transmission of various data and commands. The design of the 230MHz / 433MHz backup link solves the problems of weak signal and easy interruption of 5G links in complex construction sites (building obstruction, electromagnetic interference). Its strong diffraction and anti-obstruction capabilities are suitable for various complex scenarios. The integration of data encryption, verification and packet loss retransmission mechanisms solves the problems of traditional communication data being easily stolen, tampered with, and lost, ensuring the integrity and security of data transmission. The seamless link switching design solves the problems of communication interruption and control failure during link switching, ensuring the continuity of tower crane obstacle avoidance and operation, and improving the overall reliability of the system.

[0044] It also includes a cloud-based data management unit for real-time uploading of obstacle avoidance events, operating parameters, operation videos, and fault information. It supports historical data tracing and querying, tower crane health status assessment, and intelligent operation and maintenance management. The specific operating method is as follows: The cloud-based data management unit establishes a connection with various modules on the tower crane site through the communication module, and receives environmental and attitude data from the sensing module, decision data from the path planning module, operating parameters from the motion control module, fault and early warning information from the safety protection module, and operation video data in real time. All data is encrypted and then uploaded to the cloud server. The cloud server categorizes, stores, organizes, and analyzes the uploaded data to establish a tower crane operation database. Users can log in to the management platform via PC or mobile device to query historical operating parameters, obstacle avoidance event records, fault information, and operation videos, enabling the traceability of historical data and facilitating accident tracing and operation review. The cloud management unit has a built-in tower crane health status assessment model. By analyzing the tower crane's long-term operating parameters and fault records, it assesses the health status of the tower crane's various mechanisms (lifting, luffing, slewing) and core components (frequency converter, PLC, sensors), generates a health report, predicts potential faults, and issues maintenance reminders in advance. Meanwhile, the cloud management unit supports intelligent operation and maintenance management, which can formulate personalized maintenance plans based on the tower crane's operating status and maintenance records, reminding maintenance personnel to carry out equipment inspection and component replacement in a timely manner, thereby reducing the failure rate; The system is compatible with both newly built tower cranes and retrofits of existing tower cranes. Newly built tower cranes can directly integrate all modules to achieve full functionality, while existing tower cranes can be adapted to the existing hardware structure by adding sensing, communication, and control modules without replacing the entire equipment. It also supports private 5G deployment (suitable for large construction sites and closed areas) and operator 5G network deployment (suitable for ordinary construction sites), flexibly adapting to the deployment needs of different scenarios.

[0045] In summary, the design of the end-to-end cloud-based data management unit solves the problems of traditional tower cranes, such as lack of unified data management, inability to trace historical data, difficulty in tracing faults, and reliance on manual experience for operation and maintenance. It enables end-to-end control and traceability of tower crane operation data, facilitating accident review and problem investigation. The health status assessment and fault prediction functions solve the problems of passive operation and maintenance, susceptibility to sudden failures, and high maintenance costs in traditional tower cranes, enabling proactive operation and maintenance, early warning, and reducing failure rates and maintenance costs. The intelligent operation and maintenance management function solves the problems of traditional operation and maintenance plans being extensive and lacking specificity, improving operation and maintenance efficiency and effectiveness through personalized maintenance plans. The system's compatibility with both newly built and in-service tower crane retrofits solves the problems of high difficulty, high cost, and the need for complete equipment replacement in the automation retrofit of in-service tower cranes, reducing retrofit costs and expanding the system's application scope. Support for multiple 5G deployment modes solves the communication deployment needs of different construction sites (large enclosed construction sites, ordinary construction sites), enhancing the system's flexibility and applicability.

[0046] The intelligent load wind deflection prediction and trajectory compensation module has a built-in precast component load characteristic database, which stores various component geometric models, masses, and aerodynamic coefficients. Its specific operation is as follows: First, the precast component load characteristic database pre-inputs the geometric models, mass parameters, and aerodynamic coefficients (drag coefficient, lift coefficient, etc.) of various precast components commonly used in tower crane operations (such as steel bars, formwork, concrete components, etc.) to provide basic data support for wind deflection prediction. During operation, this module receives real-time visual data of the hook (used to obtain the three-dimensional attitude of the load, including tilt angle and rotation angle) and wind field data (wind speed and wind direction) collected by the perception module, and at the same time calls the parameters of the corresponding component in the prefabricated component load feature database; Based on the three-dimensional attitude of the load, wind field data, and aerodynamic parameters of the components, the force of the wind load on the suspended weight is estimated in real time through fluid dynamics algorithms. The horizontal drift, vertical sway, and rotational deviation of the suspended weight are predicted, and the required trajectory compensation is calculated. Subsequently, the module generates trajectory compensation instructions and simultaneously sends them to the path planning module and motion control module: the path planning module corrects the hoisting path according to the compensation instructions to avoid dangerous areas caused by wind deflection; the motion control module adjusts the output parameters of the frequency converter according to the compensation instructions to adjust the operating speed and acceleration of each mechanism of the tower crane, so as to realize real-time correction of the hoisting posture. Meanwhile, the module has an adaptive adjustment function, which can dynamically update compensation commands based on real-time wind field changes and load attitude changes, continuously correct trajectory and anti-sway parameters, ensure the accuracy and safety of tower crane hoisting in windy environments, and solve problems such as collisions and excessive swaying of the load caused by wind deflection in traditional tower cranes.

[0047] The wind deflection prediction fluid dynamics algorithm uses aerodynamic load calculation formulas. F = 0.5 × ρ × V 2 ×C×S In the formula, ρ is the air density, V is the real-time wind speed, C is the load aerodynamic coefficient, and S is the windward area; the horizontal drift of the suspended weight is calculated as ΔX = F × t. 2 / (2m) is calculated, where m is the weight of the load; the algorithm outputs the three-dimensional coordinate correction (ΔX,ΔY,ΔH) and the anti-sway gain correction value K. The correction is directly superimposed on the path planning target point and the motion control speed loop to realize the adaptive correction of wind disturbance trajectory.

[0048] In summary, the design of the precast component load characteristic database solves the problem of different wind deflection characteristics for different types of loads and the inability to accurately predict wind deflection, providing accurate data support for wind deflection prediction and adapting to various precast component hoisting scenarios. The linkage design of wind deflection prediction and trajectory compensation solves the problems of traditional tower cranes lacking a dedicated wind deflection compensation mechanism, large load drift under wind disturbance, easy collision, and low hoisting accuracy, realizing adaptive trajectory correction in wind disturbance environments. The application of fluid dynamics algorithms solves the problems of inaccurate wind deflection estimation and untimely compensation, ensuring the accuracy of compensation commands. The adaptive adjustment function solves the problem of compensation failure caused by dynamic changes in wind field and load attitude, realizing continuous and dynamic trajectory correction, improving the accuracy and safety of tower crane hoisting under wind disturbance environments, and making up for the shortcomings of traditional anti-sway mechanisms in dealing with the influence of wind deflection, further ensuring operational safety.

[0049] This system achieves closed-loop control of the entire tower crane operation process through the organic cooperation and coordinated operation of the sensing module, path planning module, motion control module, safety protection module, communication module, and intelligent load wind deviation prediction and trajectory compensation module.

[0050] This system, through the combined use of multi-source sensing and intelligent path planning, can acquire three-dimensional environmental and tower crane attitude information in real time, dynamically complete obstacle identification, hazard level determination and optimal path planning, enabling tower cranes to avoid danger in advance and automatically detour in multi-tower operations and complex working conditions, significantly improving operational safety and environmental adaptability. This system combines intelligent path planning with high-precision motion control to quickly convert planning commands into smooth and precise mechanical movements. Combined with sensorless anti-sway and zero-speed hovering functions, it effectively reduces load swaying and positioning deviation, improves lifting efficiency and placement accuracy, and reduces the intensity of manual intervention. This system combines motion control with multi-level safety protection, and can trigger protection actions according to the operating status and degree of danger. It ensures operating efficiency under normal working conditions, and can quickly brake and suspend heavy objects under fault or extremely dangerous working conditions, forming an integrated closed loop of control and protection, and comprehensively reducing safety risks such as collisions, falling objects, hook slippage, and top collisions. This system combines environmental perception and intelligent wind deflection compensation to simultaneously correct the operating trajectory based on real-time wind field, load attitude and obstacle position. In windy conditions, it can simultaneously achieve anti-sway, anti-deflection and anti-collision, thus expanding the applicable working conditions and operating time window of the tower crane. This system uses highly reliable dual-link communication to work with all system modules to ensure stable transmission of sensing data, control commands and protection signals. Even in the obstruction and interference environment of the construction site, the link can still be kept open to ensure the continuous effectiveness of intelligent control and safety mechanisms. This system enables tower cranes to achieve fully automated operation, precise control, full-process safety, wind resistance, and stable communication through the coordinated operation of all modules. It effectively improves the problems of traditional tower cranes, such as lagging perception, coarse control, single protection, and susceptibility to environmental and communication issues. At the same time, it is compatible with both newly built and in-service tower crane retrofits. Combined with cloud data management, it enables traceable operation, assessable health, and intelligent operation and maintenance, thus comprehensively improving the safety, stability, automation level, and engineering applicability of tower crane operations.

[0051] The embodiments described above are merely preferred embodiments of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications, improvements, and substitutions without departing from the concept of the present invention, and these all fall within the protection scope of the present invention.

Claims

1. A tower crane automatic obstacle avoidance path planning system based on machine vision and frequency converter, characterized in that... It includes a perception module, a path planning module, a motion control module, a safety protection module, a communication module, and an intelligent load wind deviation prediction and trajectory compensation module; The sensing module is used to collect data on the construction site environment and tower crane posture to achieve obstacle recognition, three-dimensional spatial modeling, and dynamic electronic fence construction. The path planning module is used to perform dynamic obstacle avoidance path planning and hazard level determination based on perception information and operational constraints. The motion control module is used to achieve precise speed regulation of multiple mechanisms of the tower crane, sensorless anti-swaying, and zero-speed hovering through a PLC and a vector closed-loop frequency converter. The security protection module is used to perform multi-level linkage security risk avoidance and fault protection; The communication module is used to construct a primary and backup dual link to achieve reliable data and command transmission; The intelligent load wind deflection prediction and trajectory compensation module is used to estimate wind load based on load aerodynamic characteristics, real-time attitude and wind field data and generate trajectory compensation instructions to realize adaptive correction of hoisting trajectory under wind disturbance.

2. The tower crane automatic obstacle avoidance path planning system based on machine vision and frequency converter according to claim 1, characterized in that, The sensing module includes a multi-channel visual acquisition unit, a lidar ranging unit, a centimeter-level positioning unit, a tower crane attitude sensor, and a wind speed and direction sensor; the hook position visual acquisition unit integrates automatic target tracking, automatic focusing, and wireless charging functions.

3. The tower crane automatic obstacle avoidance path planning system based on machine vision and frequency converter according to claim 1, characterized in that, The perception module uses deep learning algorithms to classify and identify obstacles such as buildings, scaffolding, openings, tower crane booms, personnel, and vehicles at the construction site. It integrates positioning and attitude data to establish a three-dimensional coordinate system with the tower crane's rotation center as the origin, updates the three-dimensional electronic fence in real time, and outputs information on obstacle location, hazard level, and intrusion trend.

4. The tower crane automatic obstacle avoidance path planning system based on machine vision and frequency converter according to claim 1, characterized in that, The path planning module triggers full-speed operation, automatic deceleration, path replanning, parking, emergency braking, and zero-speed hovering actions in stages according to four threshold levels: safe distance, warning distance, danger distance, and critical distance. The path planning is based on constraints such as torque limit, wind speed limit, turning angle limit, amplitude range, height limit, and interference zone of multiple towers.

5. The tower crane automatic obstacle avoidance path planning system based on machine vision and frequency converter according to claim 4, characterized in that, The path planning module supports fixed-point positioning, fixed-distance operation, one-click return, and multi-point sequential operation modes, enabling automatic hoisting, continuous operation, and multi-point cyclic execution, adapting to the needs of multi-tower collaboration and fixed working conditions.

6. The tower crane automatic obstacle avoidance path planning system based on machine vision and frequency converter according to claim 1, characterized in that, The motion control module uses digital communication to control the frequency converter, realizing vector closed-loop continuous speed regulation; the PLC uses an input shaping algorithm to calculate the anti-sway parameters in real time, realizing sensorless anti-sway under load.

7. The tower crane automatic obstacle avoidance path planning system based on machine vision and frequency converter according to claim 1, characterized in that, The safety protection module includes a five-level safety obstacle avoidance mechanism, namely, audible and visual warning and slight deceleration, speed limit and safety distance tightening, automatic detour trajectory generation, stop mechanism braking, brake failure protection and zero-speed hovering of heavy objects; it also includes soft limit virtual protection and multiple safety protections such as anti-hook slippage, anti-overrun, and anti-overload.

8. The tower crane automatic obstacle avoidance path planning system based on machine vision and frequency converter according to claim 7, characterized in that, The communication module uses high-speed mobile communication as the main link and long-distance radio as the backup link, and integrates data encryption, verification and packet loss retransmission mechanisms. When the main link is interrupted, the backup link seamlessly takes over the obstacle avoidance control.

9. The tower crane automatic obstacle avoidance path planning system based on machine vision and frequency converter according to claim 1, characterized in that, It also includes a cloud-based data management unit for uploading obstacle avoidance events, operating parameters, and operation videos in real time, supporting historical data tracing, tower crane health status assessment, and intelligent operation and maintenance management; the system is compatible with newly built tower cranes and the retrofitting of existing tower cranes, and supports the deployment of various communication networks.

10. The tower crane automatic obstacle avoidance path planning system based on machine vision and frequency converter according to claim 1, characterized in that, The intelligent load wind deviation prediction and trajectory compensation module has a built-in load feature database. It integrates load three-dimensional attitude and wind field data to estimate aerodynamic load in real time, predicts load drift and rotation deviation and generates compensation commands, and simultaneously corrects the hoisting trajectory and anti-sway parameters.

Images