Crane self-adaptive control method and control system thereof

Abstract

The invention belongs to the field of crane control, and particularly relates to a crane self-adaptive control method and a control system thereof.The crane self-adaptive control method includes the following steps that a three-dimensional digital map of a crane working area is obtained, and a hoisting point, a discharging point, a static obstacle point and a dynamic obstacle point are marked in the map; obstacles are accurately marked through a three-dimensional digital map, a safe distance is added, path obstacle avoidance is ensured, dynamic obstacles are detected and tracked in real time, a local planner performs real-time obstacle avoidance, and the collision risk is reduced. Multiple optimal paths are generated through multi-objective optimization, intelligent selection is carried out according to work requirements, and the method adapts to different scenes. The global and local plans are combined to generate a smooth and executable trajectory, reducing idle travel and dwell time. Self-adaptive tracking control compensates interference in real time through sensor feedback and an algorithm, and tracks accurately; by integrating sensing, planning and control technologies, safe, efficient, stable and self-adaptive control over hoisting operation of the crane is achieved, and the method is suitable for diversified industrial scenes.

Description

Technical Field

[0001] This invention relates to the field of crane control, and more specifically to a crane adaptive control method and its control system. Background Technology

[0002] Crane adaptive control is a technology that adjusts control strategies in real time to cope with environmental changes and system uncertainties, aiming to improve the stability and safety of cranes under complex working conditions.

[0003] Avoiding static and dynamic obstacles is a challenge in traditional crane operation, especially dynamic obstacles such as people and vehicles, which can easily lead to collisions. In addition, load swing may cause collisions or inaccurate landing. Crane operation requires optimization of time, path, sway, and energy consumption, but traditional methods often cannot optimize these factors simultaneously, resulting in low operational efficiency. Crane dynamics are complex, and load swing is difficult to control. Traditional control methods may not be able to adapt to external disturbances such as wind load or load changes. Path planning needs to handle obstacles in a 3D environment and generate smooth, executable trajectories. Traditional planning may not be able to adapt to changes in the environment in real time. Summary of the Invention

[0004] (a) Purpose of the invention To address the technical problems existing in the background art, this invention proposes an adaptive control method for cranes, which features improved safety, improved operating efficiency, and improved control accuracy.

[0005] (II) Technical Solution To address the aforementioned technical problems, this invention provides an adaptive control method for cranes, comprising the following steps: Step 1: Obtain a 3D digital map of the crane's working area and mark the lifting points, unloading points, static obstacle points, and dynamic obstacle points on the map; Step 2: Constrain and optimize the hoisting path. Constraints include safety constraints and load swing constraints. Optimization includes shortest time optimization, shortest path optimization, minimum swing optimization, and minimum energy consumption optimization. Step 3: Perform global and local planning of the cargo hoisting path on the 3D digital map. Define at least one safe path from the start point to the end point that avoids all obstacle points. Local planning will smooth the safe path generated by global planning and make it meet the dynamic constraints of the crane. Step 4: The obtained safe paths include the time-optimal path, the swing-optimal path, and the energy-optimal path, which are selected according to the work requirements; Step 5: Convert the geometric information of the safe path into a time-parameterized trajectory that the crane can execute, and perform adaptive tracking control during the operation of the trajectory.

[0006] Preferably, step one further includes, when acquiring a three-dimensional digital map of the crane's working area, the information acquisition sensors for the crane's working area include LiDAR, binocular / depth camera, industrial camera, decoder / GPS / UWB; Point cloud data collected at different times and from different perspectives are stitched together to form a globally consistent 3D digital map, and noise points are removed. The three-dimensional space is divided into uniform cubic grids, and each cubic grid is marked as occupied, free, or unknown; Semantic information is added to each marker point on the constructed 3D digital map to help the operating system recognize and understand each marker point.

[0007] Preferably, the marking of hoisting and unloading points includes: pre-marking the coordinates (X, Y, Z) of all standard workstations in a three-dimensional digital map; Visual recognition technology enables cameras to automatically identify specific signs or features and calculate their position relative to the crane, updating the data in real time to a 3D digital map. Static obstacle labeling includes: automatically separating each obstacle from a 3D digital map using a point cloud clustering algorithm, and adding labels to the obstacles by combining visual information or a predefined model library; Add a safety distance to static obstacle points. Based on the crane's safety margin and positioning error, add a safety distance outward to the geometric model of each static obstacle to generate a doubled virtual obstacle. Subsequent path planning will be carried out in the obstacle space with the added safety distance. Dynamic obstacle detection and tracking includes: continuously comparing sensor data of the current frame with a static background map, and identifying moving objects, such as ground personnel, vehicles, and other crane booms, through subtraction or deep learning object detection algorithms; For detected dynamic obstacles, a tracking algorithm is used to estimate their speed and direction, and to predict their trajectory in the next few seconds.

[0008] Preferably, in step two, the static obstacle avoidance constraint is: any point on the path, including the entire spatial envelope of the crane trolley, crane hook, and load, must maintain a minimum safe distance from all static obstacles, including virtual obstacles for which a safety distance has been added. In a cube grid-based map, this is simplified to the path point not falling into a grid marked as occupied or unknown. Dynamic obstacle avoidance constraint: The path must avoid the predicted trajectory area of ​​all dynamic obstacles within the prediction period. The local planner will consider the predicted position and bounding box of the dynamic obstacle in the next few seconds in real time and use it as a temporary obstacle for obstacle avoidance planning. Load sway constraints are designed to control the swaying of loads during transportation, preventing collisions and inaccurate landings.

[0009] Preferably, the shortest time optimization objective is to find a trajectory with the shortest time while satisfying the oscillation and safety constraints; The goal of shortest path optimization is to find a collision-free path with the shortest geometric length. Minimum sway optimization aims to minimize the sway amplitude or sway energy of the load throughout the transportation process. This involves adding brief pauses or corresponding speed change curves to counteract the sway and setting speed constraints at corresponding locations along the path. The goal of minimum energy consumption optimization is to minimize the total energy consumed by the crane to complete the lifting task. Avoiding unnecessary starts and stops and keeping the crane running at high speed helps to reduce energy consumption.

[0010] Preferably, in step three, global planning involves searching for at least one connected, collision-free geometric path from the starting point hoisting point to the ending point unloading point on the three-dimensional digital map established in steps one and two, which includes all static obstacles, hoisting points, and unloading points. When performing global planning, the 3D digital map is gridded into cubes, with each cell marked with the 3D coordinates of occupied, free, unknown, start and end points. Using a graph search algorithm, the shortest path is found by evaluating the cost of the path already taken and the estimated cost to the end point. In the 3D grid, a shortest feasible path that bypasses obstacles can be found. Local planning transforms the shortest feasible path given by global planning into a fine trajectory that allows the crane to execute smoothly, stably, and safely.

[0011] Preferably, in step four, an optimal path is selected from the crane paths generated in the above steps and executed based on actual working conditions such as task urgency, load characteristics, and energy consumption requirements.

[0012] Preferably, in step five, the geometric information of the path is transformed into a trajectory with time parameters. That is, each path point is associated with specific timestamps, speeds, accelerations and other dynamic parameters. This ensures that the trajectory not only meets spatial constraints, but also conforms to the dynamic characteristics of the crane, such as the maximum speed and acceleration limits of the trolley, hoisting mechanism and the crane itself. This makes the trajectory physically executable, smooth and continuous, and avoids sudden stops and starts.

[0013] Preferably, in step five, during the actual operation of the crane, the status of the crane body and load, such as position, speed, and swing angle, is monitored in real time by sensors such as encoders, IMUs, and vision cameras, and compared with the predetermined trajectory. Based on the feedback control algorithm, the system adaptively adjusts the output of the drive mechanism to compensate for external disturbances such as wind load, dynamic obstacles, model errors, or load changes, ensuring that the trajectory is accurately tracked while maintaining the stability and safety of the load.

[0014] The present invention also provides an adaptive control system for a crane, comprising: The map building module acquires a 3D digital map of the crane's working area and marks lifting points, unloading points, static obstacle points, and dynamic obstacle points on the map. The constraint optimization module constrains and optimizes the hoisting path. Constraints include safety constraints and load swing constraints, while optimizations include shortest time optimization, shortest path optimization, minimum swing optimization, and minimum energy consumption optimization. The path planning module performs global and local planning of cargo hoisting paths on a 3D digital map, defining at least one safe path from the start point to the end point that avoids all obstacle points. The local planning smooths the safe path generated by the global planning and makes it meet the dynamic constraints of the crane. The path decision module obtains safe paths including the time-optimal path, the swing-optimal path, and the energy-optimal path, which are selected according to work requirements; The adaptive tracking module converts the geometric information of the safe path into a time-parameterized trajectory that the crane can execute, and performs adaptive tracking control during the operation of the trajectory.

[0015] The above-described technical solution of the present invention has the following beneficial technical effects: 1. Obstacles are accurately marked using a 3D digital map, and safe distances are added to ensure obstacle avoidance on the path. Dynamic obstacles are detected and tracked in real time, and a local planner performs real-time obstacle avoidance to reduce the risk of collision.

[0016] 2. Multi-objective optimization generates multiple optimal paths, which are intelligently selected according to work requirements to adapt to different scenarios.

[0017] 3. Combining global and local planning generates smooth, executable trajectories, reducing idle travel and pause time.

[0018] 4. Adaptive tracking control uses sensor feedback and algorithms to compensate for interference in real time and accurately track the trajectory.

[0019] 5. Automated processes enable even novice operators to complete challenging lifting operations, reducing reliance on driver experience and improving overall operational reliability.

[0020] By integrating sensing, planning, and control technologies, safe, efficient, stable, and adaptive control of crane lifting operations is achieved, making it suitable for diverse industrial scenarios. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the adaptive control method for cranes according to the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments and the accompanying drawings. It should be understood that these descriptions are merely exemplary and not intended to limit the scope of the invention. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0023] like Figure 1 As shown, the present invention proposes an adaptive control method for cranes, comprising the following steps: Step 1: Obtain a 3D digital map of the crane's working area and mark the lifting points, unloading points, static obstacle points, and dynamic obstacle points on the map; Step 2: Constrain and optimize the hoisting path. Constraints include safety constraints and load swing constraints. Optimization includes shortest time optimization, shortest path optimization, minimum swing optimization, and minimum energy consumption optimization. Step 3: Perform global and local planning of the cargo hoisting path on the 3D digital map. Define at least one safe path from the start point to the end point that avoids all obstacle points. Local planning will smooth the safe path generated by global planning and make it meet the dynamic constraints of the crane. Step 4: The obtained safe paths include the time-optimal path, the swing-optimal path, and the energy-optimal path, which are selected according to the work requirements; Step 5: Convert the geometric information of the safe path into a time-parameterized trajectory that the crane can execute, and perform adaptive tracking control during the operation of the trajectory.

[0024] In step one, when acquiring a three-dimensional digital map of the crane's working area, the information acquisition sensors for the crane's working area include LiDAR, binocular / depth camera, industrial camera, decoder / GPS / UWB; The lidar acquires three-dimensional point cloud data of the surrounding environment by emitting laser beams. It is installed on the main beam of the crane and its field of view covers the area where the trolley is running and below it. Binocular / depth cameras acquire depth information through visual computing or structured light, generate RGB-D images, and integrate with the hook trolley to provide close-range perception of the load's surroundings. Industrial cameras are used for visual recognition to assist in positioning and status monitoring, and to monitor lifting points and unloading points to assist in precise positioning. Encoders / GPS / UWB measure the position of the crane's trolley and crane, providing the crane's own positioning information. They serve as a reference system for map stitching and are deployed on the crane's traveling mechanism and in the factory area. The Kalman filter algorithm is used to fuse the acquired 3D point cloud data, visual information, and the encoder pose of the crane itself to generate a consistent environment model in real time.

[0025] As an example of constructing and representing a 3D digital map: Point cloud data collected at different times and from different perspectives are stitched together to form a globally consistent 3D digital map, and noise points are removed. The three-dimensional space is divided into uniform cubic grids, each marked as occupied, free, or unknown. The purpose is to make the cubic grid division method computationally efficient and applicable to real-time obstacle avoidance.

[0026] Examples of marking hoisting points, unloading points, static obstacle points, and dynamic obstacle points: Semantic information is added to each marker point on the constructed 3D digital map to help the operating system recognize and understand each marker point.

[0027] The marking of hoisting and unloading points includes: pre-marking the coordinates (X, Y, Z) of all standard workstations in a 3D digital map; and using visual recognition technology, enabling cameras to automatically identify specific markers or features and calculate their position relative to the crane, updating the 3D digital map in real time.

[0028] Static obstacle labeling includes: automatically separating individual obstacles from a 3D digital map using point cloud clustering algorithms, and adding labels to the obstacles by combining visual information or a predefined model library.

[0029] In an optional embodiment, a safety distance is added to the static obstacle points. Based on the crane's safety margin and positioning error, a safety distance is added outward to the geometric model of each static obstacle, generating a multiplied virtual obstacle. Subsequent path planning will be carried out in the obstacle space with the added safety distance, thereby ensuring physical safety.

[0030] Dynamic obstacle detection and tracking includes: continuously comparing sensor data of the current frame with a static background map, and identifying moving objects, such as ground personnel, vehicles, and other crane booms, through subtraction or deep learning object detection algorithms; For detected dynamic obstacles, a tracking algorithm is used to estimate their speed and direction, and to predict their trajectory in the next few seconds.

[0031] As an example of updating a 3D digital map: the current position, bounding box, and predicted trajectory of these dynamic obstacles are added to the 3D digital map in real time as temporary obstacles for use by the local planner.

[0032] Understandably, the working environment of cranes is not static, and the map needs to be able to be updated. Based on the existing 3D digital map, incremental updates can be made for local changes in the environment, such as the addition of temporary equipment, without having to rebuild the entire map.

[0033] In step two, safety constraints include static obstacle avoidance constraints, dynamic obstacle avoidance constraints, crane mechanical constraints, and workspace constraints. The static obstacle avoidance constraint is: any point on the path, including the entire spatial envelope of the crane trolley, crane hook, and load, must maintain a minimum safe distance from all static obstacles, including virtual obstacles for which a safety distance has been added. In a cube grid-based map, this is simplified to the path point not falling into a grid marked as occupied or unknown.

[0034] Understandably, unknown areas are considered obstacles for security reasons.

[0035] Dynamic obstacle avoidance constraint: The path must avoid the predicted trajectory area of ​​all dynamic obstacles within the prediction period. The local planner will consider the predicted position and bounding box of the dynamic obstacle in the next few seconds in real time and use it as a temporary obstacle for obstacle avoidance planning.

[0036] Crane's own mechanical constraints: The path must be within the travel limits of the trolley, hoisting mechanism, and other components.

[0037] Workspace constraints: The range of motion of the hook and load must not exceed the physical space limited by the crane boom span, lifting height, etc.

[0038] Furthermore, load sway constraints are designed to control the swaying of the load during transportation, preventing collisions and inaccurate landings.

[0039] Maximum swing angle constraint: The swing angle of the load must be limited to a maximum allowable value. within, The load oscillation is modeled as a simple pendulum, and its dynamic equations are related to the acceleration of the crane trolley / gantry. Directly related: Where L is the rope length and g is the gravitational acceleration, this constraint is a constraint on the acceleration of the crane.

[0040] Furthermore, optimization includes shortest time optimization, shortest path optimization, minimum oscillation optimization, and minimum energy consumption optimization. Optimization is to find the optimal path that best meets one or more specific objectives among feasible paths that satisfy all constraints.

[0041] The shortest time optimization objective is to minimize the total time T from the hoisting point to the unloading point; Each mechanism of the crane should accelerate to the maximum permissible speed with the maximum permissible acceleration and decelerate as late as possible. Direct acceleration and deceleration in this way will cause the load to swing violently. Therefore, time optimization must be carried out in conjunction with swing constraints. Under the premise of satisfying swing and safety constraints, the shortest trajectory should be found.

[0042] The goal of shortest path optimization is to minimize the spatial geometric length L of the path. In the absence of obstacles, the shortest path is a straight line between the hoisting point and the unloading point. However, in a three-dimensional space with obstacles, the shortest path is a smooth curve that bypasses the obstacles. Search algorithms such as A, D, and RRT* are used to find a collision-free path with the shortest geometric length.

[0043] Among them, minimum sway optimization is to minimize the sway amplitude or sway energy of the load throughout the entire transportation process; Used to minimize the maximum swing angle throughout the process; Used to minimize the oscillation energy throughout the process; On the planned path of the crane body, the sway is counteracted by adding brief pauses or corresponding speed change curves, which is equivalent to setting speed constraints at corresponding positions on the path.

[0044] Among them, the minimum energy consumption optimization objective is to minimize the total energy consumed by the crane to complete the lifting task; Where F is the driving force and v is the speed, energy consumption is related to acceleration, speed and running time. Smooth acceleration and deceleration can avoid unnecessary starts and stops and keep the crane running at high speed, which helps to reduce energy consumption.

[0045] In one embodiment, in practical applications, the above optimization objectives often need to be considered simultaneously. For example, users may want to complete the hoisting as quickly as possible while keeping the sway within 10 degrees and minimizing energy consumption. A hierarchical optimization method can be used to optimize according to priority. First priority: satisfy all safety constraints and oscillation constraints.

[0046] Second priority: Optimize time while ensuring the first priority is met.

[0047] Third priority: Among paths with similar time intervals, choose the one with the lowest energy consumption; In this way, the system can intelligently generate efficient and stable hoisting operation trajectories that adapt to different work requirements while ensuring safety.

[0048] In step three, global planning involves searching for at least one connected, collision-free geometric path from the starting point hoisting point to the ending point unloading point on the three-dimensional digital map established in steps one and two, which includes all static obstacles, hoisting points, and unloading points.

[0049] When performing global planning, the 3D digital map is gridded into cubes, with each cell marked with the 3D coordinates of occupied, free, unknown, start and end points. Using a graph search algorithm, the shortest path is found by evaluating the cost of the path already taken and the estimated cost to the end point. A shortest feasible path around obstacles can be found in the 3D grid.

[0050] Understandably, global planning is based on information about the entire environment, focusing on whether it is possible to pass rather than at what speed or smoothly.

[0051] Local planning transforms the shortest feasible path given by global planning into a fine trajectory that allows the crane to execute smoothly, stably, and safely.

[0052] Taking local planning as an example: input the path generated by global planning and real-time sensor data, i.e., dynamic obstacle information; Using a curve fitting algorithm, sharp corners in the global path are smoothed out. A smooth path helps reduce sudden stops and starts of the crane, reduces energy consumption and load sway, and achieves path smoothing. Considering the maximum speed and acceleration of the crane's trolley, hoisting mechanism, the output of local programming is not a geometric line, but a trajectory with time parameters. That is, each point corresponds to a timestamp, speed, and acceleration, which ensures that the trajectory is physically executable. The local planner continuously receives sensor information and detects unforeseen dynamic obstacles such as people and vehicles in the global plan. When an obstacle is found to have encroached on the predetermined path, the local planner will immediately perform a small-scale local detour under the guidance of the global path to generate a new, collision-free local trajectory for real-time obstacle avoidance.

[0053] By combining global and local planning, the polygonal path is smoothed into a curve and dynamic constraints are incorporated to generate a trajectory with continuous speed. This significantly suppresses load sway, improves positioning accuracy and safety, reduces mechanical shock during smooth movement, extends equipment life, and allows for higher speed operation under stable conditions. The entire planning process is completed automatically, eliminating the need for repeated judgments and micro-operations by the driver. Even novices can complete challenging hoisting tasks, allowing them to focus more on overall monitoring and emergency response.

[0054] Step four is performed after the global and local planning in step three. The obtained safe paths include the time-optimal path, the swing-optimal path, and the energy-optimal path. Step three has generated multiple safe paths from the hoisting point to the unloading point. These paths all satisfy safety constraints such as obstacle avoidance, mechanical constraints, and load swing constraints. The role of step four is to select the optimal path from these paths based on actual working conditions such as task urgency, load characteristics, and energy consumption requirements.

[0055] Among them, the time-optimal path minimizes the total time (T) of the hoisting task. This means that the crane's trolley, hoisting mechanism and hoisting mechanism will run at the maximum allowable acceleration and speed to minimize idle or deceleration time. However, in order to meet the swing constraints and safety constraints, time optimization is not a simple acceleration and deceleration, but a shortening of time under the premise of ensuring stability through collaborative planning, such as the hierarchical optimization described in step 2.

[0056] Applicable scenarios: When the task is urgent and hoisting needs to be completed quickly, choosing this route can improve efficiency.

[0057] Among them, the optimal swing path minimizes the swing amplitude or swing energy of the load throughout the transportation process. This is achieved by limiting the maximum swing angle or minimizing the swing energy. During planning, brief pauses or specific speed curves may be added to actively cancel the swing and ensure the load is stable.

[0058] Applicable scenarios: When the load is fragile, delicate, or requires high-precision positioning, choosing this path can reduce the risk of collision and improve the accuracy of the landing point.

[0059] Among them, the energy-optimal path minimizes the total energy consumed by the crane to complete the lifting task. Energy consumption is related to acceleration and speed curves. Energy consumption is reduced by smooth acceleration and deceleration and avoiding unnecessary starts and stops. The planning tends to use uniform motion or optimize the speed curve.

[0060] Applicable scenarios: When energy saving is the primary consideration, choosing this path can reduce operating costs.

[0061] Understandably, selecting methods based on work requirements reflects the adaptability and flexibility of the approach. The selection process can be manual, with operators choosing through an interface, or automatic, with the system making decisions based on preset rules. Specific selection criteria include: Task priority: If time is tight, choose the time-optimal path first; if safety and stability are the priorities, choose the swing-optimal path.

[0062] Load characteristics: The weight, shape, fragility, etc. of the load will affect the selection. For example, heavy loads focus on swing control, while light loads focus on time.

[0063] Environmental factors: Are there dynamic obstacles in the work area, such as people or vehicles? If so, the system will automatically tend to swing to the optimal path to ensure safety.

[0064] Economic efficiency: If energy consumption costs are high, the energy-optimal path becomes the default choice.

[0065] In one embodiment, step four enables the control system to adapt to changing work scenarios and make intelligent decisions instead of using a fixed path, thereby achieving a balance between efficiency, safety, and economy. By generating diverse paths through multi-objective optimization and selecting them in combination with actual needs, the adaptability, efficiency, and safety of crane lifting operations are ensured.

[0066] In step five, the geometric information of the path is transformed into a trajectory with time parameters. That is, each path point is associated with specific timestamps, speeds, accelerations and other dynamic parameters. This ensures that the trajectory not only meets spatial constraints, but also conforms to the dynamic characteristics of the crane, such as the maximum speed and acceleration limits of the trolley, hoisting mechanism, etc., making the trajectory physically executable, smooth and continuous, and avoiding sudden stops and starts.

[0067] Furthermore, during the actual operation of the crane, sensors such as encoders, IMUs, and vision cameras monitor the status of the crane body and load in real time, including position, speed, and swing angle, and compare them with the predetermined trajectory. Based on feedback control algorithms, the system adaptively adjusts the output of the drive mechanism to compensate for external disturbances such as wind load, dynamic obstacles, model errors, or load changes, ensuring that the trajectory is accurately tracked while maintaining the stability and safety of the load.

[0068] As an example, the system uses encoders, GPS / UWB, IMU, and vision sensors to acquire the actual position and speed information of the crane's trolley, crane hook, etc. in real time, and uses cameras or lidar to directly measure the swing angle of the load. Because sensor data contains noise and delay, a state observer or Kalman filter algorithm is used to make optimal estimates of quantities that cannot be directly measured or are not measured accurately, such as load swing speed and the true speed of the system, so as to obtain a reliable and complete view of the current state. By continuously comparing the error between the predetermined trajectory and the estimated actual state using an adaptive controller, and employing PID control, fuzzy control, etc., the basic control force is calculated based on the position and velocity errors. The controller's built-in algorithm identifies changes in system parameters online to resist unknown interference.

[0069] Based on the current state and system model, the behavior of the system is predicted in the near future, and a series of optimal control commands are calculated to minimize the deviation from the predetermined trajectory, actively anticipate and compensate for sway, and adjust the controller's own parameters in real time to adapt to changes in the dynamic characteristics of the crane.

[0070] Actuators: The control quantities calculated by the controller are used to send speed or torque commands to the motor drivers and actuators of the trolley, hoisting mechanism, etc., to drive them to act and eliminate tracking errors.

[0071] It should be understood that the specific embodiments described above are merely illustrative or explanatory of the principles of the invention and do not constitute a limitation thereof. Therefore, any modifications, equivalent substitutions, improvements, etc., made without departing from the spirit and scope of the invention should be included within the protection scope of the invention. Furthermore, the appended claims are intended to cover all variations and modifications falling within the scope and boundaries of the appended claims, or equivalent forms of such scope and boundaries.

Claims

1. A method of adaptive control of a crane, characterized in that, The method comprises the following steps: Step one, obtain a three-dimensional digital map of the crane working area, mark the lifting points, unloading points, static obstacle points and dynamic obstacle points in the map; Step two, constraint and optimization of the lifting path, the constraints include safety constraints and load swing constraints, the optimization includes shortest time optimization, shortest path optimization, minimum swing optimization and lowest energy consumption optimization; Step three, global planning and local planning of the cargo lifting path in the three-dimensional digital map, define at least one safe path from the starting point to the end point avoiding all obstacle points, the local planning smooths the safe path generated by the global planning and makes it meet the dynamics constraints of the crane; Step four, the obtained safe path includes the time optimal path, the swing optimal path and the energy optimal path, select according to the work demand; Step five, convert the geometric information of the safe path into a crane executable trajectory with time parameter, and adaptively track and control during the running trajectory.

2. A crane adaptive control method according to claim 1, characterized in that, Step one also includes, when obtaining the three-dimensional digital map of the crane working area, the information acquisition sensor of the crane working area includes laser radar, binoculars / depth camera, industrial camera, decoder / GPS / UWB; Splice the point cloud data collected at different times and from different angles into a globally consistent three-dimensional digital map, and remove the noise points; Divide the three-dimensional space into uniform cubic cells, and mark each cubic cell as occupied, idle or unknown; Add semantic information to each marked point on the constructed three-dimensional digital map, which is used to assist the operating system to recognize and understand each marked point.

3. A crane adaptive control method according to claim 2, characterized in that, The marking of lifting points and unloading points includes: pre-accurately marking the coordinates (X, Y, Z) of all standard stations in the three-dimensional digital map; Through visual recognition technology, the camera automatically recognizes specific marks or features and calculates their position relative to the crane, and updates it in real time to the three-dimensional digital map; The marking of static obstacle points includes: using point cloud clustering algorithm to automatically separate each obstacle from the three-dimensional digital map, and adding labels to the obstacles by combining visual information or a pre-defined model library; Add a safety distance to the static obstacle points, according to the safety margin and positioning error of the crane, add a safety distance to the geometric model of each static obstacle, generate a virtual obstacle multiplied, and the subsequent path planning will be carried out in the space of the obstacle with the added safety distance; Dynamic obstacle detection and tracking includes: continuously comparing the sensor data of the current frame with the static background map, and identifying moving objects such as ground personnel, vehicles, other crane jibs, etc. through subtraction or deep learning target detection algorithm; For the detected dynamic obstacles, estimate their speed, direction, and short-term predict their trajectory in the next few seconds.

4. A crane adaptive control method according to claim 1, characterized in that, In step two, static obstacle avoidance constraint: the entire space envelope of any point on the path including the crane trolley, trolley, hook and load must maintain a minimum safety distance from all static obstacles including virtual obstacles with added safety distance, in the map based on cubic grid, it is simplified that the path point cannot fall into the grid marked as occupied or unknown; Dynamic obstacle avoidance constraint: the path must avoid the predicted trajectory area of all dynamic obstacles within the prediction period, and the local planner will consider the predicted position and bounding box of dynamic obstacles in the next few seconds in real time and treat them as temporary obstacles for obstacle avoidance planning; Load swing constraint is to control the swing of the load during transportation, prevent collision and inaccurate landing point.

5. The crane adaptive control method according to claim 1, characterized in that, The shortest time optimization objective is to find a trajectory with the shortest time under the premise of meeting the swing and safety constraints; The shortest path optimization objective is to find a collision-free path with the shortest geometric length; The minimum swing optimization is to minimize the swing amplitude or swing energy of the load during the entire transportation process, and to add a short pause or a corresponding speed change curve to cancel the swing, and to set the speed constraint at the corresponding position of the path; The minimum energy consumption optimization objective is to minimize the total energy consumption of the crane to complete the lifting task, and to avoid unnecessary start and stop and to maintain high-speed operation of the crane to help reduce energy consumption.

6. A crane adaptive control method according to claim 1, characterized in that, In step three, the global planning is performed on the three-dimensional digital map established in steps one and two, which contains all static obstacles, lifting points, and unloading points, to search for at least one connected collision-free geometric path from the starting lifting point to the ending unloading point; During global planning, the three-dimensional digital map is cubed, each grid is marked as occupied, free, unknown, starting point, and ending point three-dimensional coordinates, and a graph search algorithm is used to find the shortest path by evaluating the cost of the path taken and the estimated cost to the end point, and a shortest feasible path that bypasses obstacles can be found in the three-dimensional grid; Local planning is to process the at least one shortest feasible path given by global planning into a fine trajectory that can be smoothly, stably, and safely executed by the crane.

7. A crane adaptive control method according to claim 1, characterized in that, In step four, from the crane path generated in the above steps, an optimal path is selected for execution according to actual working conditions such as task urgency, load characteristics, energy consumption requirements, etc.

8. A crane adaptive control method according to claim 1, characterized in that, In step five, the geometric information position coordinates of the path are converted into a trajectory with time parameters, i.e. each path point is associated with specific time stamps, speed, acceleration, etc. dynamic parameters, ensuring that the trajectory not only meets the spatial constraints, but also meets the dynamics characteristics of the crane such as the maximum speed and acceleration limits of the trolley, the trolley and the lifting mechanism, making the trajectory physically executable, smooth and continuous, and avoiding sudden stops and starts.

9. A crane adaptive control method according to claim 1, characterized in that, In step five, during the actual operation of the crane trajectory, the state of the crane body and the load such as position, speed, swing angle is monitored in real time through sensors such as encoders, IMUs, and vision cameras, and compared with the predetermined trajectory, based on feedback control algorithm, the system adaptively adjusts the output of the driving mechanism to compensate for external disturbances such as wind load, dynamic obstacles, model errors or load changes, ensuring that the trajectory is accurately tracked while maintaining the stability and safety of the load.

10. A crane adaptive control system according to any one of claims 1-9, characterized in that, It includes: A map construction module that obtains a three-dimensional digital map of the crane's working area, and marks lifting points, unloading points, static obstacle points, and dynamic obstacle points in the map; The constraint optimization module, which is used to constrain and optimize the lifting path, includes safety constraint and load swing constraint, and the optimization includes shortest time optimization, shortest path optimization, minimum swing optimization and lowest energy consumption optimization; The path planning module, which is used to plan the global and local path of the cargo lifting in the three-dimensional digital map, defines at least one safe path from the starting point to the end point avoiding all obstacle points, and the local planning smooths the safe path generated by the global planning and makes it meet the dynamics constraint of the crane; The path decision module, which is used to select the time optimal path, swing optimal path and energy optimal path according to the work demand; The adaptive tracking module, which is used to convert the geometric information of the safe path into the crane executable trajectory with time parameter, and adaptively tracks and controls in the running trajectory process.

Images