Dual-mode stacking and automatic alignment control method and system for stacker-reclaimer

Abstract

The invention relates to a dual-mode stacking and automatic alignment control method and system for a stacker-reclaimer. The method comprises the steps that real-time sensing and modeling are carried out; intelligent mode decision making; performing dynamic alignment calculation; trajectory planning and execution; and carrying out closed-loop calibration and anti-collision control. The system comprises a three-dimensional sensing subsystem; an edge calculation and control subsystem; an equipment driving subsystem; a cloud collaborative optimization platform; the edge calculation and control subsystem comprises a data fusion and modeling module, a mode decision module, an intelligent alignment calculation module and a trajectory planning and motion control module. The problems that in the prior art, the control mode is single, and alignment depends on fixed coordinates are solved, intelligent decision making, accurate dynamic alignment and closed-loop safety control are achieved, and the overall efficiency and quality of material stacking and taking operation are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of automation technology for bulk cargo terminal and yard material handling, and in particular to a dual-mode stacking and automatic alignment control method and system for a stacker-reclaimer. Background Technology

[0002] As a key piece of equipment in bulk cargo yards, the level of automation and intelligence of stacker-reclaimers directly affects operational efficiency and safety. Currently, the automatic control of stacker-reclaimers mainly revolves around path pre-setting and sensor feedback.

[0003] For example, Chinese invention patent CN113830569B discloses a stacking control method and stacking system. This solution uses ground-based photographic equipment, height sensors, and pressure sensors to acquire stacking information (such as height and weight). A PLC system then determines whether the stacking conditions are met and controls the stacker to sequentially complete the stacking operations at each location. This technology represents the current mainstream automation approach, realizing a shift from manual to automated processes.

[0004] However, analysis reveals the following limitations of this existing technology:

[0005] 1. Single control mode: It adopts a fixed and sequential stacking logic, which cannot adaptively switch strategies according to the operation stage (such as rapid stacking and later shaping) or the final goal (such as preparing for subsequent efficient material retrieval), making it difficult to take into account the overall efficiency of the operation chain.

[0006] 2. Discrete sensing and alignment methods: Relying on multiple discrete sensors for single-point or two-dimensional judgments (such as whether a preset height has been reached), lacking continuous and accurate sensing of the overall three-dimensional geometry of the material pile. The "alignment" in its technical solution essentially refers to driving the equipment to a preset fixed spatial coordinate, rather than dynamic and intelligent spatial alignment based on the real-time shape of the material pile.

[0007] 3. Insufficient decision-making dimensions: Control decisions are based on limited local state information (such as whether a single stack is full) and are not deeply integrated with material characteristics, global operational tasks (such as urgency) and better stacking objectives, resulting in limited intelligence.

[0008] Therefore, existing technologies struggle to achieve synergistic optimization of efficiency, quality, and safety in stack-reclaim operations, particularly failing to address the core issue of "how to make intelligent decisions and accurately align based on real-time three-dimensional stockpile morphology and global task objectives." Summary of the Invention

[0009] The present invention aims to overcome the shortcomings of the prior art and provides a dual-mode stacking and automatic alignment control method and system for stacker-reclaimers.

[0010] To achieve the above objectives, the present invention adopts the following technical solution:

[0011] The dual-mode stacking and automatic alignment control method for stacker-reclaimers includes the following steps:

[0012] S1. Real-time perception and modeling: Obtain the real-time 3D point cloud model of the work area and the preset work task parameters;

[0013] S2. Intelligent Mode Decision-Making: Based on the 3D point cloud model and task parameters, it adaptively selects between an efficiency-first mode and a regularity-first mode; among which...

[0014] The efficiency-first model aims to maximize the material throughput per unit time.

[0015] The regularity priority mode aims to optimize the formation of a regular stockpile profile that facilitates subsequent full-section material extraction.

[0016] S3. Dynamic alignment calculation: Based on the selected target mode, the alignment algorithm corresponding to the mode is called to analyze the three-dimensional point cloud model, calculate and output an optimal alignment point that matches the current dynamic geometric features of the stockpile and the operation target in real time.

[0017] S4. Trajectory Planning and Execution: Based on the optimal alignment point, plan the motion trajectory of the stacker-reclaimer boom and control it to perform stacking or reclaiming operations.

[0018] Specifically, in step S1, the work area is rotated and scanned by a lidar or depth camera mounted on the stacker-reclaimer arm to obtain a real-time three-dimensional point cloud model; the work task parameters include at least the material type, target stack height, and work urgency level indicator.

[0019] Specifically, the adaptive selection logic in step S2 includes: if the task parameters indicate that the average height of the stockpile is lower than the first threshold, the efficiency priority mode is selected; if the task parameters indicate that the stockpile shape is arranged or the average height of the stockpile is higher than the second threshold, the regularization priority mode is selected, wherein the second threshold is greater than or equal to the first threshold.

[0020] Specifically, in step S3, when the operation is stacking, the real-time calculation includes: based on the selected mode and the three-dimensional shape of the current stockpile, if it is the efficiency priority mode, the first alignment algorithm is called, and if it is the regularity priority mode, the second alignment algorithm is called. The material falling motion is simulated through the corresponding alignment algorithm to determine the cantilever end falling focus that can optimize the current mode target of the new stockpile material as the optimal alignment point.

[0021] Specifically, the first alignment algorithm determines the material drop focus by minimizing the material roll-off distance and cantilever travel time; the second alignment algorithm determines the material drop focus by simulating the material drop sequence required to form a preset standard stack cross-section.

[0022] Specifically, in step S3, when the operation is material retrieving, the real-time calculation includes: identifying the slope and contour of the material retrieving surface based on the three-dimensional point cloud model, calling the third alignment algorithm, and dynamically calculating the starting point and cutting angle of the material retrieving device cutting into the material pile to achieve the maximum material retrieving efficiency and avoid the risk of material pile collapse, as the optimal alignment point.

[0023] In particular, the methods also include:

[0024] S5. Closed-loop calibration and collision avoidance control: During the operation, step S1 is executed cyclically to update the three-dimensional point cloud model, and the optimal alignment point and motion trajectory are adjusted in real time based on the updated model. At the same time, the distance between the cantilever, the material handling device and the material pile and surrounding facilities is monitored in the model. If it is less than the safety threshold, protective action is triggered.

[0025] A dual-mode stacking and automatic alignment control system for a stacker-reclaimer is used to implement a dual-mode stacking and automatic alignment control method for the stacker-reclaimer, including:

[0026] The three-dimensional perception subsystem, fixedly installed on the stacker-reclaimer cantilever, includes at least one lidar scanner and / or depth camera for periodically acquiring raw point cloud data of the work area;

[0027] The edge computing and control subsystem, deployed in the control room of the stacker-reclaimer, is connected to the 3D sensing subsystem via industrial Ethernet communication. It includes:

[0028] The data fusion and modeling module is used to filter and register the raw point cloud data to generate a real-time 3D point cloud model.

[0029] The mode decision module has embedded dual-mode switching logic, which is used to perform adaptive selection based on the 3D point cloud model and externally input task parameters;

[0030] The intelligent alignment calculation module has embedded a first alignment algorithm and a second alignment algorithm corresponding to the efficiency-first mode and the regularization-first mode, respectively, which are used to calculate the optimal alignment point based on the selected mode.

[0031] The trajectory planning and motion control module is used to generate drive commands for each mechanism based on the optimal alignment point and the kinematic model of the stacker-reclaimer.

[0032] The equipment drive subsystem includes a walking drive unit, a slewing drive unit, a pitch drive unit, and a material handling device drive unit, which receives drive commands and drives the corresponding mechanisms to move.

[0033] Specifically, it also includes a cloud-based collaborative optimization platform, with the edge computing and control subsystem connected to the cloud-based collaborative optimization platform via a wireless communication network; the cloud-based collaborative optimization platform is used for:

[0034] It integrates historical operation data, 3D models, and performance indicators from multiple devices;

[0035] Based on big data analysis, the switching logic in the mode decision module and the alignment algorithm parameters in the intelligent alignment calculation module are globally optimized.

[0036] The optimized algorithm model or parameter package is then distributed to the corresponding edge computing and control subsystem for updating.

[0037] Specifically, the intelligent alignment calculation module integrates a material property knowledge base, which stores parameters such as the angle of repose, density, and coefficient of friction corresponding to different material types. When the first and second alignment algorithms are solving the problem, they call the property parameters of the current material as physical constraints.

[0038] The beneficial effects of this invention are:

[0039] Adaptive intelligent decision-making is achieved: by introducing a dual mode of "efficiency priority" and "organization priority" and establishing a switching logic based on a three-dimensional model and task parameters, the stacker-reclaimer can adaptively select the optimal strategy according to different operation stages and objectives, fundamentally solving the contradiction that a single fixed mode cannot take into account both stacking speed and reclaiming efficiency.

[0040] High-precision dynamic alignment is achieved: abandoning the traditional fixed coordinate point alignment method, it perceives the geometry of the material pile through a real-time 3D point cloud model and calls dedicated algorithms (first, second, and third alignment algorithms) bound to the mode to dynamically calculate the optimal working point. This allows the alignment point to change in real time with the shape of the material pile, significantly improving the accuracy and adaptability of the operation, especially effectively handling irregular material piles.

[0041] A closed-loop safety control system was constructed: by continuously updating the model and adjusting the trajectory in real time during operation, a closed-loop control of "perception-decision-execution-re-perception" was formed, improving the system's robustness. Simultaneously, active collision avoidance monitoring based on the 3D model provides early warning and intervention for potential collision risks, enhancing system safety.

[0042] A cloud-edge-device collaborative architecture has been formed: the system adopts a layered architecture of "cloud optimization, edge control, and local execution". The edge side ensures real-time performance, while the cloud side realizes global data aggregation and algorithm iteration optimization, enabling the system to have both local rapid response and continuous evolutionary learning capabilities, and continuously improving its intelligence level.

[0043] It deeply integrates process knowledge: by integrating a material property knowledge base, it combines alignment calculation with the physical properties of materials (angle of repose, density, etc.), making intelligent control more in line with actual processes and improving the practicality and professionalism of the solution. Attached Figure Description

[0044] Figure 1 This is a flowchart of the method of the present invention;

[0045] Figure 2 This is a system architecture diagram of the present invention;

[0046] The following will describe in detail, with reference to the accompanying drawings, embodiments of the present invention. Detailed Implementation

[0047] The present invention will be further described below with reference to embodiments:

[0048] like Figure 1 As shown, a dual-mode stacking and automatic alignment control method for a stacker-reclaimer includes the following steps:

[0049] S1. Real-time perception and modeling: Obtain the real-time 3D point cloud model of the work area and the preset work task parameters; use a lidar or depth camera mounted on the stacker reclaimer arm to perform a rotational scan of the work area to obtain the real-time 3D point cloud model; the work task parameters include at least the material type, target stack height, and work urgency level indicator.

[0050] Specifically, this step is completed collaboratively by the data fusion and modeling modules of the 3D perception subsystem and the edge computing and control subsystem. Task parameters are typically transmitted from the higher-level production management system (such as a manufacturing execution system or scheduling system) to the edge computing and control subsystem via standard industrial communication protocols.

[0051] A lidar unit is rigidly mounted on the upper part of the stacker-reclaimer's cantilever, overlooking the work area. This lidar can perform a 360-degree rotating scan at a frequency of 10Hz, generating approximately 600,000 points per second, forming raw point cloud data covering a radius exceeding 100 meters. The scan data is transmitted in real-time via armored industrial Ethernet cables to the edge computing and control subsystem cabinet located behind the operator's cab.

[0052] After receiving the raw point cloud, the data fusion and modeling module (which can be hosted by a high-performance industrial control computer) executes the following processing pipeline in sequence:

[0053] Filtering: A statistical outlier removal algorithm is used to filter out noise points caused by dust, rain, and snow.

[0054] Registration and coordinate transformation: Using real-time encoder data from the stacker-reclaimer's walking, rotating, and pitching mechanisms, each frame of point cloud data is registered from the radar coordinate system to the global world coordinate system based on the yard ground through a coordinate transformation matrix.

[0055] Fusion and Generation: The point clouds registered from multiple consecutive frames are fused together, and a dense, real-time 3D point cloud digital model of the working area is generated using triangulation or voxel meshing algorithms. This model can accurately reflect the height, slope, volume, and surrounding environment of the material pile.

[0056] Meanwhile, the system receives the task parameter message for this operation from the upper-level manufacturing execution system, and obtains the operation task parameters after parsing, namely the material type (such as "iron ore"), target stack height (such as "15m"), and operation urgency level indicator (such as "normal").

[0057] This invention utilizes a cantilevered lidar to perform global, continuous, and high-precision three-dimensional scanning, acquiring complete spatial morphological information of the material pile, rather than just the state of a few points. This provides a data foundation far exceeding traditional methods for subsequent intelligent decision-making and precise alignment, solving the problems of low perception dimensionality and incomplete information in existing technologies.

[0058] S2. Intelligent Mode Decision-Making: Based on the 3D point cloud model and task parameters, it adaptively selects between an efficiency-first mode and a regularity-first mode; among which...

[0059] The efficiency-first model aims to maximize the material throughput per unit time.

[0060] The regularity priority mode aims to optimize the formation of a regular stockpile profile that facilitates subsequent full-section material extraction.

[0061] The adaptive selection logic includes: if the task parameters indicate that the average height of the stockpile is lower than the first threshold, the efficiency priority mode is selected; if the task parameters indicate that the stockpile shape is arranged or the average height of the stockpile is higher than the second threshold, the regularization priority mode is selected, where the second threshold is greater than or equal to the first threshold.

[0062] Specifically, this step is executed by the mode decision module. The module has pre-defined dual-mode switching logic, which can be represented as a decision function:

[0063] Input: Real-time 3D point cloud model, task parameters.

[0064] Processing: Extract the point cloud of the current target stack area from the point cloud model and calculate its average height H. 平均 (For example, by averaging the Z-coordinates of all points in the region, H is calculated in this case.) 平均 =8 meters).

[0065] Read the job identifier (“General”) and target stack height (H) from the task parameters. 目标 =15 meters).

[0066] Applying decision rules:

[0067] Rule 1 (Efficiency Trigger): If the job identifier is "Initialization" or H 平均 < (H 目标 If the distance is 50% × 7.5 meters, then select the "efficiency priority" mode.

[0068] Rule 2 (Rectification Trigger): If the job is identified as "Rectification" or H 平均 > (H 目标 If the value is 80% (12 meters), then select the "Orderliness Priority" mode.

[0069] Rule 3 (Default and Manual Intervention): If neither is satisfied, the efficiency-first mode will be selected by default, but the interface for manually switching modes via HMI (Human-Machine Interface) will be retained.

[0070] Output and transmission: According to rule 1, due to H 平均 =8 meters > 7.5 meters, and is marked as "normal", the system decides to use the efficiency-first mode. This decision result (mode flag) is sent to the intelligent alignment calculation module in real time.

[0071] This invention introduces a dual-mode adaptive decision-making mechanism, completely changing the single, sequential, and fixed process described in the prior art. It can intelligently switch between two strategies, "rapid filling" and "fine shaping," based on the actual progress of the operation (pile height) and macroscopic requirements (task identifier). This fundamentally solves the core contradiction in existing technologies that cannot simultaneously balance the efficiency of the piling process with the final quality of the pile shape (for material retrieval), achieving optimization of the global operation chain.

[0072] S3. Dynamic alignment calculation: Based on the selected target mode, the alignment algorithm corresponding to the mode is called to analyze the three-dimensional point cloud model, calculate and output an optimal alignment point that matches the current dynamic geometric features of the stockpile and the operation target in real time.

[0073] When the operation is material stacking, the real-time calculation includes: based on the selected mode and the three-dimensional shape of the current material stack, if it is the efficiency priority mode, the first alignment algorithm is called, and if it is the regularity priority mode, the second alignment algorithm is called. The material falling motion is simulated through the corresponding alignment algorithm to determine the cantilever end falling focus that can optimize the current mode target of the new material stack as the optimal alignment point.

[0074] The first alignment algorithm determines the material drop focus by minimizing the material roll-off distance and the cantilever travel time; the second alignment algorithm determines the material drop focus by simulating the material drop sequence required to form a preset standard stack cross-section.

[0075] When the operation is material retrieving, the real-time calculation includes: identifying the slope and contour of the material retrieving surface based on the 3D point cloud model, calling the third alignment algorithm, and dynamically calculating the starting point and cutting angle of the material retrieving device cutting into the material pile to achieve the maximum material retrieving efficiency and avoid the risk of material pile collapse, as the optimal alignment point.

[0076] Specifically, this step is performed by the intelligent alignment calculation module. The module encapsulates three core algorithms and integrates a material property knowledge base (such as storing parameters like the angle of repose of iron ore being 35° and its bulk density being 2.5 tons / cubic meter).

[0077] Scenario A: Material stacking operation (efficiency-first mode);

[0078] Algorithm call: Receives the "efficiency priority mode" instruction and calls the first pairing algorithm.

[0079] Algorithm Execution: The algorithm takes the current 3D point cloud model as input. Its core optimization objective function is to minimize (material roll-off horizontal distance + K × cantilever idle travel time), where K is a weighting coefficient. The algorithm workflow is as follows:

[0080] a. Define a candidate material drop area near the current top surface of the stockpile.

[0081] b. Using the angle of repose parameter of iron ore, simulate the natural roll-off and accumulation process of material after falling from the end of the cantilever into each candidate point (using a simplified discrete element or geometric projection model).

[0082] c. Evaluate the additional stockpile volume (efficiency) and the path length (time) of the cantilever moving from its current position to that point for each candidate point.

[0083] d. Find the optimal candidate point that minimizes the objective function value through optimization search (such as gradient descent or heuristic search).

[0084] Output: Output a three-dimensional spatial coordinate P1(x1, y1, z1) as the optimal alignment point, which ensures the maximum accumulation of material per unit time.

[0085] Scenario B: Material stacking operation (organization priority mode);

[0086] Algorithm call: If it is a regularization mode, the second alignment algorithm is called.

[0087] Algorithm Execution: The primary goal of this algorithm is to make the final stockpile cross-section approximate the preset "trapezoidal shape". Its workflow is as follows:

[0088] a. Extract the contour line C of a key cross section from the current 3D model. 当前 .

[0089] b. Place C 当前 With the target trapezoidal profile C 目标 Compare and calculate the contour difference map.

[0090] c. Analyze the difference map to identify the recessed areas that most need to be "filled" or the raised areas that need to be "trimmed".

[0091] d. Plan a sequential drop point. The goal of the current cycle is to calculate the first drop point in the sequence that most effectively reduces contour differences. The algorithm will simulate the new contour formed after drop to ensure it is closer to C. 目标 .

[0092] Output: Output an optimal alignment point P2(x2, y2, z2) for "shaping".

[0093] Scenario C: Material handling operation algorithm call: Call the third alignment algorithm.

[0094] Algorithm execution: This algorithm takes into account both material handling efficiency and safety.

[0095] a. Analyze the point cloud of the material intake surface, fit its slope equation, and calculate the slope angle.

[0096] b. Calculate the maximum allowable cutting depth to maintain stability, taking into account the angle of repose (35°) of the iron ore.

[0097] c. Within the safe depth range, with the goal of "maximizing the volume of material obtained in a single cut", optimize the calculation of the optimal starting point and cutting angle of the bucket wheel or scraper.

[0098] Output: Output the alignment point P3(x3, y3, z3) and the angle of inclination α.

[0099] This invention pioneers dynamic alignment based on real-time 3D models and dedicated optimization algorithms. Three algorithms are implemented for different operational objectives and scenarios, respectively:

[0100] The intelligence of alignment: The alignment point is not fixed, but calculated based on the current state of the material pile.

[0101] Precision of alignment: Combining the physical properties of materials makes the control more in line with natural laws.

[0102] The purpose of alignment: in efficiency mode, it seeks "speed"; in regularization mode, it seeks "shape"; and in material handling mode, it seeks "stability and quantity", thus realizing true task-oriented intelligent control.

[0103] The above descriptions of the first, second, and third alignment algorithms focus on their core logic, input-output relationships, and technical effects defined to solve specific industrial control problems (efficiency, regularity, and safe material handling). Those skilled in the art should understand that the aforementioned steps of 'simulation,' 'evaluation,' 'optimization search,' and 'contour comparison' can be implemented using various computer program methods known in the field (e.g., projection methods based on geometric calculations, simplified models using the discrete element method, gradient descent methods, contour difference algorithms in image processing, etc.). This invention does not limit the specific software coding implementation of these steps; any program module capable of processing input data and outputting the results according to the described logic is an equivalent implementation of the technical solution of this invention.

[0104] S4. Trajectory Planning and Execution: Based on the optimal alignment point, plan the motion trajectory of the stacker-reclaimer boom and control it to perform stacking or reclaiming operations.

[0105] Specifically, this step is completed by the trajectory planning and motion control module and the device drive subsystem.

[0106] Trajectory planning: The module receives the optimal alignment point P1(x1, y1, z1). It embeds the kinematic model of the stacker-reclaimer. The planning process is as follows:

[0107] P1 is solved inversely as the target distance D of the traveling mechanism, the target angle θ of the slewing mechanism, and the target elevation angle φ of the pitching mechanism.

[0108] Considering the maximum speed and acceleration constraints of each motor, a smooth motion trajectory with optimal time is generated (usually using S-curve acceleration and deceleration planning), and decomposed into a series of position and speed command sequences with time steps.

[0109] Command Issuance and Execution: The command sequence is sent to the device drive subsystem via a real-time Ethernet bus. The servo drives of the traveling, rotating, pitching, and material handling devices receive the commands, precisely control the synchronous movement of the motors, drive the cantilever to move smoothly and efficiently to the alignment point, and start the belt conveyor or material handling device to perform the operation.

[0110] S5. Closed-loop calibration and collision avoidance control: During the operation, step S1 is executed cyclically to update the three-dimensional point cloud model, and the optimal alignment point and motion trajectory are adjusted in real time based on the updated model. At the same time, the distance between the cantilever, the material handling device and the material pile and surrounding facilities is monitored in the model. If it is less than the safety threshold, protective action is triggered.

[0111] Specifically, this step forms a closed loop that runs throughout the entire operation.

[0112] Closed-loop calibration: During the cantilever movement, the 3D sensing subsystem continuously scans (10Hz). The data fusion and modeling module continuously updates the model. If the updated model shows that the actual stockpile shape differs significantly from the expected shape due to material landing point deviations, the system will automatically trigger the recalculation in step S3 to fine-tune the next alignment point, achieving a closed-loop correction of "sensing-execution-re-sensing".

[0113] Active collision avoidance:

[0114] The real-time 3D point cloud model not only includes the material pile, but also incorporates the static point cloud of surrounding fixed facilities (such as transfer station corridors and other equipment) through pre-modeling or scanning.

[0115] The system calculates in real time the minimum distance d between the moving parts of the stacker-reclaimer, such as the boom and bucket wheel (generated in real time based on the kinematic model) and the stockpile and static facilities. 最小 .

[0116] Set three levels of safety thresholds: warning threshold (e.g., 2 meters), deceleration threshold (e.g., 1 meter), and emergency stop threshold (e.g., 0.5 meters).

[0117] When d 最小 When the system enters the deceleration threshold, it automatically sends instructions to the trajectory planning and motion control module to reduce the speed of each mechanism. If the system enters the emergency stop threshold, it immediately issues an emergency stop signal, stops all drives, and issues an audible and visual alarm, thus achieving proactive and preventative safety protection based on three-dimensional spatial perception.

[0118] like Figure 2 As shown, the stacker-reclaimer dual-mode stacking and automatic alignment control system is used to realize the stacker-reclaimer dual-mode stacking and automatic alignment control method, including:

[0119] The three-dimensional perception subsystem, fixedly installed on the stacker-reclaimer cantilever, includes at least one lidar scanner and / or depth camera for periodically acquiring raw point cloud data of the work area;

[0120] The edge computing and control subsystem, deployed in the control room of the stacker-reclaimer, is connected to the 3D sensing subsystem via industrial Ethernet communication. It includes:

[0121] The data fusion and modeling module is used to filter and register the raw point cloud data to generate a real-time 3D point cloud model.

[0122] The mode decision module has embedded dual-mode switching logic, which is used to perform adaptive selection based on the 3D point cloud model and externally input task parameters;

[0123] The intelligent alignment calculation module has embedded a first alignment algorithm and a second alignment algorithm corresponding to the efficiency-first mode and the regularity-first mode, respectively, which are used to solve the optimal alignment point based on the selected mode. The intelligent alignment calculation module integrates a material characteristic knowledge base, which stores the angle of repose, density, and friction coefficient parameters corresponding to different material types. When solving, the first alignment algorithm and the second alignment algorithm call the characteristic parameters of the current material as physical constraints.

[0124] The trajectory planning and motion control module is used to generate drive commands for each mechanism based on the optimal alignment point and the kinematic model of the stacker-reclaimer.

[0125] The data fusion and modeling module interacts with the pattern decision module and the intelligent alignment calculation module through shared memory or message queues; the intelligent alignment calculation module sends the optimal alignment point to the trajectory planning and motion control module, and the trajectory planning and motion control module sends the standard drive instructions generated by the optimal alignment point to each servo drive in the device drive subsystem through the real-time industrial Ethernet bus.

[0126] The equipment drive subsystem includes a walking drive unit, a slewing drive unit, a pitch drive unit, and a material handling device drive unit, which receives drive commands and drives the corresponding mechanisms to move.

[0127] It also includes a cloud-based collaborative optimization platform, with the edge computing and control subsystem connected to the cloud-based collaborative optimization platform via a wireless communication network; the cloud-based collaborative optimization platform is used for:

[0128] It integrates historical operation data, 3D models, and performance indicators from multiple devices;

[0129] Based on big data analysis, the switching logic in the mode decision module and the alignment algorithm parameters in the intelligent alignment calculation module are globally optimized.

[0130] The optimized algorithm model or parameter package is then distributed to the corresponding edge computing and control subsystem for updating.

[0131] The platform regularly (e.g., daily) collects operational data from all stacker-reclaimers (mode switching records, alignment point trajectories, final stack shape quality, operation cycle, etc.). Through big data analysis, optimization points that might otherwise go unnoticed by humans can be identified. For example, historical data analysis reveals that when ambient humidity exceeds 80%, the actual effective angle of repose of iron ore materials decreases by 2°-3° compared to the standard value. Based on this, the platform automatically generates correction coefficients for regularization mode algorithm parameters under high humidity conditions and pushes the updates. The platform automatically trains the updated algorithm model, and after testing and verification, it can be deployed to the corresponding edge computing and control subsystem for one-click updates. This gives the entire system the ability to continuously learn and evolve globally, which is unmatched by stand-alone automated systems.

[0132] This system is a typical "cloud-edge-device" collaborative architecture.

[0133] End-side (equipment layer): It consists of a 3D perception subsystem (LiDAR) and an equipment drive subsystem (each mechanism drive unit), and is responsible for the signal acquisition and action execution at the lowest level.

[0134] Edge (Control Layer): The edge computing and control subsystem is the core, deployed in an industrial control cabinet next to the equipment, undertaking all real-time processing, decision-making, and control tasks. It communicates with edge devices at high speed via industrial Ethernet to ensure the real-time performance and determinism of control.

[0135] Cloud-side (Optimization Layer): The cloud-based collaborative optimization platform connects to edge subsystems on multiple stacker-reclaimers via wireless networks (such as 5G). It does not participate in real-time control but periodically aggregates operational data from each device. Utilizing big data and machine learning algorithms, it analyzes which mode-switching logic and algorithm parameters deliver superior overall performance under different operating conditions. It continuously sends optimized "strategy packages" or "model parameters" to the edge side for updates, enabling the entire system to continuously learn and evolve.

[0136] This invention constructs a highly intelligent, adaptive, precise, and safe stacker-reclaimer control system through dual-mode decision-making, three-dimensional dynamic alignment, and cloud-edge-device collaboration.

[0137] The present invention has been described above by way of example. Obviously, the specific implementation of the present invention is not limited to the above-described manner. Any improvements made by adopting the inventive concept and technical solution of the present invention, or direct application to other occasions without modification, are all within the protection scope of the present invention.

Claims

1. A dual-mode stacking and automatic alignment control method for a stacker-reclaimer, characterized in that, Includes the following steps: S1. Real-time perception and modeling: Obtain the real-time 3D point cloud model of the work area and the preset work task parameters; S2. Intelligent Mode Decision-Making: Based on the 3D point cloud model and task parameters, it adaptively selects between an efficiency-first mode and a regularity-first mode; among which... The efficiency-first model aims to maximize the material throughput per unit time. The regularity priority mode aims to optimize the formation of a regular stockpile profile that facilitates subsequent full-section material extraction. S3. Dynamic alignment calculation: Based on the selected target mode, the alignment algorithm corresponding to the mode is called to analyze the three-dimensional point cloud model, calculate and output an optimal alignment point that matches the current dynamic geometric features of the stockpile and the operation target in real time. S4. Trajectory Planning and Execution: Based on the optimal alignment point, plan the motion trajectory of the stacker-reclaimer boom and control it to perform stacking or reclaiming operations.

2. The dual-mode stacking and automatic alignment control method for a stacker-reclaimer according to claim 1, characterized in that, In step S1, the work area is rotated and scanned by a lidar or depth camera mounted on the stacker-reclaimer arm to obtain a real-time three-dimensional point cloud model; the work task parameters include at least the material type, target stack height, and work urgency level indicator.

3. The dual-mode stacking and automatic alignment control method for a stacker-reclaimer according to claim 1, characterized in that, The adaptive selection logic in step S2 includes: if the task parameters indicate that the average height of the stockpile is lower than the first threshold, the efficiency priority mode is selected; if the task parameters indicate that the stockpile shape is arranged or the average height of the stockpile is higher than the second threshold, the regularization priority mode is selected, wherein the second threshold is greater than or equal to the first threshold.

4. The dual-mode stacking and automatic alignment control method for a stacker-reclaimer according to claim 1, characterized in that, In step S3, when the operation is stacking, the real-time calculation includes: based on the selected mode and the three-dimensional shape of the current stockpile, if it is the efficiency priority mode, the first alignment algorithm is called, and if it is the regularity priority mode, the second alignment algorithm is called. The material falling motion is simulated through the corresponding alignment algorithm to determine the cantilever end falling focus that can optimize the current mode target of the new stockpile material as the optimal alignment point.

5. A dual-mode stacking and automatic alignment control method for a stacker-reclaimer according to claim 4, characterized in that, The first alignment algorithm determines the material drop focus by minimizing the material roll-off distance and the cantilever travel time; the second alignment algorithm determines the material drop focus by simulating the material drop sequence required to form a preset standard stack cross-section.

6. The dual-mode stacking and automatic alignment control method for a stacker-reclaimer according to claim 1, characterized in that, In step S3, when the operation is material retrieving, the real-time calculation includes: identifying the slope and contour of the material retrieving surface based on the three-dimensional point cloud model, calling the third alignment algorithm, and dynamically calculating the starting point and cutting angle of the material retrieving device cutting into the material pile to achieve the maximum material retrieving efficiency and avoid the risk of material pile collapse, as the optimal alignment point.

7. The dual-mode stacking and automatic alignment control method for a stacker-reclaimer according to claim 1, characterized in that, The method also includes: S5. Closed-loop calibration and collision avoidance control: During the operation, step S1 is executed cyclically to update the three-dimensional point cloud model, and the optimal alignment point and motion trajectory are adjusted in real time based on the updated model. At the same time, the distance between the cantilever, the material handling device and the material pile and surrounding facilities is monitored in the model. If it is less than the safety threshold, protective action is triggered.

8. A stacker-reclaimer dual-mode stacking and automatic alignment control system, used to implement the stacker-reclaimer dual-mode stacking and automatic alignment control method according to any one of claims 1-7, characterized in that, include: The three-dimensional perception subsystem, fixedly installed on the stacker-reclaimer cantilever, includes at least one lidar scanner and / or depth camera for periodically acquiring raw point cloud data of the work area; The edge computing and control subsystem, deployed in the control room of the stacker-reclaimer, is connected to the 3D sensing subsystem via industrial Ethernet communication. It includes: The data fusion and modeling module is used to filter and register the raw point cloud data to generate a real-time 3D point cloud model. The mode decision module has embedded dual-mode switching logic, which is used to perform adaptive selection based on the 3D point cloud model and externally input task parameters; The intelligent alignment calculation module has embedded a first alignment algorithm and a second alignment algorithm corresponding to the efficiency-first mode and the regularization-first mode, respectively, which are used to calculate the optimal alignment point based on the selected mode. The trajectory planning and motion control module is used to generate drive commands for each mechanism based on the optimal alignment point and the kinematic model of the stacker-reclaimer. The equipment drive subsystem includes a walking drive unit, a slewing drive unit, a pitch drive unit, and a material handling device drive unit, which receives drive commands and drives the corresponding mechanisms to move.

9. The stacker-reclaimer dual-mode stacking and automatic alignment control system according to claim 8, characterized in that, It also includes a cloud-based collaborative optimization platform, with the edge computing and control subsystem connected to the cloud-based collaborative optimization platform via a wireless communication network; The cloud-based collaborative optimization platform is used for: It integrates historical operation data, 3D models, and performance indicators from multiple devices; Based on big data analysis, the switching logic in the mode decision module and the alignment algorithm parameters in the intelligent alignment calculation module are globally optimized. The optimized algorithm model or parameter package is then distributed to the corresponding edge computing and control subsystem for updating.

10. The stacker-reclaimer dual-mode stacking and automatic alignment control system according to claim 8, characterized in that, The intelligent alignment calculation module integrates a material property knowledge base, which stores parameters such as the angle of repose, density, and coefficient of friction corresponding to different material types. When solving the problem, the first alignment algorithm and the second alignment algorithm call the property parameters of the current material as physical constraints.

Images