Gantry crane remote control platform access system and method

Abstract

The invention relates to the field of automatic crane control, in particular to a gantry crane remote control platform access system and method, and the system comprises a task prediction module, a task distribution module, a task execution module and an exception handling module. The task prediction module generates task prediction data by collecting the task pool and the equipment state data; the task allocation module dynamically generates a task allocation scheme by using a reinforcement learning optimization algorithm and a simulated annealing algorithm; the task execution module generates control path data through real-time data acquisition and multi-modal data fusion, and controls equipment to execute tasks; and the exception handling module predicts a fault through a time sequence model and implements a task relay and dynamic recovery mechanism. According to the method, the efficiency of task scheduling, the precision of a control path and the stability of system operation are improved, and the execution efficiency of a remote control task is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of automated crane control, and in particular to a gantry crane remote control platform access system and method. Background Art

[0002] The gantry crane remote control platform access system is a key technology for automated operation of modern container yards. As one of the core equipment in the yard, the gantry crane undertakes the main tasks of container loading, unloading and stacking. The traditional local manual operation mode is not only limited by the physical range of equipment distribution, but also has problems such as low efficiency, safety hazards and difficulty in ensuring real-time control. The introduction of remote control technology has made it possible to improve the efficiency of yard operations. However, the remote control system needs to achieve efficient scheduling and seamless docking of equipment and tasks in a complex environment, which puts higher requirements on system architecture and data interaction.

[0003] The existing container yard automated rail crane remote control system (China invention patent, publication number: CN118164371A, name: A container yard automated rail crane remote control system and method) mainly relies on the collaborative work of multiple subsystems (such as ROCS, BMS, RCCS). This approach has the following defects: Existing methods require ROCS to interact with multiple systems such as BMS and ACCS, resulting in a complex and lack of uniformity in the architecture, which is prone to inconsistent data interaction. Since the data link is long and relies on multiple layers of transit, once a problem occurs, it is difficult to quickly locate the source of the fault, affecting the stability and real-time response capabilities of the system. The design of independent systems requires a large amount of repetitive resources, such as separate databases and monitoring interfaces, which not only increases deployment costs, but also reduces the overall collaboration efficiency of the system. Summary of the invention

[0004] In view of the many problems existing in the above-mentioned prior art, the present invention provides a gantry crane remote control platform access system and method. The present invention realizes efficient scheduling and execution of remote control tasks through the collaborative work of four modules: task prediction, task allocation, task execution and exception handling. The core principles include the global optimization capability of the reinforcement learning algorithm, the path control accuracy of multimodal data fusion, and the task continuity guarantee of the dynamic exception recovery mechanism. Its effect is significantly reflected in improving equipment operation efficiency, reducing operation errors and enhancing system robustness.

[0005] A gantry crane remote control platform access system for realizing intelligent task scheduling and execution, including: The task prediction module is used to collect task status data and automated gantry crane equipment status data in the task pool, standardize the task status data, and generate task prediction data; The task allocation module is used to generate a task allocation plan based on the task prediction data and the equipment status data, dynamically adjust the task allocation plan using the reinforcement learning optimization algorithm, and generate a backup task allocation plan based on the equipment status data during the task execution process; A task execution module is used to bind the task to the automated gantry crane device based on the task allocation plan, collect the real-time video data and real-time status data of the automated gantry crane device, and generate control path data, and send a control signal to the automated gantry crane device according to the control path data to execute the task; The exception handling module is used to predict exceptions based on task execution feedback data and equipment status data. When an abnormality is detected in the automated gantry crane equipment, tasks are reallocated through a collaborative recovery mechanism and an optimized task allocation plan is generated.

[0006] Preferably, the task prediction module maps the task priority to a fixed interval by standardizing the task status data, and generates task prediction data in a data format adapted to the task allocation module.

[0007] Preferably, the task allocation module generates a task allocation plan through a reinforcement learning optimization algorithm, wherein the reward function defined by the reinforcement learning optimization algorithm is:

[0008] in, Assign reward values ​​to tasks; The time for task completion; is the standard deviation of the equipment load; Prioritize the tasks; , , They are adjustable weight coefficients, which are used to balance task completion efficiency, device load balancing, and task priority optimization.

[0009] Preferably, the task assignment module combines a simulated annealing algorithm to locally optimize the task assignment scheme generated by reinforcement learning, wherein the simulated annealing algorithm generates a new task assignment state by randomly perturbing the task assignment scheme, and calculates the acceptance probability based on the following formula:

[0010] in, is the probability of acceptance; is the performance difference between the new state and the current state; It is the temperature parameter in the simulated annealing algorithm, which is used to control the state jump probability during the search process.

[0011] Preferably, the task execution module collects real-time video data and real-time status data of the automated gantry crane equipment, and generates control path data based on multimodal data fusion, wherein the control path data includes the motion trajectory of the equipment and the grasping action planning.

[0012] Preferably, the task execution module adjusts the control signal sent to the automated gantry crane equipment in real time through a signal smoothing mechanism. The signal smoothing mechanism reduces high-frequency noise through a filtering algorithm and optimizes the amplitude change of the control signal to reduce the motion error during the operation of the equipment.

[0013] Preferably, the exception handling module analyzes the task execution feedback data and the real-time status data of the automated gantry crane equipment and uses a time series model to predict the type of equipment failure and the time range of the failure.

[0014] Preferably, the exception handling module reallocates the tasks of the abnormal devices through a collaborative recovery mechanism, and gives priority to selecting spare devices with lower loads to take over the tasks. The control parameters of the devices after the task relay are synchronously updated in real time.

[0015] Preferably, after completing the task relay, the exception handling module updates the task priority queue and the device status monitoring parameters by dynamically adjusting the task scheduling logic.

[0016] A method for executing the gantry crane remote control platform access system comprises the following steps: Collect task status data in the task pool and equipment status data of the automated gantry crane equipment, where the task status data includes task type and task priority, and the equipment status data includes equipment location and equipment load information; perform standardization on the task status data to generate task prediction data; Generate task allocation plans based on task prediction data and equipment status data, and dynamically adjust task allocation plans using reinforcement learning optimization algorithms. The optimization goals include minimizing task completion time, balancing equipment loads, and giving priority to high-priority tasks. During task execution, generate backup task allocation plans based on real-time collected equipment status data. Bind the task to the automated gantry crane device based on the task allocation plan, and collect the real-time video data and real-time status data of the automated gantry crane device; generate control path data by fusing the real-time video data and the real-time status data, and send a control signal to the automated gantry crane device according to the control path data to execute the task; Anomalies are predicted based on task execution feedback data and equipment status data to detect equipment operation anomalies. When an anomaly is detected, tasks are reallocated and an optimized task allocation plan is generated through a collaborative recovery mechanism.

[0017] Compared with the prior art, the advantages and beneficial effects of the present invention are: The present invention adopts a reinforcement learning optimization algorithm and a simulated annealing local optimization algorithm to dynamically generate a task allocation scheme, optimize task completion time and equipment load balancing, and solve the problems of low efficiency and poor adaptability of task allocation in the prior art. This solution uses multimodal data fusion to generate accurate control paths, reducing motion errors during equipment operation; The present invention dynamically adjusts the task scheduling logic and preferentially selects backup equipment to perform task relay, which significantly improves the task continuity and response efficiency of the system in the event of a failure. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a structural block diagram of the system of the present invention; Figure 2 A logical flow diagram of the task allocation module in the present invention; Figure 3 This is a schematic diagram of the control data flow of the task execution module in the present invention; Figure 4 Schematic diagram of the recovery mechanism of the exception handling module in the present invention; Figure 5 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0019] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is obvious that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0020] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0021] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0022] like Figure 1As shown, a gantry crane remote control platform access system is used to realize intelligent task scheduling and execution, including: The task prediction module is used to collect task status data and automated gantry crane equipment status data in the task pool, standardize the task status data, and generate task prediction data; Preferably, the task prediction module maps the task priority to a fixed interval by standardizing the task status data, and generates task prediction data in a data format adapted to the task allocation module.

[0023] The core principle of the task prediction module lies in the collection, processing and conversion of data. Specifically, the module first extracts task status data from the task pool, including task type (such as cargo loading and unloading tasks or equipment maintenance tasks) and task priority (reflecting the urgency of task processing). At the same time, equipment status data is collected from the automated gantry crane equipment, including the current location, load condition and operating status of the equipment. These data are uploaded to the task prediction module in real time through the sensor network and stored in the module's short-term memory for further processing.

[0024] The task prediction module standardizes the task status data to ensure that the data is consistent and comparable in different allocation scenarios. The main means of standardization is to normalize the task priority data and map it to a fixed interval (for example, [0,1]). This process uses a linear transformation formula to effectively compare the urgency of tasks in a multi-tasking environment. The standardized data is not only convenient for the task allocation module to use directly, but also adjusts the data format to be consistent with the data interface of subsequent modules, thereby improving the transmission efficiency of the overall data flow of the system.

[0025] In order to achieve efficient operation of the task allocation module, the task prediction module performs structured processing on the standardized data and generates a data format suitable for the task allocation module, such as a table or vector form. For the task type field, the task type can be converted into a discrete numerical expression through a classification coding method (for example, "loading and unloading task" is coded as 1, and "maintenance task" is coded as 2), so as to facilitate subsequent computer processing.

[0026] Through the design and implementation of the task prediction module, the gantry crane remote control platform access system can accurately and efficiently complete the preprocessing of task status data, providing reliable data support for subsequent task allocation. Data standardization and format conversion greatly reduce the time for the task allocation module to perform compatibility processing on the original data, enabling the task allocation plan to be generated more quickly. The normalization of task priorities enables the system to accurately judge the relative importance of tasks in multi-task competition scenarios and prioritize key tasks. The standardized task prediction data generated by the task prediction module can adapt to the needs of different allocation algorithms, providing compatibility for the subsequent introduction of new task allocation strategies.

[0027] In this example, a cargo loading and unloading task is used as an example: Assume that there are three tasks in the task pool, task A is a normal loading and unloading task, task B is an emergency equipment maintenance task, and task C is a regular inspection task. The task prediction module first collects the task types ("loading and unloading", "maintenance", and "inspection") and priorities (3, 5, and 2, respectively) of these three tasks. At the same time, the status data of two automated gantry cranes are collected, including the current position (10, 15), load (80%), and operating status (normal) of device 1, and the current position (20, 25), load (60%), and operating status (standby) of device 2.

[0028] The task prediction module normalizes the task priorities and maps priorities 3, 5, and 2 to the interval [0,1], obtaining 0.5, 1.0, and 0.3, respectively. Then, the task types are classified and coded, with "loading and unloading" coded as 1, "maintenance" coded as 2, and "inspection" coded as 3. The processed data is structured in the following table:

[0029] This table, together with the equipment status data, is passed to the task allocation module to support efficient matching of task priorities and equipment loads, and ultimately generate the optimal task allocation plan.

[0030] Task allocation module, such as Figure 2 As shown, it is used to generate a task allocation plan based on task prediction data and equipment status data, dynamically adjust the task allocation plan using a reinforcement learning optimization algorithm, and generate a backup task allocation plan based on the equipment status data during task execution; The workflow of the task allocation module includes three main steps: task allocation plan generation, plan optimization, and backup plan generation. First, the task prediction data (such as task type, priority, etc.) and equipment status data (such as equipment location, load information, etc.) output by the task prediction module are input into the task allocation module. The module uses heuristic rules to make preliminary allocations based on the requirements of the task and the current status of the equipment. For example, high-priority tasks are allocated to equipment with lower loads and closer to the task target point. This preliminary allocation process is implemented through logical rules and can be customized according to specific business scenarios.

[0031] Subsequently, the task allocation module dynamically adjusts the initially generated task allocation scheme through the reinforcement learning optimization algorithm. Reinforcement learning is an adaptive decision-making algorithm that evaluates the pros and cons of the allocation scheme by defining a reward function. In this scheme, the reward function usually includes the following goals: minimizing task completion time, balancing device load, and improving the completion efficiency of high-priority tasks. The module optimizes the task allocation scheme through multiple rounds of reinforcement learning algorithms, adjusts the matching relationship between tasks and devices, and improves the overall performance of the system. For example, during the optimization process, if it is found that a device has a high load due to too many task assignments, the system will reallocate some tasks to other devices with lower loads to achieve dynamic balance.

[0032] During the task execution phase, the module dynamically generates backup task allocation plans by monitoring device status data in real time (such as the device's operating status and load changes). When an abnormality is detected in a device (such as a failure or response delay), the module can immediately activate the backup plan and transfer the unfinished tasks of the device to other normally operating devices. This process uses the real-time update of device status data combined with the online adjustment capabilities of the reinforcement learning model to ensure the continuity and reliability of the task execution process.

[0033] The task allocation module comprehensively considers multiple factors such as task priority, equipment load and location, making task allocation more reasonable, reducing equipment idle time and task completion time, and improving resource utilization. The introduction of reinforcement learning optimization algorithm enables the system to have the ability to adaptively adjust task allocation. It can dynamically adjust the allocation plan according to real-time data during operation, significantly improving the operating efficiency and flexibility of the system. The generation and real-time activation mechanism of the backup task allocation plan ensures the continuity of task execution in the event of equipment failure or abnormality, reduces the task failure rate, and provides high reliability guarantee for system operation.

[0034] In an embodiment, in a port cargo loading and unloading scenario, the task pool contains three tasks: Task A is an emergency unloading task with a priority of 5, Task B is a common unloading task with a priority of 3, and Task C is an equipment maintenance task with a priority of 2. The equipment status data includes that the position of the automated gantry crane equipment 1 is (10, 15) and the load is 50%; the position of equipment 2 is (20, 25) and the load is 70%; the position of equipment 3 is (30, 35) and the load is 20%.

[0035] The initial allocation plan may assign Task A to Device 1, Task B to Device 2, and Task C to Device 3. Subsequently, the task allocation module optimizes the allocation plan based on the reinforcement learning algorithm: because the load of Device 2 is higher and Task B has a lower priority, the system adjusts to assign Task B to Device 3 with a lower load, and postpones the execution of Task C. This adjustment makes the device load more balanced and gives priority to the completion of high-priority tasks.

[0036] During the task execution process, if device 1 experiences a short-term abnormality, the module will immediately transfer task A to device 2 for execution through the backup task allocation plan, and dynamically update the device control instructions to ensure that task execution is not interrupted.

[0037] The task allocation module realizes efficient, dynamic and robust task allocation through three functions: task allocation plan generation, reinforcement learning optimization and backup plan generation. The module not only improves the efficiency of task completion and resource utilization, but also enhances the adaptability and reliability of the system. It is an important part of the gantry crane remote control platform access system to ensure task execution.

[0038] Preferably, the task allocation module generates a task allocation plan through a reinforcement learning optimization algorithm, wherein the reward function defined by the reinforcement learning optimization algorithm is:

[0039] in, The reward value for the task allocation plan. The larger the reward value, the better the allocation plan. The task completion time. The smaller the value, the faster the task is completed and the higher the corresponding reward value. is the standard deviation of the device load, which reflects the degree of load balancing among all devices. The smaller the standard deviation, the more balanced the device load is, and the higher the reward value is; The higher the priority of the task, the higher the reward for the task, ensuring that the high-priority task is executed first; , , They are adjustable weight coefficients, which are used to balance task completion efficiency, device load balancing, and task priority optimization.

[0040] The reinforcement learning optimization algorithm dynamically optimizes the matching scheme of tasks and devices based on the current state and historical feedback of task allocation through an interactive learning mechanism. In this scheme, the reward function is the core indicator for evaluating the pros and cons of the task allocation strategy, and the size of the reward value directly affects the generation and optimization direction of the allocation scheme.

[0041] During the allocation process, the reinforcement learning algorithm generates an initial task allocation plan based on the current state and calculates the corresponding reward value. It adjusts the allocation strategy through an iterative optimization process to gradually increase the reward value, thereby improving the overall superiority of the task allocation plan.

[0042] To optimize task allocation, this module takes task prediction data and equipment status data as input and executes through the following steps: 1. State modeling: Mapping the attributes of tasks (such as type, priority) and equipment (such as location, load) to the state space of the reinforcement learning algorithm.

[0043] 2. Initial assignment: Generate an initial task assignment plan based on simple rules, such as assigning high-priority tasks to the nearest and lightly loaded devices.

[0044] 3. Reward calculation: Calculate the reward value based on the completion time, load balance and task priority of the initial allocation plan .

[0045] 4. Iterative optimization: Use the policy gradient method of reinforcement learning to gradually optimize the allocation plan by updating the matching relationship between tasks and devices, so that the reward value continues to increase.

[0046] 5. Dynamic adjustment: During the task execution process, the device status data (such as load changes and failures) is monitored in real time, the allocation strategy is recalculated based on the reward function, and a backup plan is generated to deal with emergencies.

[0047] The reward function comprehensively considers the task completion time, device load balance and task priority, optimizing task allocation while ensuring efficiency, making the system run more smoothly. Through the continuous iteration of the reinforcement learning algorithm, the system can dynamically adjust the allocation strategy to adapt to the real-time changing device status and task requirements. Task priority parameter in the reward function The introduction of ensures that resources can be allocated first to emergency tasks, improving the system's ability to respond to high-priority tasks. The module generates backup task allocation plans to ensure the continuity of task execution in the event of equipment anomalies or task interruptions, significantly improving the reliability of the system.

[0048] In the example, it is assumed that there are three tasks in the task pool: Task A is an emergency cargo loading and unloading task with a priority of 5, Task B is a general cargo transportation task with a priority of 3, and Task C is a regular equipment inspection task with a priority of 1. The status of the automated gantry crane equipment is: Equipment 1 is at (10, 15) with a load of 50%; Equipment 2 is at (20, 25) with a load of 70%; Equipment 3 is at (30, 35) with a load of 20%.

[0049] 1. Initial state modeling: Map the task and device attributes to the state space of reinforcement learning and construct an initial allocation scheme: Task A is assigned to device 1, Task B is assigned to device 2, and Task C is assigned to device 3.

[0050] 2. Reward calculation: Calculate the task completion time based on the initial allocation plan , Equipment load balancing and task priorities , calculate the reward value .

[0051] 3. Optimization and adjustment: Since the load of device 2 is high and task B has a low priority, the reinforcement learning algorithm adjusts the allocation plan and transfers task B to device 3 for execution. At the same time, task A is reallocated to device 1, further improving the load balance of the equipment and the efficiency of completing urgent tasks.

[0052] 4. Real-time adjustment: During the task execution process, if a load overload alarm occurs on device 1, the module immediately activates the backup plan, transfers task A to device 3, and dynamically updates the task allocation strategy to ensure uninterrupted task execution.

[0053] The task allocation module achieves efficient generation and dynamic adjustment of task allocation schemes through the introduction of reinforcement learning optimization algorithm and reward function. The multi-objective optimization mechanism of reward function enables the system to balance the requirements of task completion time, equipment load balancing and task priority, providing efficient, intelligent and reliable task allocation capabilities for the gantry crane remote control platform access system.

[0054] Preferably, the task assignment module combines a simulated annealing algorithm to locally optimize the task assignment scheme generated by reinforcement learning, wherein the simulated annealing algorithm generates a new task assignment state by randomly perturbing the task assignment scheme, and calculates the acceptance probability based on the following formula:

[0055] in, is the probability of acceptance; is the performance difference between the new state and the current state; It is the temperature parameter in the simulated annealing algorithm, which is used to control the state jump probability during the search process.

[0056] The simulated annealing algorithm is a stochastic optimization algorithm based on the physical annealing process. Its core idea is to explore the solution space with high randomness in the initial state, and gradually reduce the randomness as the search process progresses, so as to gradually converge to the global optimal solution. In this scheme, the simulated annealing algorithm optimizes the initial task allocation scheme generated by reinforcement learning, generates a new allocation state through random perturbations, and determines whether to adopt the new state based on the acceptance probability function.

[0057] Specifically, the task allocation module first generates an initial task allocation scheme based on the reinforcement learning algorithm, which takes into account factors such as task priority, device load balancing, and task completion time. However, since the optimization process of the reinforcement learning algorithm in a complex solution space may be limited by the initial conditions, the generated scheme may only be a local optimum. To further optimize the allocation scheme, the module introduces a simulated annealing algorithm for local adjustment.

[0058] The optimization process of the simulated annealing algorithm in the present invention includes the following key steps: 1. Random perturbation: Based on the current task allocation plan, a new task allocation state is generated by randomly changing the allocation relationship between tasks and devices. For example, task A is reallocated from device 1 to device 2 to form a new allocation plan.

[0059] 2. Performance evaluation: Calculate the performance difference between the new allocation state and the current state, and use the performance difference Indicates that The performance indicators can comprehensively consider factors such as task completion time, load balancing and task priority.

[0060] 3. Acceptance probability calculation: Calculate the probability of the new state being accepted according to the following formula:

[0061] 4. State jump: If , that is, the new state is better than the current state, then directly accept the new state; if , that is, the new state performance is poor, then The probability of accepting the new state can prevent the algorithm from falling into the local optimum.

[0062] 5. Temperature update: As the number of optimization iterations increases, gradually reduce the temperature parameter , reduce the frequency of random jumps and make the allocation scheme converge to a stable state.

[0063] Through the above steps, the simulated annealing algorithm achieves global optimization of the task allocation scheme and improves the robustness of the allocation results to a certain extent.

[0064] By combining the simulated annealing algorithm, the task allocation module shows significant advantages in optimization performance and allocation stability: the introduction of random perturbations and acceptance probabilities enables the system to jump out of the local optimum and explore a better allocation scheme, thereby significantly improving the global performance of task allocation. The dynamic adjustment of temperature parameters enables the algorithm to explore the solution space extensively in the early stage and gradually converge to a stable solution in the later stage, which not only ensures the exploration ability but also takes into account the stability of the solution. The simulated annealing algorithm comprehensively considers the task completion time, equipment load balance and task priority during the optimization process, making the optimized allocation scheme more efficient and reasonable. When uneven load or sudden tasks occur during task execution, the module can quickly adjust the task allocation strategy through the backup allocation scheme generated by simulated annealing optimization to ensure the continuity and stability of system operation.

[0065] In a port scenario, the task pool contains three tasks: Task A is an emergency cargo loading and unloading task with a priority of 5, Task B is a general cargo transportation task with a priority of 3, and Task C is an equipment maintenance task with a priority of 1. The initial task allocation scheme is generated by a reinforcement learning algorithm, with Task A allocated to Equipment 1, Task B allocated to Equipment 2, and Task C allocated to Equipment 3. However, since the load of Equipment 2 is close to the upper limit, the initial scheme performs poorly in terms of load balancing.

[0066] Through the simulated annealing algorithm, the task allocation module is optimized: 1. Random perturbation: reallocate task B from device 2 to device 3 to generate a new allocation state; 2. Performance evaluation: Calculate new load balancing indicators and obtain performance differences This indicates that the new state is better than the current state in terms of load balancing; 3. Acceptance probability calculation: based on temperature parameters Calculate the acceptance probability and get , a higher acceptance probability makes the system tend to choose the new state; 4. State jump: accept the new state and update the task allocation plan; 5. Temperature update: lower the temperature parameters To reduce randomness in subsequent iterations.

[0067] After the optimization is completed, task A is still assigned to device 1, but task B and task C are both assigned to device 3. The overall load balancing and allocation efficiency are significantly improved.

[0068] The task allocation module combines the simulated annealing algorithm to locally optimize the allocation scheme generated by reinforcement learning through random perturbations and state jump mechanisms, avoiding the local optimal problem and further improving the global performance of the allocation scheme. Through reasonable parameter settings and dynamic adjustments, the module demonstrates excellent performance in terms of efficiency, adaptability and stability of task allocation, and is a key technology for realizing intelligent task allocation.

[0069] Task execution modules, such as Figure 3 As shown, it is used to bind the task to the automated gantry crane device based on the task allocation scheme, collect the real-time video data of the automated gantry crane device and the real-time status data of the automated gantry crane device, and use them to generate control path data, and send a control signal to the automated gantry crane device according to the control path data to execute the task; The working process of the task execution module includes three main links: task binding, real-time data collection, path planning and control signal generation. First, the module receives the task allocation plan, which clarifies the matching relationship between tasks and equipment. According to the content of the plan, the module binds the task to the specified automated gantry crane equipment, for example, assigning cargo loading and unloading task A to equipment 1.

[0070] During the task execution phase, the module uses the built-in sensors and cameras of the automated gantry crane equipment to collect real-time equipment operation status data (such as equipment location, load conditions, movement speed, etc.) and video data (such as equipment surroundings, cargo placement, etc.). These data are transmitted to the task execution module through the communication network and processed locally.

[0071] The task execution module integrates and analyzes real-time video data and operating status data to generate control path data. The path data describes the trajectory and action sequence that the equipment must follow during task execution, including the equipment's moving path, the lifting position of the sling, and the specific parameters of the grabbing action. During the path planning process, the module comprehensively considers the task objectives, equipment status, and operating environment, and generates the optimal control path through the path optimization algorithm. For example, in the task of loading and unloading cargo, the module will plan the equipment's moving path based on the initial position and target position of the cargo to ensure the accuracy of the grabbing and placing operations.

[0072] In the control signal generation phase, the module converts the equipment's operating parameters into executable control signals based on the generated control path data. These signals are transmitted to the automated gantry crane equipment through industrial communication protocols (such as Ethernet / IP), commanding the equipment to complete operations such as moving, grabbing, and placing according to the path data. Throughout the process, the module will continue to collect real-time status data and dynamically adjust the control signal based on the actual feedback from the equipment, thereby improving the accuracy and stability of task execution.

[0073] The task execution module generates high-precision control path data by integrating and analyzing real-time video data and operating status data, ensuring that the equipment can complete the task along the optimal path. The ability to dynamically adjust the control signal enables the module to quickly adapt to changes in equipment status, reduce equipment idle time, and improve resource utilization efficiency. During task execution, the module can monitor the equipment status in real time and adjust the operation strategy to ensure that the task is not interrupted due to environmental changes or equipment deviations. The module uses a real-time feedback mechanism to correct equipment operations, effectively reducing the risk of misoperation and improving the overall stability and robustness of system operation.

[0074] In the embodiment, it is assumed that the allocation scheme generated by the task allocation module is: Task A (cargo loading and unloading task, priority 5) is bound to the automated gantry crane device 1. The task execution module starts the task execution process according to the scheme content.

[0075] 1. Task binding: The module matches the target cargo information of task A with the operating parameters of device 1 and initializes the task execution process.

[0076] 2. Real-time data acquisition: The camera of device 1 collects real-time image data of the goods, including the shape, position and posture of the goods; at the same time, the device status sensor collects the operating parameters of device 1, such as position (10, 15), load (60%) and speed (1.5 m / s).

[0077] 3. Path planning: The task execution module generates the optimal path data based on the real-time location of the cargo and the status data of device 1. For example, device 1 needs to move to the target cargo location (20, 25), lower the sling to the specified height, grab the cargo, and move to the target unloading point (30, 35).

[0078] 4. Control signal generation and dynamic adjustment: The module generates control signals based on the path data to command device 1 to move along the planned path. At the same time, the module monitors the real-time feedback signals of device 1, such as position deviation and speed change, and corrects the operating parameters of the device in real time through a dynamic adjustment mechanism to ensure accurate execution of the path.

[0079] Finally, device 1 successfully completes the cargo loading and unloading tasks according to the planned path, and returns the task completion status to the task execution module to update the system's task pool and device status.

[0080] The task execution module achieves efficient execution of the task allocation plan through four key links: task binding, real-time data collection, path planning, and control signal generation. The module's real-time feedback mechanism and dynamic adjustment capabilities ensure the accuracy of equipment operation and the stability of task completion, providing strong support for the overall operation of the gantry crane remote control platform access system. Through efficient resource management and precise task execution, the task execution module plays an indispensable role in the system.

[0081] Preferably, the task execution module collects real-time video data and real-time status data of the automated gantry crane equipment, and generates control path data based on multimodal data fusion, wherein the control path data includes the motion trajectory of the equipment and the grasping action planning.

[0082] The core of the task execution module is to jointly analyze and process data from different sensors and camera devices through multimodal data fusion technology. Multimodal data includes real-time video data of automated gantry crane equipment (capturing the surrounding environment of the equipment, the position and posture of the target object, etc.) and real-time status data (such as the current position, speed, load condition, and hoisting device status of the equipment). These data have different physical meanings and formats. Through fusion analysis, they can provide a comprehensive decision-making basis for task execution.

[0083] The cameras and status sensors on the automated gantry crane equipment transmit real-time video data and real-time status data to the task execution module through the industrial communication network. The video data provides visual information about the equipment's operating environment, including the shape of the goods, their placement, and the location of surrounding obstacles; the status data includes the equipment's operating parameters, such as its current position, movement speed, and equipment load. These data are processed synchronously in the task execution module.

[0084] The core of multimodal data fusion is to jointly model and extract features from data from different sources to generate control path data for the device. Specifically, the module extracts key features (such as the bounding box, center of gravity position, and rotation angle of the target object) from the video data through computer vision algorithms (such as convolutional neural networks, CNN), and combines the device status data (such as current speed and position information) to establish a unified environmental model. The fusion process usually adopts a weighted fusion strategy or a decision-level fusion method to ensure that the information from different data sources complements each other and avoids the limitations of a single data.

[0085] The task execution module generates control path data through a path planning algorithm based on the multi-modal fused environment model. The control path data mainly includes the motion trajectory of the device and the grasping action planning: Equipment motion trajectory: describes the optimal path of the gantry crane from the current position to the target position, including moving direction, speed adjustment and stop position.

[0086] Grab action planning: define the lifting position, grab force, release time and other operating parameters of the spreader to ensure safe and smooth loading and unloading of goods.

[0087] During the process of path data generation and task execution, the task execution module monitors the device feedback in real time and dynamically adjusts the path planning. For example, when new obstacles appear in the environment, the module can recalculate the obstacle avoidance path; when the load status of the device changes, the movement speed and grasping force are adjusted to adapt to the new operation requirements.

[0088] Through the multimodal data fusion and path planning function of the task execution module. The module generates accurate equipment paths and grasping action plans through joint analysis of real-time video and status data to ensure the accuracy of cargo loading and unloading operations, especially for complex operating environments and high-priority tasks. Multimodal data fusion greatly improves the accuracy and reliability of path planning, reduces errors and interruptions in task execution, thereby improving the overall utilization of equipment and task completion efficiency. The dynamic adjustment mechanism enables the module to flexibly respond to environmental changes. For example, when obstacles suddenly appear or the equipment status is abnormal, new path planning can be quickly generated to ensure task continuity and safety. Through real-time monitoring and adjustment, the module reduces the risk of equipment operation errors and cargo damage, and enhances the robustness and stability of system operation.

[0089] In the embodiment, taking the port cargo loading and unloading scenario as an example, the task execution module binds the loading and unloading task A to device 1 according to the allocation plan generated by the task allocation module. The module starts to collect data in real time: the camera captures the real-time position (such as the coordinate position is (10,15)) and posture (rotation angle is 30°) of the cargo, and the device status sensor reports the current position ((5,10)), running speed (1.2 m / s) and load (80%) of the device.

[0090] 1. Multimodal data fusion: The module extracts the location and shape features of the goods through visual algorithms, integrates them with the real-time status data of the equipment, and generates a complete operating environment model.

[0091] 2. Path planning, based on the environment model, the module plans the movement trajectory of the equipment, guides the equipment 1 to move from the current position (5,10) to the target point (10,15), and designs the grasping path so that the sling approaches the cargo at the optimal angle (30°) and height (5 meters).

[0092] 3. Dynamic adjustment. During the execution of the path, a new obstacle (coordinate position (8,12)) is detected in the environment. The module immediately adjusts the device trajectory to avoid the obstacle and updates the path to the target point. At the same time, it adjusts the grasping angle to adapt to the existence of the obstacle.

[0093] Finally, equipment 1 successfully completed the task according to the updated path and loaded and unloaded the goods to the designated location. The whole process was smooth and efficient.

[0094] Preferably, the task execution module adjusts the control signal sent to the automated gantry crane equipment in real time through a signal smoothing mechanism. The signal smoothing mechanism reduces high-frequency noise through a filtering algorithm and optimizes the amplitude change of the control signal to reduce the motion error during the operation of the equipment.

[0095] The signal smoothing mechanism in the task execution module optimizes the device control signal in real time through a filtering algorithm. The control signal is usually a signal sequence that instructs the device to complete a specific action, including parameters such as displacement, speed, and direction. However, due to sensor errors, environmental interference, or data fluctuations, there may be high-frequency noise or uneven amplitude changes in the control signal, which may cause vibration, delay, or motion errors in the operation of the device.

[0096] The signal smoothing mechanism processes the control signal through a low-pass filter. The low-pass filter can effectively filter out high-frequency noise in the signal while retaining the low-frequency components in the signal (i.e. the main control instructions for the device movement). This process is implemented in real time by the computer to ensure that the noise is suppressed in time. For example, when a jitter signal appears in the device movement path instruction, the filter can weaken these invalid fluctuations to generate a smoother position signal.

[0097] In actual tasks, the device needs to perform multi-stage operations based on the control signal, such as quickly moving to the target area and then slowly grabbing the goods. In these dynamic processes, the amplitude change of the control signal may be too abrupt, causing drastic changes in the acceleration of the device. The signal smoothing mechanism optimizes the amplitude through interpolation algorithms or recursive smoothing methods, smoothing the sudden signal into a gradual signal. For example, when the signal switches from "stop" to "full speed", the module adjusts the signal's amplification rate so that the device accelerates gradually to avoid unstable device load due to excessive acceleration.

[0098] The module dynamically adjusts the control signal by collecting real-time feedback from the device's status (such as position, speed, and load). For example, when the deviation between the device's actual running trajectory and the target trajectory increases, the module will correct the output parameters of the control signal in real time to adjust the device back to the target trajectory. In addition, the system uses a recursive filtering strategy, which allows the filtering parameters to be adjusted in real time according to the device status, such as enhancing the degree of smoothness under high load conditions.

[0099] The task execution module significantly improves the system's execution capability in complex task scenarios through a signal smoothing mechanism. The signal smoothing mechanism makes the equipment move more smoothly by eliminating high-frequency noise and sudden signal changes, reducing equipment vibration and unnecessary errors caused by poor signals. The optimized signal amplitude change enables the equipment to run accurately according to the predetermined path and speed, especially in complex environments (such as when the goods are placed irregularly or there are many obstacles), which significantly improves the accuracy of task completion. Smooth control signals reduce the risk of mechanical stress and fatigue of the equipment under high load and high-speed operation conditions, extend the service life of the equipment, and reduce the possibility of cargo damage or equipment failure caused by sudden signal changes. The real-time adjustment mechanism ensures that the equipment can respond quickly to environmental and task changes while maintaining an efficient and stable operating state, improving the overall task execution efficiency of the system.

[0100] In an example, in a cargo loading and unloading task, the task execution module needs to move the automated gantry crane device from the current position (10, 15) to the target position (30, 35) and complete the action of grabbing the cargo. During the execution process, the operation of the device is affected by environmental interference (such as wind) and sensor data fluctuations, resulting in high-frequency jitter and amplitude mutations in the control signal.

[0101] 1. In the signal processing stage, the module processes the input signal through a low-pass filter, reducing the high-frequency jitter in the command, making the device's moving path smoother and avoiding multiple unnecessary adjustments before reaching the target position.

[0102] 2. During the amplitude optimization stage, when the equipment accelerates from a stationary state to a moving speed of 2 m / s, the signal smoothing mechanism will gradually increase the amplitude of the control signal from the initial value instead of directly jumping to the target value, ensuring that the equipment accelerates smoothly at a gradual speed and preventing the cargo from tilting due to excessive acceleration.

[0103] 3. In the dynamic adjustment stage, when the device approaches the target position, the status feedback shows that the device deviates from the predetermined path by about 0.3 meters. The module immediately adjusts the control signal to correct the deviation of the moving direction back to the target trajectory. At the same time, in order to reduce the shaking when grabbing the goods, the module enhances the signal smoothing parameters and reduces the descending speed of the spreader to make the grabbing action more precise.

[0104] Finally, the equipment reaches the target position smoothly and completes the cargo grabbing, with no significant vibration or position deviation during the entire process.

[0105] Exception handling modules, such as Figure 4 As shown, it is used to predict anomalies based on task execution feedback data and equipment status data. When an abnormality is detected in the automated gantry crane equipment, tasks are reallocated through a collaborative recovery mechanism and an optimized task allocation plan is generated.

[0106] The core of the exception handling module lies in the design and implementation of exception prediction and collaborative recovery mechanisms. The module captures potential anomalies in equipment operation in real time through multi-layer data monitoring and analysis, and quickly adjusts task allocation and execution strategies based on collaborative recovery mechanisms.

[0107] The exception handling module first monitors the task execution feedback data and device status data in real time. Task execution feedback data includes the device's task completion progress, path deviation, response delay, etc.; device status data includes the device's location, speed, load information, and operating status (such as whether overload or hardware failure occurs). These data are uploaded to the module through the device's sensors and communication network. The module uses a machine learning-based anomaly detection model to analyze the historical trends and current fluctuation characteristics of the data to determine whether the device is in an abnormal state. For example, when the device path deviation exceeds the set threshold, or the device load exceeds the safe range, the module will predict possible failures.

[0108] When the anomaly prediction model detects an abnormal event (such as the device cannot complete the current task, the device communication is interrupted, or the load exceeds the limit), the module will immediately trigger the collaborative recovery mechanism. The core of the collaborative recovery mechanism is to reallocate the affected tasks to ensure that the system's task execution is not interrupted. The recovery process includes the following steps: 1. Abnormal task marking: Mark the affected tasks as unfinished tasks and extract spare task allocation resources from the current task pool.

[0109] 2. Dynamic allocation optimization: Based on the remaining tasks and available equipment status, the optimized task allocation plan is regenerated. The optimization goals include giving priority to high-priority tasks, balancing equipment loads, and reducing task delays.

[0110] 3. Task migration and execution: Migrate the affected tasks from abnormal devices to normal devices and update the task execution plan. The module switches the task to the new device through real-time control signals and dynamically adjusts the path planning and task parameters.

[0111] The process of generating an optimized task allocation scheme combines the functions of the task prediction module and the task allocation module. The module uses a reinforcement learning optimization algorithm to quickly calculate new task allocation strategies while dynamically evaluating the current device status and task requirements. The generated allocation scheme prioritizes high-priority tasks while minimizing task delays caused by abnormal events.

[0112] Through real-time anomaly prediction and collaborative recovery mechanisms, the module can quickly respond to sudden anomalies in equipment operation, reduce the risk of system operation interruption, and improve the continuity of task execution. By optimizing the task allocation plan, the module quickly switches the affected tasks to normal equipment for execution, reducing task delay time and ensuring that the system remains efficient under abnormal conditions. The real-time monitoring and prediction functions enable the module to identify potential faults in advance, thereby reducing the impact of equipment failures on task execution, and providing decision-making basis for equipment maintenance and resource scheduling. In complex scenarios of multi-task parallelism and multi-device collaboration, the module can flexibly adjust the task allocation strategy to achieve dynamic resource scheduling, significantly improving the adaptability of the system.

[0113] Example, in a port loading and unloading scenario, the task pool contains task A (priority 5), task B (priority 3) and task C (priority 2). Task A and task B have been assigned to equipment 1 and equipment 2, and task C is assigned to equipment 3. During the task execution, equipment 2 triggers an exception due to overload, resulting in the inability to complete task B normally.

[0114] The exception handling module detects that the load of device 2 exceeds the limit (the load value exceeds 90%) through the device status data, and confirms that device 2 is abnormal in combination with the task execution feedback data (task B is not completed).

[0115] The module marks task B as an abnormal task and starts the collaborative recovery mechanism. The module extracts the current task pool and device status, re-evaluates the available resources, and finds that device 3 has a low load and is available.

[0116] The module regenerates the task allocation plan and assigns Task B to Device 3. It also adjusts the priority of Task C and updates its execution path to ensure that the completion of Task C is not affected.

[0117] The module switches the execution instructions of Task B from Device 2 to Device 3 through real-time control signals and dynamically adjusts the path planning parameters of Task B. Device 3 smoothly receives Task B and completes the operation as planned.

[0118] In the end, both Task B and Task C were successfully completed according to the optimization plan, the task delay time was controlled within the minimum range, and the system operation was not significantly affected.

[0119] Preferably, the exception handling module analyzes the task execution feedback data and the real-time status data of the automated gantry crane equipment and uses a time series model to predict the type of equipment failure and the time range of the failure.

[0120] The core of the exception handling module is the prediction mechanism based on the time series model. The time series model analyzes the time characteristics of the equipment operation data, extracts the trend, periodicity and random fluctuation characteristics of the data, and then predicts the possible fault type and time range of the equipment. This module has the following specific working principles in terms of data collection, feature extraction, model training and prediction: The module collects task execution feedback data and device status data in real time, including the device's operating speed, position, load, temperature, current and voltage fluctuations, and task completion time. This data is continuously uploaded through the sensor network and stored in a time series format. For example, the device load data is recorded as ,in is the timestamp, is the load value. The data preprocessing process includes outlier removal, noise filtering and missing data filling.

[0121] The module extracts key features from time series data, including long-term trends (such as an overall increase in device load), short-term fluctuations (such as rapid changes in load when tasks are executed), and periodic patterns (such as recurring operating states). For example, temperature data may show a periodic increase trend, which may be associated with a decrease in the device's heat dissipation performance.

[0122] The module uses forecasting models based on time series analysis, such as the Autoregressive Integrated Moving Average (ARIMA) model or the Long Short-Term Memory (LSTM) model. These models can capture the dynamic changes of data and infer future trends from historical data. For example, the LSTM model captures the long-term dependencies of equipment load data through its memory cells and predicts future load values.

[0123] The module compares the predicted data with the preset equipment operation safety threshold to identify possible fault types. For example, a load prediction value exceeding the equipment load limit may indicate a load overload fault; an abnormal temperature rise rate may indicate a fault in the equipment cooling system. In addition, the module uses historical fault records combined with classification models to identify the type of predicted fault (such as mechanical fault, electrical fault, etc.).

[0124] The module uses time series models to predict the time range in which a fault may occur. For example, by analyzing the load increase trend, the module predicts that a load overload may occur in the next 15 to 30 minutes, and generates an early warning message containing the fault type and time range.

[0125] By introducing the time series model, the exception handling module realizes the early prediction and risk assessment of equipment failures. The module uses multidimensional data analysis and time series modeling to identify subtle changes in the operating status of the equipment, achieve high-precision prediction of potential failures, and effectively reduce false alarms and missed reports. The time range of failure occurrence is predicted in advance, and sufficient time is reserved for task scheduling and equipment maintenance, which significantly shortens the failure response time and reduces the risk of system operation interruption. By predicting potential risks, the module coordinates the task allocation module to adjust the allocation plan in advance, reduces task delays or failures caused by equipment failures, and improves the continuity and reliability of task execution. The early warning function reduces the need for emergency maintenance and equipment downtime caused by failures, and reduces the cost of system operation and equipment maintenance.

[0126] In a cargo loading and unloading scenario, the exception handling module monitors the load data of equipment 1. Historical data shows that the load has a slow upward trend when high-intensity tasks are performed, and the short-term fluctuation is large. The module analyzes the load data through a time series model to identify the potential risk of overload of equipment 1.

[0127] The module collects the load data of device 1 in the recent task execution and finds that the load value gradually increases from 60% to 85%, accompanied by frequent small fluctuations. After preprocessing, the module standardizes the data into a time series format.

[0128] The module uses the LSTM model to predict the load trend. The results show that the load value will reach the 90% equipment safety threshold within the next 15 minutes and exceed the 100% limit within 20 minutes.

[0129] The module compares the prediction results with the equipment load safety threshold, determines that the equipment may experience a load overload failure within 15 to 20 minutes, and generates an early warning message to prompt the task allocation module to readjust the task.

[0130] According to the early warning information, the task allocation module reallocates the unfinished tasks of device 1 to device 2. At the same time, the exception handling module notifies the maintenance personnel to arrange load detection and maintenance of device 1.

[0131] Finally, when the load of device 1 was close to the limit, the failure did not actually occur and the system operation was not affected.

[0132] Preferably, the exception handling module reallocates the tasks of the abnormal devices through a collaborative recovery mechanism, and gives priority to selecting spare devices with lower loads to take over the tasks. The control parameters of the devices after the task relay are synchronously updated in real time.

[0133] The core of the collaborative recovery mechanism is to quickly identify abnormal equipment by monitoring equipment status and task execution, and complete the relay allocation and execution of tasks in the shortest time. The module workflow mainly includes the following steps: When a device abnormality occurs (such as device communication interruption, load overload, or hardware failure), the module marks the abnormal device as an unavailable device based on task execution feedback data and device status data, and extracts its unfinished tasks. Task information includes task type, priority, current execution progress, and generated control path data.

[0134] Based on the task information and the current operating status of the device pool, the module prioritizes the backup devices with lower load as the target devices for task relay. The device load is measured based on the real-time load ratio, the task queue length, and the location matching degree of the device. The core goal of task relay is to ensure the rapid continuation and smooth transition of tasks by minimizing task delays and uneven resource utilization.

[0135] After the task is relayed, the module synchronizes the task-related control parameters to the relay device in real time, including the device's movement path, task priority, execution progress, etc. To adapt to the operating characteristics of the relay device, the module will also dynamically adjust the control path data, such as replanning the movement trajectory to match the initial position of the relay device, or adjusting the operating speed to take into account the device load capacity.

[0136] During the task execution of the relay equipment, the module continuously monitors its status data and task completion progress, and dynamically adjusts the task execution parameters. For example, if the load of the relay equipment suddenly increases, the module can trigger the task adjustment mechanism again and further allocate part of the task to other backup equipment.

[0137] The module can quickly complete task relay after an abnormality occurs, avoiding task interruption caused by abnormal equipment downtime, and significantly improving the continuity of task execution. By giving priority to spare equipment with lower load, the module effectively avoids unbalanced task distribution and reduces the risk of equipment overload or resource waste. The dynamic control parameter synchronization and real-time adjustment mechanism enable the relay equipment to quickly enter the working state, reducing the delay in the task relay process. Through continuous feedback monitoring and dynamic adjustment, the module ensures that the relay equipment maintains operational stability while efficiently executing tasks, thereby improving the robustness of the entire system.

[0138] Example, suppose that in a port cargo loading and unloading task, the task pool contains task A (priority 5), task B (priority 3) and task C (priority 2). Task A is executed by device 1, and devices 2 and 3 are responsible for task B and task C respectively. During the task execution process, device 1 cannot continue to execute task A due to communication interruption.

[0139] The exception handling module detects the communication interruption of device 1, marks device 1 as an abnormal device, and extracts relevant information of task A, including task priority, completion progress (such as 50%), target location, etc.

[0140] The module selects backup devices with lower loads based on the operating status of the device pool. The monitoring results show that the current load of device 3 is 40%, the location is close to the target point of task A, and task C has a lower priority, so the module selects device 3 as the relay device for task A.

[0141] The module synchronizes the control path data (such as moving path and grasping position) of task A to device 3, and adjusts the path data according to the initial position and running speed of device 3. For example, the starting point of the moving path is changed to the current position of device 3, and the path curve is optimized to reduce the moving distance.

[0142] While device 3 is executing task A, the module monitors its status data (such as speed, load) and task completion progress in real time. When the load of device 3 is close to 80%, the module adjusts its execution speed to prevent overload and reserves execution time for task C.

[0143] Finally, device 3 successfully completed task A, and task C was also carried out smoothly as planned. During the entire process, task delays were kept to a minimum, and system operation was not significantly affected.

[0144] Preferably, after completing the task relay, the exception handling module updates the task priority queue and the device status monitoring parameters by dynamically adjusting the task scheduling logic.

[0145] After the task relay is completed, the core function of the exception handling module is to dynamically adjust the task scheduling logic and device status to restore and optimize the system's task execution capability. This process mainly includes the following key steps: 1. The module adjusts the task priority queue in real time based on task execution feedback and current task completion status. The reordering of task priorities is based on the following factors: Task completion status: Completed tasks are removed from the queue, and the priority of unfinished tasks is recalculated based on the remaining task time and importance.

[0146] Device status: Adjust the priority of task allocation based on the real-time load and operation status of the device. For example, tasks of high-load devices can be reassigned to low-load devices.

[0147] Compensation strategy for abnormal tasks: For tasks that are interrupted due to device abnormalities, their priority will be appropriately increased based on their impact scope and delay to reduce the impact of the overall task delay on the system.

[0148] 2. After the task relay is completed, the module updates the monitoring parameters of the equipment in real time, including equipment load, operating efficiency, task completion rate, etc. The update of these parameters enables the system to dynamically track the health status and task execution capabilities of the equipment, providing a reliable basis for subsequent task allocation.

[0149] 3. The module combines the updated task priority queue and device monitoring parameters to dynamically adjust the task scheduling strategy to ensure the rationality and efficiency of task allocation. For example: Assign higher-performance devices to high-priority tasks; use delayed scheduling for low-priority tasks to give priority to meeting the resource requirements of high-priority tasks; when resources are tight, achieve more efficient task execution by splitting or merging tasks.

[0150] The module monitors the results of task scheduling adjustments in real time, collects execution feedback, and further optimizes the scheduling logic. For example, when dynamic adjustments cause a device to be overloaded, the module will adjust the queue priority again or reallocate some tasks.

[0151] Dynamic priority adjustment ensures that key tasks are completed first, reduces task delays caused by abnormal devices, and optimizes the utilization of device resources. The module ensures load balancing of task allocation through real-time monitoring and adjustment, reducing the risk of overload or long-term idleness of a single device. The task scheduling logic adjustment combined with the real-time status of the device enables the system to quickly adapt to changes in the operating environment, especially in complex scenarios with multiple tasks in parallel and multiple devices collaborating. The dynamic update of priority and compensation strategy effectively reduce the negative impact of abnormal tasks on the overall system operation, ensuring the efficiency and stability of the system after abnormal recovery.

[0152] In this example, assume that the task pool contains task A (priority 5), task B (priority 3), and task C (priority 2). During the task execution, device 1 completes the relay operation of task A, and the exception handling module makes the following dynamic adjustments to the system: 1. The module removes Task A from the task queue and reorders the priorities of Task B and Task C based on the remaining time of the current task. For example, since Task B has a shorter remaining time, the module raises its priority from 3 to 4, while Task C remains at priority 2.

[0153] 2. The module updates the operating parameters of device 1 in real time, including the current load (from 80% to 50%), task completion rate (increased to 95%), and current operating efficiency (increased to 85%) after completing task A. The update of these parameters provides an important basis for subsequent task allocation.

[0154] 3. The module combines the updated task queue and device status to assign task B to device 1 first, and assign task C to device 2 with a lower load. During the task assignment process, the module dynamically adjusts the execution time of task C so that it does not affect other tasks of device 2.

[0155] 4. After adjusting the task allocation, the module detected that the load of device 1 increased slightly when executing task B, but the operating efficiency remained above 85%, so no further adjustment was made. At the same time, the module tracked the operation of device 2 in real time to ensure the smooth completion of task C.

[0156] Finally, through dynamic adjustment, both Task B and Task C were successfully completed, task delays were controlled within the minimum range, and the overall operating efficiency of the system was significantly improved.

[0157] like Figure 5 As shown, a method for executing the gantry crane remote control platform access system comprises the following steps: Collect task status data in the task pool and equipment status data of the automated gantry crane equipment, where the task status data includes task type and task priority, and the equipment status data includes equipment location and equipment load information; standardize the task status data to generate task prediction data; collect task status data (such as task type, task priority) and equipment status data (such as equipment location, equipment load information) in the task pool, generate task prediction data through data standardization, and ensure the rationality and operability of task allocation.

[0158] Generate task allocation plans based on task prediction data and equipment status data, and use reinforcement learning optimization algorithms to dynamically adjust task allocation plans. The optimization goals include minimizing task completion time, balancing equipment loads, and giving priority to high-priority tasks. During task execution, generate backup task allocation plans based on real-time collected equipment status data. Based on task prediction data and equipment status data, use reinforcement learning optimization algorithms to dynamically generate task allocation plans. The optimization goals include minimizing task completion time, balancing equipment loads, and giving priority to high-priority tasks. Collect equipment status data in real time to generate backup task allocation plans to ensure scheduling flexibility.

[0159] Based on the task allocation plan, the task is bound to the automated gantry crane equipment, and the real-time video data and real-time status data of the automated gantry crane equipment are collected; control path data is generated by fusing real-time video data and real-time status data, and control signals are sent to the automated gantry crane equipment according to the control path data to perform the task; control paths are generated by fusing real-time video data and equipment status data, the equipment operation trajectory is accurately planned, and control signals are sent to the equipment to perform the task.

[0160] Based on task execution feedback data and equipment status data, abnormalities are predicted to detect abnormal equipment operation; when abnormalities are detected, tasks are reallocated through a collaborative recovery mechanism and an optimized task allocation plan is generated. Based on task execution feedback data and equipment status data, equipment abnormalities are predicted, tasks are reallocated in real time through a collaborative recovery mechanism, and an optimized task allocation plan is generated to ensure task continuity and system stability.

[0161] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware.

[0162] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A gantry crane remote control platform access system for realizing intelligent task scheduling and execution, characterized in that: include: The task prediction module is used to collect task status data and automated gantry crane equipment status data in the task pool, standardize the task status data, and generate task prediction data; The task allocation module is used to generate a task allocation plan based on the task prediction data and the equipment status data, dynamically adjust the task allocation plan using the reinforcement learning optimization algorithm, and generate a backup task allocation plan based on the equipment status data during the task execution process; A task execution module is used to bind the task to the automated gantry crane device based on the task allocation plan, collect the real-time video data and real-time status data of the automated gantry crane device, and generate control path data, and send a control signal to the automated gantry crane device according to the control path data to execute the task; The exception handling module is used to predict exceptions based on task execution feedback data and equipment status data. When an abnormality is detected in the automated gantry crane equipment, tasks are reallocated through a collaborative recovery mechanism and an optimized task allocation plan is generated.

2. The gantry crane remote control platform access system according to claim 1, characterized in that: The task prediction module normalizes the task status data, maps the task priority to a fixed interval, and generates task prediction data in a data format that is suitable for the task allocation module.

3. The gantry crane remote control platform access system according to claim 1, characterized in that: The task allocation module generates a task allocation plan through the reinforcement learning optimization algorithm, where the reward function defined by the reinforcement learning optimization algorithm is: in, Assign reward values ​​to tasks; The time for task completion; is the standard deviation of the equipment load; Prioritize the tasks; , , They are adjustable weight coefficients, which are used to balance task completion efficiency, device load balancing, and task priority optimization.

4. The gantry crane remote control platform access system according to claim 1, characterized in that: The task assignment module combines the simulated annealing algorithm to locally optimize the task assignment scheme generated by reinforcement learning. The simulated annealing algorithm generates a new task assignment state by randomly perturbing the task assignment scheme and calculates the acceptance probability based on the following formula: in, is the probability of acceptance; is the performance difference between the new state and the current state; It is the temperature parameter in the simulated annealing algorithm, which is used to control the state jump probability during the search process.

5. The gantry crane remote control platform access system according to claim 1, characterized in that: The task execution module collects real-time video data and real-time status data of the automated gantry crane equipment, and generates control path data based on multimodal data fusion. The control path data includes the motion trajectory of the equipment and the grasping action planning.

6. The gantry crane remote control platform access system according to claim 1, characterized in that: The task execution module makes real-time adjustments to the control signal sent to the automated gantry crane equipment through a signal smoothing mechanism. The signal smoothing mechanism reduces high-frequency noise through a filtering algorithm and optimizes the amplitude change of the control signal to reduce the motion error during the operation of the equipment.

7. The gantry crane remote control platform access system according to claim 1, characterized in that: The exception handling module analyzes the task execution feedback data and the real-time status data of the automated gantry crane equipment, and uses the time series model to predict the type of equipment failure and the time range of the failure.

8. The gantry crane remote control platform access system according to claim 1, characterized in that: The exception handling module reallocates the tasks of abnormal devices through a collaborative recovery mechanism, and gives priority to backup devices with lower loads for task relay. The control parameters of the devices after task relay are updated synchronously in real time.

9. The gantry crane remote control platform access system according to claim 8, characterized in that: After completing the task relay, the exception handling module dynamically adjusts the task scheduling logic and updates the task priority queue and device status monitoring parameters.

10. A method for executing the gantry crane remote control platform access system according to any one of claims 1 to 9, characterized in that: The following steps are involved: Collect task status data in the task pool and equipment status data of the automated gantry crane equipment, where the task status data includes task type and task priority, and the equipment status data includes equipment location and equipment load information; perform standardization on the task status data to generate task prediction data; Generate task allocation plans based on task prediction data and equipment status data, and use reinforcement learning optimization algorithms to dynamically adjust task allocation plans. The optimization goals include minimizing task completion time, balancing equipment loads, and giving priority to high-priority tasks. During the task execution process, a backup task allocation plan is generated based on the real-time collected equipment status data; Bind the task to the automated gantry crane device based on the task allocation plan, and collect real-time video data and real-time status data of the automated gantry crane device; Generate control path data by fusing real-time video data and real-time status data, and send control signals to the automated gantry crane equipment to perform tasks according to the control path data; Anomalies are predicted based on task execution feedback data and equipment status data to detect equipment operation anomalies. When an anomaly is detected, tasks are reallocated and an optimized task allocation plan is generated through a collaborative recovery mechanism.

Images