A digital-twin-based intelligent port bulk cargo yard handling system

Abstract

The application relates to the technical field of smart ports, and particularly discloses a port bulk cargo yard intelligent loading system based on digital twinning, which comprises a task separation module, a coordination optimization module, an execution and modeling module and a deviation readjustment module, all operation plan data in an operation period are acquired and arranged into subtask units, dynamic resource states are acquired, global resource conflict coordination and operation optimization are performed on all subtask units in the operation period, global operation instruction sets and twinning data service orders are generated, automatic execution of the subtask units of the global operation instruction sets is performed, a dynamically updated three-dimensional digital twinning model is constructed based on the twinning data service orders, and self-adaptive deviation correction is performed, whether deviation early warning is triggered in the execution process is judged, if the deviation early warning is triggered, a dynamic regulation and control process is started, and therefore, full-process intelligent closed-loop management from planning, scheduling and execution to monitoring and adjustment is realized, and operation efficiency, safety and self-adaptive capacity are improved.

Description

Technical Field

[0001] This invention relates to the field of smart port technology, specifically to a smart cargo handling system for port bulk cargo yards based on digital twins. Background Technology

[0002] As a crucial hub connecting ship loading and unloading with land transportation, the level of intelligence in port tallying operations directly impacts the overall operational efficiency and safety of the port. Traditional tallying and operations management methods mainly rely on macro-level instructions from the Terminal Operating System (TOS) and manual on-site coordination. However, when faced with ever-increasing throughput and complex multi-ship parallel operations, a series of technical bottlenecks have gradually been exposed.

[0003] The existing technical solutions and their problems are mainly reflected in the following aspects: First, there is a disconnect between work plans and on-site execution. The work plans generated by existing TOS systems are mostly static instruction sequences based on experience and fixed rules, lacking accurate prediction and proactive coordination of dynamic resource conflicts within the yard. This leads to frequent instances of empty runs, waiting, and path conflicts for on-site equipment, making it difficult to improve the overall equipment utilization rate. Second, the process monitoring and status perception are shallow and coarse-grained. Current monitoring relies heavily on video surveillance and manual reports, which cannot obtain the precise pose and operational status of equipment, containers, and stacks in three-dimensional space in real time and accurately, making it difficult for the dispatch center to grasp the true overall operational situation. Third, the ability to cope with dynamic disturbances is weak. When equipment failures, work delays, or priority changes occur, existing systems lack rapid simulation and rescheduling capabilities, often relying on manual experience for local adjustments, leading to the accumulation of task delay costs.

[0004] In recent years, digital twin technology has provided new ideas for port digitalization. However, existing applications mostly focus on 3D visualization or individual equipment performance monitoring, failing to deeply integrate the twin model into the closed loop of operation scheduling and control. Model building is often a continuous, high-energy-consuming process across the entire scenario, lacking on-demand, dynamic, and refined data collection and modeling mechanisms, resulting in low data value density and wasted computing resources.

[0005] To address the aforementioned problems, this invention proposes an intelligent cargo handling system for port bulk cargo yards based on digital twins. Summary of the Invention

[0006] The purpose of this invention is to provide an intelligent cargo handling system for port bulk cargo yards based on digital twins, in order to solve the aforementioned background problems.

[0007] The objective of this invention can be achieved through the following technical solution: an intelligent cargo handling system for port bulk cargo yards based on digital twins, comprising:

[0008] Task separation module: Acquires all job plan data within the job cycle and organizes them into sub-task units;

[0009] Coordination and optimization module: acquires dynamic resource status, performs global resource conflict coordination and job optimization for all sub-task units within the job cycle, and generates a global job instruction set and twin data service order;

[0010] Execution and Modeling Module: Automates the execution of sub-task units of the global job instruction set, serves orders based on digital twin data, constructs a dynamically updated three-layer digital twin model, and performs adaptive correction;

[0011] Deviation readjustment module: During execution, it determines whether a deviation warning is triggered. If triggered, it initiates the dynamic control process.

[0012] Furthermore, the method for obtaining sub-task units is as follows:

[0013] The operation plan data of each planned vessel is initially decomposed, and the complete cargo loading and unloading instructions are decomposed into the minimum operation actions for a single container or a single bulk cargo stack. A multi-objective clustering algorithm is adopted, with the concentration of operation area and the continuity of equipment operation as the core optimization objectives. Based on the planned time window and planned operation path of the minimum operation action, all decomposed minimum operation actions are aggregated into multiple sub-task units that can be independently scheduled, monitored and settled under the constraints of business rules.

[0014] Furthermore, the planned time window and planned job path are obtained as follows:

[0015] The planned time window is based on the planned vessel and the estimated berthing time included in the operation plan data. It is estimated by calling the standard operation time of the minimum operation action estimated by the preset standard operation cycle experience library to obtain the planned time window of the minimum operation action. The planned operation path is obtained by querying the preset standard path topology network, extracting and matching feasible paths that comply with safety regulations and one-way traffic rules between the starting coordinates and the ending coordinates of the minimum operation action according to the shortest path principle.

[0016] Furthermore, the methods for global resource conflict coordination and job optimization are as follows:

[0017] Obtain the subtask unit list, dynamic resource status, and preset basic rule base for each subtask unit. Construct a conflict detection model that combines three-dimensional space and time dimensions to automatically detect conflicts within the job cycle. Employ a heuristic scheduling algorithm based on priority rules and conflict resolution. Using the current dynamic resource status as the initial condition, in the constructed fast simulation engine, optimize all subtask units based on the defined optimization objective function, including job order and resource allocation, to generate a comprehensive optimal global scheduling scheme and output a global job instruction set based on this scheme.

[0018] Furthermore, optimizing the objective function includes:

[0019] The optimization objective function is defined as a weighted combination of the normalized calculation results of the main objective function. The main objectives corresponding to the main objective function include minimizing the total operation time, maximizing the comprehensive utilization rate of equipment, minimizing the overall energy consumption, and minimizing the cost of task delay. The optimization variable set defining the optimization objective function includes the planned time window of each sub-task unit, the specific equipment number dynamically allocated to each sub-task unit, and the planned operation path of the equipment corresponding to the sub-task unit.

[0020] Furthermore, the method for generating twin data service orders is as follows:

[0021] The twin data service order is generated by reverse engineering based on the global job instruction set. This includes dynamically determining the dynamic update depth modeling range according to the planned time window and planned job path of each sub-task unit in the global job instruction set, specifying the sensing resource scheduling plan of the sensing devices for twin data collection for each sub-task unit, defining the twin data update frequency and triggering event frequency and event of twin data collection, and the twin data service timeliness obtained by extending the preset buffer time for both the forward and backward phases of the corresponding planned time window.

[0022] Furthermore, the three-layer digital twin model is constructed as follows:

[0023] The three-layer digital twin model includes a basic static layer model, a dynamic business topology layer model, and a high-fidelity operational surface layer model. Based on the digital twin data service order, the construction and updating of the three-layer digital twin model are driven as follows: A monthly updated port 3D geographic information model is loaded as the basic static layer model; the digital twin corresponding to the active sub-task unit is activated based on the digital twin data service timeline, establishing a real-time data link to drive the dynamic business topology layer model's status update; and sensing equipment is scheduled according to the sensing resource scheduling plan to perform high-precision data acquisition within the digital twin data service timeline, generating a high-fidelity operational surface layer model through a 3D reconstruction algorithm.

[0024] Furthermore, the adaptive correction method is as follows:

[0025] A dual monitoring mechanism is adopted, which combines real-time driving of dynamic business topology layer model and periodic verification of high-fidelity operation surface layer model. For the high-fidelity operation surface layer model updated during the execution of sub-task unit, it is automatically compared and analyzed with the high-fidelity operation surface layer model before the update to verify whether the actual position, posture and equipment gripping point of the cargo unit are consistent with the expected instructions. If a deviation is found, a pose fine-tuning instruction is generated and sent to the equipment control system for adaptive correction.

[0026] Furthermore, the method for determining whether a deviation warning has been triggered is as follows:

[0027] Based on the real-time data stream provided by the dynamically updated three-layer digital twin model, and taking the global operation instruction set and twin data service order as the benchmark, the real-time status comparison engine continuously compares and analyzes the twin data of the physical world based on the preset multi-level deviation warning rule library. If any rule in the multi-level deviation warning rule library is matched, a deviation warning is triggered.

[0028] Furthermore, the dynamic control process is as follows:

[0029] A rescheduling algorithm based on rolling time-domain simulation is adopted. The current moment is taken as the new optimization starting point, and the list of affected unfinished subtask units is taken as the main optimization object. The optimization objective function and basic rule base are reused, and the adjustment range minimization objective is added. Local iterative search is performed in the fast simulation engine to generate a dynamically adjusted instruction set and update the twin data service order. The dynamically adjusted instruction set is sent to the affected equipment control system to overwrite and correct the original instruction.

[0030] The beneficial effects of this invention are as follows:

[0031] 1. This invention achieves a leap from static planning to dynamic, executable solutions by intelligently decomposing and aggregating work plans into optimized sub-task units, and generating a global work instruction set based on conflict detection and simulation optimization. Its significant advantages lie in its ability to systematically avoid resource conflicts related to equipment, routes, and yard space, maximizing continuous equipment operation capabilities, significantly reducing empty runs and waiting times, thereby improving overall operational efficiency and resource utilization, and reducing vessel port time and overall operational energy consumption.

[0032] 2. This invention generates twin data service orders that are synchronized with the work instructions, driving the construction of a three-layer digital twin model. This achieves high-fidelity synchronization and closed-loop interaction between physical operations and digital mirrors. Its core benefits lie in providing real-time monitoring, verification, and adaptive correction capabilities for the work process with centimeter-level precision, saving data acquisition and computing resources, and enabling rapid and intelligent response to sudden deviations based on rolling time-domain rescheduling. This greatly enhances the transparency, security, and robustness of the system in dealing with dynamic disturbances during the sorting process. Attached Figure Description

[0033] The invention will now be further described with reference to the accompanying drawings.

[0034] Figure 1 This is a modular architecture diagram of a port bulk cargo yard intelligent tallying system based on digital twin, as described in an embodiment of the present invention.

[0035] Figure 2 This is a flowchart illustrating the specific steps of an intelligent cargo handling system for a port bulk cargo yard based on digital twins, as described in an embodiment of the present invention. Detailed Implementation

[0036] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0037] Example 1

[0038] Please see Figure 1 and Figure 2 As shown in the embodiment of the present invention, an intelligent cargo handling system for port bulk cargo yards based on digital twins aims to solve the problems of existing port yard cargo handling relying on manual labor, static and rigid planning leading to equipment conflicts and efficiency limitations, as well as opaque operation processes and weak ability to cope with sudden deviations. By acquiring operation plan data and decomposing it into sub-task units, and then using a conflict detection model and discrete event simulation optimization algorithm, global resource conflict coordination and operation optimization are performed to generate a global operation instruction set and twin data service orders. Subsequently, the sub-task units are driven to execute automatically, and a three-layer digital twin model—a basic static layer, a dynamic business topology layer, and a high-fidelity operation surface layer—is built and updated as needed to achieve dual monitoring and verification. Finally, real-time deviation judgment is performed. If a deviation warning is triggered, a dynamic control process based on rolling time-domain simulation is initiated, thereby realizing intelligent closed-loop management of the entire process from planning, scheduling, execution to monitoring and adjustment, improving operation efficiency, safety, and adaptability. Specifically, it includes the following modules:

[0039] Task separation module: Acquires all job plan data within the job cycle and organizes them into sub-task units;

[0040] Specifically, at the start of the work cycle, the work plan data of all planned vessels in the current work cycle is received uniformly through a standard data interface. The work plan data is a structured data set, including basic vessel information, berthing and departure plans, cargo details list, and work priority coefficients specified by the Terminal Operating System (TOS) or manually.

[0041] The basic information of the vessel includes the vessel name, voyage number, length, and draft. The berthing and departure plan includes the estimated berthing time, the planned departure time, and the designated berth. The cargo list details all cargo units to be loaded and unloaded in a list format. There are two types of cargo units: containerized cargo and bulk cargo. For containerized cargo, the cargo unit includes the container number, dimensions, container type, weight, and port of destination. For bulk cargo, the cargo unit includes the cargo type, bill of lading number, planned tonnage, and storage requirements.

[0042] The operation plan data of each planned vessel is initially decomposed, and the complete cargo loading and unloading instructions included in each operation plan data are broken down into the minimum operation actions for a single container or a single bulk cargo stack. A planned time window and planned operation path are estimated for each minimum operation action.

[0043] Specifically, the planned time window is estimated based on the planned operating vessel. The operation sequence is determined according to the cargo details list of the planned operating vessel and the basic operation logic between the minimum operation actions, forming the minimum operation action sequence of the planned operating vessel. The estimated berthing time is used as the time starting point of the minimum operation action sequence. The standard operation time of each minimum operation action is estimated by calling the preset standard operation cycle experience library, and the planned time window of each minimum operation action is estimated. The standard operation cycle experience library includes the action type, cargo attributes and standard equipment efficiency of the minimum operation action.

[0044] For each minimum operation, extract the starting point and ending point coordinates, query the preset standard path topology network, and match a feasible path between the starting point and ending point coordinates that complies with safety regulations and one-way traffic rules according to the shortest path principle, which is used as the planned operation path for the minimum operation.

[0045] It should be noted that the standard route topology network is a standardized road network model composed of nodes and directed edges, which is pre-constructed based on the geographic information model of port yards and berths. The planned operation route is composed of directed edges between nodes in the standard route topology network. The planned time window and the planned operation route are a set of estimates based on rules and average efficiency with a certain conservative margin, and are not unchangeable instructions.

[0046] Using a multi-objective clustering algorithm as the core technology, with the concentration of work areas and the continuity of equipment operation as the core optimization objectives, all the decomposed minimum work actions are aggregated into multiple sub-task units that can be independently scheduled, monitored and settled under the constraints of business rules.

[0047] For the concentration of work areas, the multi-objective clustering algorithm performs spatiotemporal proximity clustering. By calculating the spatial distance between the minimum work actions and the proximity of the planned time window, multiple minimum work actions that are spatially close and have consecutive planned time windows are packaged into a sub-task unit to reduce equipment idle running and cross interference. For the continuity of equipment operation, the multi-objective clustering algorithm performs equipment coordination clustering. By analyzing the planned operation path and equipment operation cycle, it identifies the minimum work actions that can maximize the continuous operation of a single equipment and are continuously executed by the same equipment in one uninterrupted operation cycle, and packages them into a sub-task unit.

[0048] For each subtask unit, compile a list of subtask units including subtask unit ID, preset planned time window, planned operation path, required equipment type and quantity, target yard location, and subtask unit type. The subtask unit type is divided into container type, bulk cargo type and mixed type according to cargo type.

[0049] It should be noted that business rule constraints represent the hard logical restrictions on safety, compliance and standard operating procedures that must be prioritized and strictly followed throughout the entire intelligent task processing process. Subtask units are the smallest logical units for resource allocation, path planning and efficiency calculation.

[0050] Coordination and optimization module: acquires dynamic resource status, performs global resource conflict coordination and job optimization for all sub-task units within the job cycle, and generates a global job instruction set and twin data service order;

[0051] Specifically, a conflict detection model combining three-dimensional space and time dimensions is constructed. The input of the conflict detection model is a list of sub-task units for each sub-task unit, dynamic resource status, and basic rule base. The dynamic resource status includes equipment resource pool, yard space status, and path network status. The equipment resource pool includes the real-time location, status, and rated capacity of various types of equipment, with three statuses: idle, in operation, and fault. The yard space status includes the real-time occupancy status and planned occupancy sequence of each yard partition and silo. The path network status includes the real-time traffic congestion status of each node and edge in the standard path topology network. The basic rule base is based on safety specifications and includes equipment safe operating distance, minimum time interval, and mutual exclusion rules for operating areas.

[0052] The conflict detection model automatically detects conflicts within a work cycle starting from the current moment by projecting and simulating each subtask unit list in a unified four-dimensional coordinate system.

[0053] For example, the conflict types include equipment time conflict, spatial path conflict and yard location conflict. Equipment time conflict is when the same equipment is assigned multiple sub-task units at the same time. Spatial path conflict is when the planned operation paths of different equipment are at the same node in the standard path topology network at the same time. Yard location conflict is when the same yard location is planned to store multiple cargo units at the same time.

[0054] Global resource conflict coordination is performed on detected conflicts, and job optimization, including job order and resource allocation, is performed on all sub-task units.

[0055] Specifically, taking the current dynamic resource status as the initial condition, a high-fidelity fast simulation engine is established. The optimization objective function is defined as a weighted combination of the normalized calculation results of the main objective function. The main objectives corresponding to the main objective function include minimizing the total operation time, maximizing the comprehensive utilization rate of equipment, minimizing the overall energy consumption, and minimizing the cost of task delay.

[0056] Among them, minimizing the total operation time minimizes the total completion time of all planned operation vessels, maximizing the comprehensive utilization of equipment balances the load of each piece of equipment and reduces the time when the equipment is idle, minimizing the overall energy consumption optimizes the planned operation path and start-up and shutdown of equipment to reduce the total energy consumption, and minimizing the task delay cost is based on the delay cost of the sub-task unit weighted by the operation priority coefficient to calculate the task delay cost by weighted aggregation, thereby reducing the delay of each planned operation vessel.

[0057] The set of optimization variables for the objective function includes the planned time window for each subtask unit, the specific equipment number dynamically assigned to each subtask unit, and the planned operation path of the equipment corresponding to the subtask unit. A heuristic scheduling algorithm based on priority rules and conflict resolution is adopted as the optimization algorithm, and iterative search is performed in the simulation environment.

[0058] In each iteration, the optimization algorithm generates a candidate set of optimization variables, drives the fast simulation engine to run, and calculates the result of the optimization objective function. After multiple iterations, the optimization algorithm converges to a comprehensive optimal set of optimization variables as the global scheduling scheme.

[0059] Based on the global scheduling scheme, output a global job instruction set and twin data service orders;

[0060] Specifically, the global operation instruction set is a structured, time-ordered list of instructions. Each instruction in the global operation instruction set corresponds to a subtask unit and includes corresponding optimized execution parameters: subtask unit list, planned time window, equipment number, planned operation path, and target warehouse coordinates.

[0061] The twin data service order is generated based on the global job instruction set and is clearly defined as the twin data required for the precise monitoring and backtracking of each sub-task unit, including: dynamically updated deep modeling scope, perception resource scheduling plan, twin data collection frequency and events, and twin data service timeliness;

[0062] It should be noted that twin data refers to the dynamic information flow collected from the physical world required to accurately reproduce, monitor, and trace port yard operations in the digital space.

[0063] Among them, the dynamically updated deep modeling range is dynamically determined based on the planned time window and planned operation path of each sub-task unit in the global operation instruction set. The perception resource scheduling plan specifies the perception devices for twin data acquisition that need to be scheduled for each sub-task unit. The twin data acquisition frequency and event definition define the update frequency or triggering event of twin data. The twin data service timeliness specifies the start and end time of each twin data requirement, realizing the on-demand allocation and release of resources. For each sub-task unit, the twin data service timeliness is a period of time in which the corresponding planned time window is extended by a preset buffer time in both the forward and backward directions.

[0064] It should be noted that the role of this module is to transform a list of sub-task units that may have resource competition into an executable solution that is conflict-free in terms of time, space, and resources and has the best overall efficiency by establishing a global conflict detection and optimization model. At the same time, by generating twin data service orders, it provides a precise blueprint for the subsequent construction of a high-fidelity digital twin model that is strictly synchronized with physical execution, thus realizing the key leap from the planned state to the executable state.

[0065] Execution and Modeling Module: Automates the execution of sub-task units of the global job instruction set, serves orders based on digital twin data, constructs a dynamically updated three-layer digital twin model, and performs adaptive correction;

[0066] Specifically, when the planned time window of a sub-task unit begins, the sub-task unit is automatically executed. The central scheduling system sends the corresponding instructions from the global operation instruction set to the equipment control system corresponding to the specific equipment number assigned to the sub-task unit through the industrial Internet of Things protocol, thereby driving the equipment to run.

[0067] During execution, the system receives and parses twin data service orders. Based on the dynamic update depth modeling range, perception resource scheduling plan, twin data collection frequency and events, and twin data service timeliness specified in the twin data service order, the system drives the construction and updating of the three-layer digital twin model on demand, including the basic static layer (L1) model, the dynamic business topology layer (L2) model, and the high-fidelity operation surface layer (L3) model.

[0068] Among them, the L1 model loads a pre-built, monthly updated 3D geographic information model (BIM) of the port yard and berths, which serves as a unified spatial reference base map for all dynamic activities.

[0069] The L2 model focuses on the equipment and cargo units corresponding to all active subtask units in the global operation instruction set. An active subtask unit is defined as a subtask unit that enters the corresponding twin data service time at the current moment.

[0070] Specifically, based on the twin data service timeliness in the twin data service order, the digital twins of the corresponding equipment and cargo units are activated before the task begins, and a real-time data link is established with the equipment's Beidou positioning terminal and status sensors. Based on the twin data collection frequency and events, the state update of the L2 model is driven: for continuous states including the position and speed of the equipment and cargo units, twin data is collected and synchronized according to a preset frequency; for key events including the start of the operation and the completion of the grabbing, an instant state transition is triggered.

[0071] For example, when the spreader of the physical quay crane successfully grabs the container, the control system immediately issues an event: Grabbing Complete (timestamp, quay crane ID: 1001, container number: ABC123), which immediately drives the digital twin of quay crane 1001 in the L2 model to change its working state from moving to grabbing, and at the same time changes the position state of the digital twin of container ABC123 to being carried by quay crane 1001.

[0072] The L3 model is generated and updated on demand. Before the execution of the sub-task unit, the sensing equipment is scheduled to go to the corresponding dynamically updated depth modeling range according to the sensing resource scheduling plan. Within the corresponding twin data service time, high-precision laser scanning and image acquisition are performed based on the preset acquisition cycle and key events. Through point cloud registration and 3D reconstruction algorithms, a centimeter-level precision real scene model of the depth modeling range is generated.

[0073] For example, sensing devices include drones, fixed cameras, lidar, and high-precision GNSS;

[0074] It should be noted that the three-layer digital twin model serves to collaboratively construct a twin environment for the automated execution of the global operation instruction set, while also facilitating remote operation by administrators.

[0075] During the execution of subtask units, a dual monitoring mechanism of real-time driving of L2 model and periodic verification of L3 model is adopted. The L2 model is updated synchronously with the event-driven digital twin state based on the twin data acquisition frequency to form a global real-time view of the subtask unit's operation progress. The L3 model serves as the basic verification and correction thread. During the execution of subtask units, the updated L3 model is automatically compared and analyzed with the previous L3 model to verify whether the actual position, posture and equipment gripping point of the cargo unit are consistent with the expected instructions. If a deviation is found, a pose fine-tuning instruction is generated and sent to the equipment control system for adaptive correction.

[0076] It should be noted that this module drives the automated execution of equipment based on the global operation instruction set. At the same time, it builds and updates a three-layer digital twin model (L1 / L2 / L3) on demand based on the twin data service order. This achieves high-fidelity synchronization between physical operations and digital mirrors, providing a complete twin environment for real-time monitoring and accurate verification. The design incorporates a dual monitoring mechanism of real-time driving of the L2 model and periodic verification of the L3 model. L2 ensures real-time synchronization of the global state, while L3 provides a factual benchmark with centimeter-level accuracy and can generate pose fine-tuning instructions to achieve adaptive correction. This deepens the control closed-loop application of digital twins. In addition, the on-demand construction of the three-layer digital twin model significantly reduces the waste of computing and acquisition resources.

[0077] Deviation readjustment module: During execution, it determines whether a deviation warning is triggered. If triggered, it initiates the dynamic control process.

[0078] Specifically, during the automated execution of sub-task units in the global job instruction set, an online closed-loop deviation monitoring and judgment mechanism is established based on the real-time, high-fidelity data stream provided by the dynamically updated three-layer digital twin model. The global job instruction set and twin data service orders are used as a benchmark to continuously compare with the twin data collected from the physical world to determine whether a deviation warning is triggered.

[0079] Specifically, the deviation monitoring and judgment mechanism uses a real-time status comparison engine to analyze based on a preset multi-level deviation early warning rule library. The multi-level deviation early warning rule library defines the judgment conditions and thresholds for different levels and types of deviations. Deviation types include time-series deviation, spatial deviation, status deviation, and resource availability deviation. Among them, time-series deviation is when the actual start time or actual end time of a subtask unit deviates from the planned time window by more than a preset threshold; spatial deviation is when the actual location of an equipment or cargo unit deviates from the planned operation path by more than the safe operation distance; status deviation is when the actual state of the equipment digital twin does not match the expected operation state; and resource availability deviation is when a yard location or path node becomes unavailable due to unforeseen circumstances.

[0080] The real-time status comparison engine runs at a preset fixed high-frequency comparison cycle. In each comparison cycle, the real-time status comparison engine obtains twin data related to all active subtask units from the L2 model and compares and calculates deviations one by one with the global job instruction set. Once a rule is matched in the multi-level deviation warning rule library, a deviation warning of the corresponding level and type is triggered.

[0081] For example, the sub-task units of the mobile containers A to B05 assigned to the yard crane are being executed. The real-time status comparison engine analyzes high-frequency positioning data and calculates that there is a continuous and increasing lateral offset between the actual movement trajectory of the yard crane and the planned operation path. The offset has exceeded the 1.5-meter threshold set in the spatial deviation-path offset rule. A secondary spatial deviation warning is immediately triggered and associated with the specific equipment number, sub-task unit ID and offset coordinates.

[0082] If a deviation warning is triggered, the dynamic control process will be initiated immediately.

[0083] Specifically, the dynamic control process inputs include: triggered deviation warnings, the latest dynamic resource status at the current moment, the list of affected unfinished subtask units, and the real-time status of the current three-layer digital twin model. The rescheduling algorithm based on rolling time-domain simulation is used as the core technical means. The current moment is taken as the new optimization starting point, the list of affected unfinished subtask units is taken as the main optimization object, and global unfinished tasks are also considered. The optimization objective function and basic rule base are reused, and the adjustment range minimization objective is added. In the fast simulation engine, the optimization variable group of the affected subtask units is locally iteratively searched to generate a dynamically adjusted instruction set and update the twin data service order.

[0084] The dynamic adjustment instruction set is directly issued to the affected equipment control system. The instruction format is the same as the global operation instruction set, but it has a higher priority. It is used to overwrite and correct the original instructions and continue the automated execution of the sub-task unit.

[0085] It should be noted that the function of this module is to detect anomalies such as timing deviation and spatial deviation in a timely manner through deviation monitoring and judgment mechanism based on the real-time data stream of the three-layer digital twin model, and to initiate dynamic control process. It uses a rescheduling algorithm based on rolling time domain simulation to quickly generate dynamic adjustment instruction set, ensuring the robustness and adaptability of the operating system.

[0086] The technical solution of this invention is as follows: acquire all work plan data within the work cycle and organize them into sub-task units; acquire dynamic resource status; coordinate global resource conflicts and optimize work for all sub-task units within the work cycle; generate a global work instruction set and a digital twin data service order; automatically execute the sub-task units of the global work instruction set; construct a dynamically updated three-layer digital twin model based on the digital twin data service order and perform adaptive correction; determine whether a deviation warning is triggered during execution; if triggered, initiate a dynamic control process.

[0087] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. A port bulk cargo yard intelligent tallying system based on digital twin, characterized in that: Includes the following modules: Task separation module: Acquires all job plan data within the job cycle and organizes them into sub-task units; Coordination and optimization module: acquires dynamic resource status, performs global resource conflict coordination and job optimization for all sub-task units within the job cycle, and generates a global job instruction set and twin data service order; Execution and Modeling Module: Automates the execution of sub-task units of the global job instruction set, serves orders based on digital twin data, constructs a dynamically updated three-layer digital twin model, and performs adaptive correction; Deviation readjustment module: During execution, it determines whether a deviation warning is triggered. If triggered, it initiates the dynamic control process.

2. The intelligent cargo handling system for port bulk cargo yards based on digital twins according to claim 1, characterized in that: The method for obtaining subtask units is as follows: The operation plan data of each planned vessel is initially decomposed, and the complete cargo loading and unloading instructions are broken down into the minimum operation actions for a single container or a single bulk cargo stack. A multi-objective clustering algorithm is adopted, with the concentration of the operation area and the continuity of equipment operation as the core optimization objectives. Based on the planned time window and planned operation path of the minimum operation action, under the constraints of business rules, all the decomposed minimum operation actions are aggregated into multiple sub-task units that can be independently scheduled, monitored and settled.

3. The intelligent cargo handling system for port bulk cargo yards based on digital twins according to claim 2, characterized in that: The planned time window and planned job path are obtained as follows: The planned time window is based on the planned vessel and the estimated berthing time included in the operation plan data. It uses the standard operation time of the minimum operation action estimated by calling the preset standard operation cycle experience library to obtain the planned time window of the minimum operation action. The planned operation path is obtained by querying the preset standard path topology network, extracting and matching feasible paths that comply with safety regulations and one-way traffic rules between the starting coordinates and the ending coordinates of the minimum operation action according to the shortest path principle.

4. The intelligent cargo handling system for port bulk cargo yards based on digital twins according to claim 1, characterized in that: The methods for global resource conflict coordination and job optimization are as follows: The system acquires the subtask unit list, dynamic resource status, and preset basic rule base for each subtask unit. It constructs a conflict detection model that combines three-dimensional space and time dimensions to automatically detect conflicts within the job cycle. Using a heuristic scheduling algorithm based on priority rules and conflict resolution, and taking the current dynamic resource status as the initial condition, the system optimizes all subtask units, including job order and resource allocation, based on the defined optimization objective function in the constructed fast simulation engine. This generates a comprehensive optimal global scheduling scheme and outputs a global job instruction set based on it.

5. The intelligent cargo handling system for port bulk cargo yards based on digital twins according to claim 4, characterized in that: The optimization objective function includes: The optimization objective function is defined as a weighted combination of the normalized calculation results of the main objective function. The main objectives corresponding to the main objective function include minimizing the total operation time, maximizing the overall equipment utilization rate, minimizing the overall energy consumption, and minimizing the task delay cost. The optimization variable set defining the optimization objective function includes the planned time window of each sub-task unit, the specific equipment number dynamically allocated to each sub-task unit, and the planned operation path of the equipment corresponding to the sub-task unit.

6. The intelligent cargo handling system for port bulk cargo yards based on digital twins according to claim 5, characterized in that: The order for twin data services is generated as follows: The twin data service order is generated by reverse engineering based on the global job instruction set. This includes dynamically determining the dynamic update depth modeling range according to the planned time window and planned job path of each sub-task unit in the global job instruction set, specifying the sensing resource scheduling plan of the sensing devices for twin data collection for each sub-task unit, defining the twin data update frequency and triggering event frequency and events, and obtaining the twin data service timeliness by extending the preset buffer time for both forward and backward of the corresponding planned time window.

7. The intelligent cargo handling system for port bulk cargo yards based on digital twins according to claim 6, characterized in that: The three-layer digital twin model is constructed as follows: The three-layer digital twin model includes a basic static layer model, a dynamic business topology layer model, and a high-fidelity operational surface layer model. Based on the twin data service order, the construction and updating of the three-layer digital twin model are driven: a monthly updated port 3D geographic information model is loaded as the basic static layer model; the digital twin corresponding to the active sub-task unit is activated according to the twin data service timeliness, and a real-time data link is established to drive the status update of the dynamic business topology layer model; and sensing equipment is scheduled according to the sensing resource scheduling plan to perform high-precision data acquisition within the twin data service timeliness, and a high-fidelity operational surface layer model is generated through 3D reconstruction algorithms.

8. The intelligent cargo handling system for port bulk cargo yards based on digital twins according to claim 1, characterized in that: The adaptive correction method is as follows: A dual monitoring mechanism is adopted, which combines real-time driving of dynamic business topology layer model and periodic verification of high-fidelity operation surface layer model. For the high-fidelity operation surface layer model updated during the execution of sub-task unit, it is automatically compared and analyzed with the high-fidelity operation surface layer model before the update to verify whether the actual position, posture and equipment gripping point of the cargo unit are consistent with the expected instructions. If a deviation is found, a pose fine-tuning instruction is generated and sent to the equipment control system for adaptive correction.

9. The intelligent cargo handling system for port bulk cargo yards based on digital twins according to claim 1, characterized in that: The method for determining whether a deviation warning has been triggered is as follows: Based on the real-time data stream provided by the dynamically updated three-layer digital twin model, and taking the global operation instruction set and twin data service orders as the benchmark, the real-time status comparison engine continuously compares and analyzes the twin data in the physical world based on the preset multi-level deviation warning rule library. If any rule in the multi-level deviation warning rule library is matched, a deviation warning is triggered.

10. A port bulk cargo yard intelligent tallying system based on digital twins according to claim 4, characterized in that: The dynamic control process is as follows: A rescheduling algorithm based on rolling time-domain simulation is adopted. The current moment is taken as the new optimization starting point, and the list of affected unfinished subtask units is taken as the main optimization object. The optimization objective function and basic rule base are reused, and the adjustment range minimization objective is added. Local iterative search is performed in the fast simulation engine to generate a dynamically adjusted instruction set and update the twin data service order. The dynamically adjusted instruction set is sent to the affected equipment control system to overwrite and correct the original instruction.

Images