Intelligent scheduling system and method for container trucks in container wharf based on pre-empty calculation

Abstract

The invention discloses an intelligent scheduling system and method for container trucks in a container terminal based on pre-empty calculation. The method comprises the steps of data acquisition, time matrix construction, pre-empty state prediction, scheduling model establishment, optimal task allocation and dynamic updating. The method comprises the following steps: predicting future idle moments and positions of container trucks before the container trucks are not reported to be idle, and realizing pre-allocation of tasks; the driving time between any operation points is calculated through a time matrix, and a globally optimal solution is obtained among task delay, driving distance and load balance in combination with a multi-objective optimization model. The method effectively reduces the vehicle empty driving and bridge crane waiting time, improves the scheduling real-time performance, the resource utilization rate and the system robustness, and is suitable for the field of port and wharf container truck operation scheduling.

Description

Technical Field

[0001] This invention relates to the field of port intelligent scheduling technology, specifically to an intelligent scheduling system and method for container trucks in container terminals based on pre-empty calculations. Background Technology

[0002] With the continued growth of global container throughput, the operational processes of container terminals are becoming increasingly complex. Terminals typically include multiple operational units such as quay cranes, yard cranes, container yards, and internal trucks. Internal trucks, as the key transportation equipment connecting quay cranes and container yards, directly impact the overall terminal's throughput and energy utilization efficiency through their scheduling efficiency.

[0003] Existing truck dispatching systems generally adopt a task-instruction-based allocation mechanism, whereby the system assigns tasks to idle trucks sequentially according to their arrival order or priority. This approach lacks prediction of vehicle operation progress and future availability, triggering the next round of allocation only after a vehicle reports "idle," representing a typical reactive dispatching model.

[0004] However, this scheduling mode has the following shortcomings when ships are berthed in concentrated periods and operational tasks are intensive:

[0005] (1) Task triggering delay and resource idleness: The system needs to wait for the vehicle to complete the task before it can proceed to the next allocation, resulting in gaps between tasks and reducing vehicle utilization.

[0006] (2) Lack of forward-looking forecasting capability: The system cannot predict future idle time based on the location of the truck, the stage of operation and the rhythm of the operation equipment, and cannot perform resource pre-scheduling;

[0007] (3) Insufficient equipment coordination: The scheduling process lacks information linkage with work units such as bridge cranes and yard bridges, resulting in an imbalance in work rhythm and the transmission of resource bottlenecks;

[0008] (4) Insufficient dynamic response capability: In the event of sudden traffic congestion, equipment failure or task changes, the scheduling strategy lacks a real-time adjustment mechanism, making it difficult to maintain stable system operation.

[0009] In view of this, the present invention proposes an intelligent scheduling system and method for container trucks in container terminals based on pre-empty calculation. Summary of the Invention

[0010] The purpose of this invention is to provide an intelligent scheduling system and method for container trucks within container terminals based on pre-empty calculations, aiming to solve the problems of lack of forward-looking prediction, insufficient equipment coordination, and delayed scheduling response in existing container terminal truck scheduling.

[0011] In the first aspect, the present invention provides an intelligent scheduling system for container trucks in a container terminal based on pre-empty calculation, including a pre-empty calculation module, a time matrix module and a TPS scheduling algorithm module, and each module connects with the terminal operation system to form a closed-loop scheduling process;

[0012] The pre-empty calculation module is used to predict the future idle time and idle location of the truck based on the operation stage, task progress and real-time driving time provided by the time matrix module before the truck reports its idle status, and to generate a corresponding pre-empty vehicle pool when the preset time threshold condition is met.

[0013] The time matrix module is used to calculate the estimated travel time between any two work points using a graph theory-based dynamic shortest path algorithm based on the wharf road network topology, lane direction, traffic conditions and crane operation location, and to feed the updated results back to the pre-empty calculation module and the TPS scheduling algorithm module in real time.

[0014] The TPS scheduling algorithm module takes the pre-empty vehicle pool and the time matrix output as joint inputs to establish a multi-objective optimization model that includes "task delay penalty" and "driving cost". When solving the problem, it satisfies conditions such as task uniqueness, vehicle uniqueness, load constraints and operation route resource constraints. It completes the allocation of the next task and path planning before the task is completed, thus forming a closed-loop optimization mechanism for advance scheduling.

[0015] As a preferred technical solution of the first aspect of the present invention, the status acquisition unit is used to periodically acquire the truck number, operation stage, current position, task start time and expected completion time from the terminal operation system;

[0016] The progress analysis unit is used to calculate the remaining operation time based on the operation stage and task progress of the truck, and combine the travel time and waiting time provided by the time matrix module to obtain the expected task completion time.

[0017] The prediction and judgment unit is used to compare the expected task completion time with the system's preset pre-empty time threshold. If the condition is met, the truck is marked as pre-empty.

[0018] The vehicle pool generation unit is used to generate a "pre-empty vehicle pool" based on the pre-empty status, and dynamically update the record truck number, expected idle time, idle location and prediction confidence.

[0019] As a preferred embodiment of the first aspect of the present invention, the pre-emptive calculation module further includes an adaptive correction unit, used to automatically adjust the pre-emptive time threshold based on the deviation between the historical task completion time and the current task, and to feed the correction result back to the next round of prediction calculation.

[0020] As a preferred technical solution of the first aspect of the present invention, a road network construction unit is used to establish a directed graph model G(V,E) with work points and road nodes, and each edge is accompanied by length, speed limit, lane direction, turning penalty time and congestion coefficient;

[0021] The weight update unit is used to periodically collect traffic status data and correct the weights of each side in real time; the path calculation unit is used to calculate the shortest travel time and path node sequence between any two work points.

[0022] The matrix generation unit is used to batch calculate the travel time of all work point pairs to form a time matrix, and triggers matrix refresh when traffic changes exceed a threshold.

[0023] As a preferred embodiment of the first aspect of the present invention, the time matrix module includes:

[0024] This is used to simultaneously feed back the updated time matrix parameters to both the pre-empty calculation module and the TPS scheduling algorithm module after the travel time has been calculated.

[0025] This allows the pre-empty calculation module to call the latest travel time data when predicting idle times, enabling the TPS scheduling algorithm to achieve spatiotemporally consistent dynamic optimization based on the latest traffic conditions during route planning.

[0026] As a preferred embodiment of the first aspect of the present invention, the TPS scheduling algorithm module constructs a multi-objective optimization model, the objective function of which is defined as:

[0027] ;

[0028] in: As the task label, , A collection of tasks; The penalty weight for unassigned tasks. This is a penalty variable, set to 1 when no task is assigned. For the task Priority; For truck identification numbers, , A collection of collectible cards; For the task The set of candidate cards; Cost weighting for travel time, As a decision variable, if the task Assigned to truck The value is 1. For trucks To the mission Travel time, For trucks Available time;

[0029] The model satisfies the following constraints when solving: unique task, unique vehicle, load constraint, operation road resource constraint, and task time window constraint.

[0030] As a preferred embodiment of the first aspect of the present invention, the TPS scheduling algorithm module executes cyclically in units of scheduling cycles, each cycle including:

[0031] Data acquisition, time matrix retrieval, pre-empty vehicle pool update, scheduling model solution, result write-back and anomaly correction;

[0032] In each scheduling cycle, the system automatically adjusts the objective function weights based on the number of vehicles in the pre-empty vehicle pool and the task delay rate.

[0033] As a preferred embodiment of the first aspect of the present invention, the TPS scheduling algorithm module applies the following comprehensive constraints simultaneously when performing multi-objective optimization:

[0034] Task allocation constraints: Each task can be assigned to at most one truck in the same scheduling cycle, and each truck can only execute one task in the same cycle;

[0035] Task integrity constraint: By introducing a penalty variable into the objective function, it is ensured that all tasks are assigned for execution, and unassigned tasks automatically have an increased penalty weight;

[0036] Operational route resource constraints: Set the maximum number of empty vehicles and pre-empty vehicles that can be allocated for each operational route to prevent local congestion and waste of resources;

[0037] Load constraints: When assigning tasks, compare the task weight with the truck's load limit to ensure that the task weight does not exceed the rated load.

[0038] Joint solution mechanism: The algorithm module inputs the task delay penalty term, travel time cost term and all constraints into the mixed integer programming solver for joint optimization, so as to achieve the minimization of task delay and global path optimization under the conditions of task scheduling, resource constraints and load safety.

[0039] As a preferred embodiment of the first aspect of the present invention, it further includes three independent computing nodes: one active node and two standby nodes.

[0040] Each node is equipped with the pre-emptive calculation module, time matrix module, and TPS algorithm module, and a distributed lock mechanism is used to achieve master node election and failover.

[0041] The standby node performs mirroring calculations and verifies the consistency of results in the background. When the primary node's heartbeat times out or the lock is released, the standby node automatically takes over the task, achieving recovery within seconds.

[0042] Secondly, the present invention provides an intelligent scheduling method for container trucks within a container terminal based on pre-empty calculation, for executing the first aspect, comprising the following steps:

[0043] S1: Obtain real-time status of container trucks, task list, and time window parameters from the terminal operation system;

[0044] S2: Establish a time matrix based on the wharf road network topology and traffic conditions, and calculate the travel time between any two work points;

[0045] S3: Based on the truck operation stage, task progress and travel time, predict its future idle time and location, and generate a pre-empty vehicle pool;

[0046] S4: Based on the pre-empty vehicle pool and time matrix, establish a multi-objective optimization model with the goal of minimizing task delay time and travel distance, and set constraints such as task uniqueness, vehicle uniqueness, work route quota and load.

[0047] S5: Solve the task—the optimal vehicle matching scheme—using a mixed integer programming algorithm to generate path planning results;

[0048] S6: Write scheduling instructions into the terminal operation system and update them in real time. When a task delay or vehicle abnormality is detected, perform a local re-solution to achieve dynamic scheduling.

[0049] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0050] This invention utilizes a pre-emptive calculation module to predict the future idle time and location of trucks before they report their availability, thus creating a pool of dispatchable vehicles in advance. This enables pre-allocation of tasks, significantly shortening task response time and reducing crane waiting time. A time matrix module comprehensively considers road length, speed limits, traffic flow, and congestion coefficients to obtain real-time travel times between any work points, providing a precise spatiotemporal reference for task scheduling and route planning, avoiding path conflicts and wasted empty runs. A multi-objective optimization model achieves a globally optimal match between task delay, travel distance, and vehicle load, improving scheduling accuracy and resource utilization. Distributed nodes and a dynamic rescheduling mechanism ensure the system continues to operate even in the event of a single node failure or task anomaly, enhancing system robustness and fault tolerance. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0052] Figure 1 This is a schematic diagram illustrating the architecture and operating mechanism of the present invention;

[0053] Figure 2 This is a schematic diagram of the system architecture and operating mechanism of the present invention;

[0054] Figure 3 This is a flowchart of the pre-emptive calculation module of the present invention;

[0055] Figure 4 This is a schematic diagram of the multi-node optimization mechanism architecture of the present invention; Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this application will be described in more detail below with reference to the accompanying drawings.

[0057] Throughout the accompanying drawings, the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions. The described embodiments are only a part of the embodiments of this application, not all of them. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application. All other embodiments obtained by those skilled in the art based on the embodiments in this application without inventive effort are within the scope of protection of this application. The embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0058] Example 1

[0059] Please see Figure 1 As shown, this embodiment provides an intelligent truck scheduling system for container terminals based on pre-empty calculation. Through technologies such as pre-empty calculation and route optimization, it improves truck operation efficiency, reduces empty runs, optimizes resource utilization, and ultimately achieves the optimal solution for the current state. It includes a pre-empty calculation module, a time matrix module, and a TPS scheduling algorithm module, and through data integration with the Terminal Operations System (TOS), it jointly supports efficient scheduling decisions.

[0060] The pre-idle calculation module obtains real-time monitoring and predicts the idle status of container trucks. Before the container trucks report their idle status, it predicts their future idle time and location, so as to realize the pre-allocation of tasks.

[0061] Specifically, the pre-empty calculation module is responsible for predicting the idle time of container trucks in real time based on their operational status and generating a "pre-empty vehicle pool" that can be used for task allocation. Its main functions include: predicting the remaining operational time of container trucks, assessing their idle status, determining task execution time, and generating pre-empty container truck data that can be used for the next task.

[0062] This can be understood as the logic for predicting travel time using the time matrix module as follows:

[0063] Step A1: Data Acquisition and Input: Periodically acquire real-time operating data of the container trucks from the Terminal Operating System (TOS) and read task data from the task scheduling system.

[0064] The real-time operation data of the container truck includes at least the container truck number, operation stage, current location, task start time, estimated task completion time, communication heartbeat status, and location update time.

[0065] The task data includes task number, task type, starting point, target point, task priority, and task time window.

[0066] Step A2: Time Matrix Data Retrieval: Call the time matrix module interface to obtain the estimated travel time between the truck's current location and the task endpoint or pre-empty location, providing a travel time prediction between any two points. The time matrix module calculates the truck's travel time in real time using factors such as road information, traffic flow, and the truck's current speed. Its output serves as one of the input data for pre-empty calculation.

[0067] Step A3: Calculation of Remaining Working Time: Based on the current work stage and task progress, calculate the remaining working time. If the vehicle is waiting for a gantry crane or yard crane operation, add the estimated waiting time. Based on the truck's current location and the remaining task time, the module combines the current time, travel time, remaining working time for the truck's current task, and waiting time to estimate the estimated completion time for the truck to finish the current task. :

[0068] ;

[0069] in: Indicates the current time of the truck; This indicates that the remaining operation time is calculated based on the task progress of the container truck; This represents the travel time from the truck's current location to the next task location, calculated by the time matrix module. This indicates some waiting time during the truck's operation.

[0070] The system dynamically predicts the future completion time of container trucks in a cyclical manner and calculates whether the trucks meet the conditions for "pre-empty status" based on the expected idle time.

[0071] Step A4: Determine the pre-empty status based on the pre-set pre-empty time threshold. If a vehicle's current estimated completion time is less than the set threshold, it can be classified as a pre-empty vehicle. If the vehicle's current estimated completion time is greater than the set threshold, its status will continue to be monitored until it becomes a pre-empty vehicle or the scheduling task is completely completed.

[0072] Step A5: By constructing a pre-empty vehicle pool, vehicles determined to be in a pre-empty state are written into the pre-empty vehicle pool, and the vehicle number, expected idle time, idle location, prediction confidence and data update timestamp are recorded; when a vehicle task is abnormally delayed, communication is interrupted or a new task is assigned, it is removed from the vehicle pool.

[0073] During task allocation, factors such as truck idle time, task priority, travel time, and road congestion are comprehensively considered to ensure optimal resource allocation. Route optimization uses travel time prediction from the time matrix module, combined with the task allocation algorithm, to select the shortest path and fastest arrival time, avoiding excessive empty runs and resource waste.

[0074] The time matrix module provides estimated travel time data between any two work points within the terminal, providing a unified time reference for pre-emptive prediction and scheduling algorithms.

[0075] It should be noted that factors such as road conditions within the dock, the current status of the trucks, and traffic flow are considered. The time matrix module is used for decision-making in the pre-emptive calculation module and path planning in the TPS scheduling algorithm, providing the estimated travel time of the trucks from their current location to their destination. The time matrix module calculates travel time based on real-time road condition data, truck speed, road conditions, and other information.

[0076] Specifically, the prediction logic for travel time data is as follows:

[0077] Step B1: Road network graph construction: With port operation points and road nodes as vertices V and road connections as directed edges E, establish a directed graph model G(V,E) of the road network. The attributes of each edge include edge length, speed limit, lane direction, turning penalty time and congestion coefficient.

[0078] Step B2: Real-time status update: Periodically collect port area traffic status data and equipment location information, and dynamically adjust the weights of each side. Among them, the congestion coefficient and turning time are used to adjust the travel time of each side.

[0079] Step B3: Edge Time Calculation: Calculate the travel time for each edge. :

[0080]

[0081] in: Let be the side length. To limit speed, The congestion coefficient, This refers to the turning time.

[0082] Step B3: Shortest path solution: Use Dijkstra's algorithm or A* algorithm to calculate the shortest travel time between the specified starting point and the destination, and output the sequence of path nodes.

[0083] Step B4: Time Matrix Generation and Update: Calculate travel time for all work point pairs in batches, construct the time matrix, and cache it. If changes in the port area's road network structure or traffic conditions exceed a threshold, a matrix refresh will be triggered.

[0084] Step B5: Data Output and Interface Call: The module provides the latest travel time matrix to the pre-emptive calculation module and the TPS scheduling algorithm module; when a query request is received, it returns the travel time, path nodes, and data freshness indicator.

[0085] In other words, the system uses a graph-based shortest path algorithm to calculate the travel time of trucks between different road segments. The system input includes information such as the truck's current location, target location, road type, traffic status, and truck speed. Then, based on the road network map, the Dijkstra algorithm or A* algorithm is used to calculate the shortest path from the current location to the destination. Finally, the estimated travel time and possible traffic delay information are output.

[0086] The TPS scheduling algorithm module schedules truck tasks in real time based on the prediction results of the pre-emptive calculation module and the travel time of the time matrix module. Under the condition of meeting the task constraints, it achieves optimal task-vehicle matching and path planning based on a multi-objective optimization model.

[0087] Specifically, the construction logic of the multi-objective optimization model is as follows:

[0088] Step C1: Input Loading: The module acquires the task dataset, vehicle dataset, job route constraint set, time matrix data, and a snapshot of the pre-empty vehicle pool.

[0089] Step C2: Variable Definition: Define the vehicle / truck Should the task be executed? Decision variables If the task Assigned to truck If the value is 1, then the penalty variable is defined. This indicates whether the task has not been assigned; where: , .

[0090] Step C3: Objective Function Construction: Establishing a multi-objective function:

[0091] ;

[0092] in: As the task label, , A collection of tasks; The penalty weight for unassigned tasks; For the task Priority; For truck identification numbers, , A collection of collectible cards; For the task The set of candidate cards; Cost weighting for travel time, For trucks To the mission Travel time, For trucks Available time.

[0093] Step C4: The constraints are as follows:

[0094] Each task can be assigned a maximum of one container truck;

[0095] Each truck can be assigned a maximum of one task;

[0096] Ensure that tasks are assigned;

[0097] This indicates the limit on the number of empty vehicles allocated to the work area. For the set of work routes, For empty vehicles to assemble, For the work route The task set For the work route The maximum number of empty cars that can be assigned;

[0098] This indicates the limit on the number of empty cars to be allocated on the work route. For pre-empty vehicle assembly, Work route The maximum number of pre-empty cars that can be allocated;

[0099] This indicates that the weight of the task cannot exceed the truck's load capacity. For the task The weight; For trucks Load limits;

[0100] In other words, each task can only be performed by one vehicle; each vehicle can only perform one task per cycle; the vehicle's load must not exceed its maximum load; the number of empty and pre-empty vehicles on the work route must not exceed the quota; and task execution must meet time window constraints.

[0101] Step C5: Candidate vehicle set filtering: Filter the candidate vehicle set for each task based on the travel time and vehicle availability status in the time matrix, and remove vehicles that are unreachable or do not meet the constraints.

[0102] Step C6: Solving and Instruction Generation: Solve the model using mixed integer programming or heuristic search algorithms to obtain the optimal matching scheme. Generate scheduling instructions based on the results, including task number, vehicle number, departure time, and suggested path node sequence.

[0103] Step C7: Write-back and Monitoring of Results: Write the scheduling results back to the TOS system and monitor task execution deviations. If the deviation exceeds a set threshold, trigger a local or global re-solution.

[0104] In other words, the TPS scheduling algorithm module is responsible for generating the optimal task allocation and path planning scheme based on real-time job tasks, truck operating status, and time matrix output results. This algorithm is based on "task time window constraints, vehicle status constraints, and job priority constraints," and achieves globally optimal scheduling of truck resources by comprehensively optimizing multiple objective indicators such as empty driving distance, task delay time, and vehicle load balance. Unlike the feasible solutions of traditional algorithms, this algorithm will derive the optimal solution for the current state of internal truck scheduling.

[0105] The TPS scheduling algorithm is designed to select the most suitable truck and generate scheduling instructions for each task, given a truck pool, op-route, and tasks, in order to minimize multiple system-level objectives (including overall operation time, task delay, and empty driving distance). The model explicitly defines decision variables (0-1 indicating whether a task is assigned to a truck), sets of tasks / vehicles / op-routes, and constraints that must be satisfied (each task can be assigned to at most one truck, each truck can be assigned to at most one set of tasks, empty / pre-empty truck quotas on the op-route, and load constraints, etc.).

[0106] The algorithm depends on several real-time / configuration inputs (these are explicitly stated in the API documentation):

[0107] Real-time input (read from TOS / monitoring): Real-time tasks and status of container trucks, status of work routes / bridge cranes, task queues, container truck pool information, GPS / arrival time, etc. Interfaces such as GetTrkTaskInfos, GetTrkPoolInfos, GetTpsTrkTaskRealTimeInfo, and write-back interfaces SaveTrkPreMatchInfo / SaveTrkStartQueuingTime are used for interaction and log write-back.

[0108] Parameter configuration (delivered via the TPS parameter interface): polling interval `schd_loop_interval_tm`, optimization mode `optimization_mode` (efficiency priority / distance priority), objective function weights (`model_obj_un_assignment`, `model_obj_task_delay`, `model_obj_travel_distance`), and thresholds related to the prefree / time matrix modules (such as `prefree_loop_interval_tm`, `prefree_time_out_criteria`, `truck_speed`, `turning_time`, etc.). These parameters directly affect algorithm tuning and scheduling strategies.

[0109] In each scheduling cycle, the TPS master controller first calls the time matrix module to obtain the travel time parameter φ from the candidate vehicle to the task location. 𝑖𝑗 Then, the pre-empty processing module is called to obtain the empty car set and the pre-empty car set, and then the scheduling solution process is entered. Pre-empty cars and empty cars are treated as different types of available resources during scheduling (the allowed number of pre-empty cars for each work route is also constrained).

[0110] The time matrix output (static + dynamic) is used to calculate 𝑡 𝑖𝑗 The source of the data is determined by the fact that TPS prioritizes reading the pre-calculated time matrix table during runtime to save computation time; and it refreshes the data as needed when dynamic updates are required, such as when the bridge crane moves.

[0111] Example 2

[0112] This embodiment, based on Embodiment 1, is an organic whole comprised of three core modules, built upon the Terminal Operating System (TOS) and the Machinery Scheduling System (CWP): a pre-emptive calculation module, a time matrix module, and a TPS scheduling algorithm module. These modules collaborate through a microservice architecture, forming a complete intelligent scheduling closed loop, such as... Figure 2 As shown.

[0113] The system operation begins with the pre-empty calculation module's real-time monitoring of the entire field's truck operation status. This module obtains key data from the TOS system, including the truck's mechanical status, operation stage, unloading time, and current location. Through built-in status judgment logic, it identifies trucks that are about to complete their current operation and marks them as pre-empty. This identification process is based on a comprehensive analysis of the truck's current operation progress and historical operation data, ensuring the accuracy of the pre-empty judgment.

[0114] Upon identifying pre-empty trucks, the system immediately initiates a multi-module collaborative workflow. The pre-empty truck calculation module transmits the pre-empty truck information to the TPS scheduling algorithm module, while the time matrix module provides accurate path time predictions for scheduling decisions. The time matrix module comprehensively considers the actual road network structure at the terminal, lane direction, real-time crane operation location, and road congestion to calculate the truck's travel time between any two points, providing a reliable time basis for scheduling decisions.

[0115] The TPS scheduling algorithm module, acting as the system's decision-making center, receives pre-empty truck information from the pre-empty calculation module and path time data from the time matrix module. Combining this with the current list of tasks to be assigned, it generates the optimal scheduling scheme through operations research and optimization algorithms. This scheme not only considers the timely completion of tasks but also comprehensively considers multiple optimization objectives such as path economy, equipment coordination, and resource balance.

[0116] To further explain, the pre-empty calculation module is based on refined monitoring and intelligent prediction of the truck's operational status. This module continuously acquires real-time data from the TOS system, including key fields such as mechanical status, operational stage, and emptying time from the truck status table. Through real-time analysis of this data, the module can accurately determine the current status and operational progress of each truck.

[0117] like Figure 3 The workflow of the pre-empty calculation module, as shown, first initializes and begins monitoring the truck's status. When a truck is detected to be in a loaded operation phase, the module further analyzes its operation progress. By comparing the average completion time of similar historical operations with the actual progress of the current operation, the module calculates the truck's remaining operation time. If the remaining operation time is less than the system-set pre-empty time threshold, and the truck's communication status is normal and its positioning data is updated in a timely manner, the truck is marked as pre-empty and added to the pre-empty queue.

[0118] The management of the pre-empty queue adopts a dynamic update mechanism, continuously updating the pre-empty truck information according to the system's set polling intervals. For each pre-empty truck in the queue, the module combines data provided by the time matrix to estimate its arrival time at various potential task points, providing basic data support for subsequent scheduling decisions. However, the system continues to monitor loaded trucks that have entered the pre-empty list. If a truck experiences a serious timeout or fails to enter the next task assigned by the system, it will be removed from the pre-empty list, and its original matching task will be given a higher priority.

[0119] In this embodiment, the TPS scheduling algorithm module serves as the decision-making core of the entire intelligent truck scheduling system. Given a set of tasks, a set of vehicles, a set of work routes, and on-site constraints, it calculates and outputs the optimal task-vehicle matching scheme to achieve the comprehensive scheduling goal of maximizing operational continuity, minimizing task delays, and minimizing empty driving distance.

[0120] The TPS intelligent scheduling algorithm runs in an independent service module, executing cyclically with a scheduling cycle (Loop) as the basic time unit. Each scheduling cycle includes the following steps in sequence:

[0121] Data Acquisition: Real-time acquisition of relevant data on tasks, vehicles, equipment, and roads from the Terminal Operating System (TOS) and monitoring system via the interface module;

[0122] Time matrix call: Retrieves the estimated travel time data between any two work points from the time matrix module to form a time constraint input;

[0123] Pre-empty vehicle pool update: Obtain the empty vehicle set and the pre-empty vehicle set from the pre-empty calculation module, and update and deduplicate the data from the previous period;

[0124] Scheduling model solution: Based on multi-objective optimization functions and constraints, task-vehicle matching is solved;

[0125] Result write-back and instruction issuance: Write the matching results back to the database and generate task instructions to issue to the vehicle control system;

[0126] Anomaly detection and dynamic correction: The system continuously monitors task execution deviations and equipment status changes, and triggers a local re-solution when the deviation exceeds a set threshold.

[0127] To further explain, the data sources and interface definitions of the TPS scheduling algorithm module include the following five types of data interfaces:

[0128] Task data interface: provided by the TOS system, including task number, task type (packing, unloading or transshipment), start and end points, task priority, container weight, operation time window and other information;

[0129] Vehicle data interface: provided by the truck monitoring system, including vehicle number, current geographical location, availability status (empty or pre-empty), vehicle load limit and availability start time;

[0130] Time matrix interface: Updated in real time by the time matrix module, outputting the travel time and path node sequence between any two work points;

[0131] Pre-empty status interface: provided by the pre-empty calculation module, outputting the empty car set and the pre-empty car set, as well as the estimated availability time and location of each car;

[0132] The operation route configuration interface is read from the dispatch system configuration file and includes the maximum number of empty cars allowed on the operation route, the maximum number of pre-empty cars, road priority, and dispatch weight parameters.

[0133] Before the start of each scheduling cycle, the system retrieves and caches the above data through the interface module to ensure data consistency and time synchronization during the solution phase.

[0134] In this embodiment, the algorithm runs in an independent service module, executing cyclically in units of scheduling cycles (Loop). Each loop includes: data acquisition; time matrix invocation; pre-empty vehicle pool update; scheduling model solving; scheduling result write-back and distribution; anomaly checking and dynamic correction. To ensure the high availability, fault tolerance, and continuous operation capability of the container terminal intelligent scheduling system, this invention introduces a multi-node parallel optimization mechanism at both the algorithm and system deployment levels.

[0135] The system consists of three computing nodes, each running independently: a TPS scheduling algorithm module; a pre-emptive computation module; and a time matrix module. The three nodes operate in a peer-to-peer cluster state. A distributed lock mechanism (DistributedLock) and a master-slave competition mechanism are used to achieve master node election and automatic failover for the intelligent scheduling system, ensuring that the system continues to provide optimized scheduling capabilities even in the event of a single node failure. The architecture diagram is shown below. Figure 4 As shown, it includes three parallel computing nodes: Node A, Node B, and Node C. Each node deploys a complete scheduling algorithm service component with identical logical functions, but differs in their operational status:

[0136] Active Node: Currently holds the master lock and is responsible for executing the actual scheduling tasks;

[0137] Supervisory Node (Standby Node): The node that has not acquired the master lock is responsible for monitoring the status of the execution node in real time and competing for the lock when the execution node fails.

[0138] During system initialization, all nodes simultaneously attempt to acquire the distributed lock (Lock Key). The node that successfully acquires the lock becomes the execution node, responsible for running the main loop of the TPS scheduling algorithm; the other two nodes automatically enter the monitoring state, periodically polling the execution node's heartbeat, forming a one-master-two-backup structure.

[0139] Furthermore, this embodiment also includes a scheduling lock management mechanism, employing a lightweight distributed lock (such as one based on database row locks or Zookeeper / Redis). During each scheduling cycle, the execution node refreshes the lock's liveness status via a heartbeat mechanism. If the heartbeat is interrupted for more than a set time threshold, the lock is automatically released.

[0140] Once the supervisory node detects that the lock has been released, it immediately initiates a contention (Acquire Lock). Whoever acquires the lock first automatically becomes the new execution node. This process is fully automated, requiring no manual intervention, ensuring that the system can complete a primary / backup switchover within seconds in the event of an execution node failure (such as crash, power outage, or disconnection). This ensures stability and scalability.

[0141] Furthermore, this embodiment also includes a multi-node TPS algorithm collaboration mechanism. When three nodes run the TPS scheduling algorithm simultaneously, the other two nodes, except for the master node, will not idle but will participate in the calculation in a "cooperative supervision" manner.

[0142] Real-time mirroring mechanism: The three nodes share the same task queue and vehicle status data. The supervisor node synchronously performs the same scheduling calculations as the master node in the background, but the results are not submitted to the database; they are only used for consistency verification and health monitoring.

[0143] Consistency verification: If the difference between the optimal solution calculated by the supervisory node and the solution calculated by the execution node exceeds a threshold, the system will record a deviation log for subsequent optimization of the scheduling model or detection of potential computational anomalies.

[0144] State drift and fast recovery: When an execution node crashes or a lock is released, the new execution node can directly use the local computing cache to continue the unfinished scheduling cycle without reloading all the data, achieving recovery in seconds.

[0145] Lock contention and task relay mechanism: During lock switching, the task execution status and scheduling instructions are cached in the shared database. After the new node takes over, it immediately continues to run from the most recently submitted instruction point without affecting job continuity.

[0146] The multi-node optimization core mechanism of this invention combines multi-objective optimization at the scheduling algorithm level with high-availability architecture design at the system level. Through the mechanism of "three-node contention lock + algorithm synchronization mirroring + automatic master-slave drift", it realizes the high availability and robustness of the scheduling system in complex port environments, ensuring the continuity, security and efficiency of truck scheduling.

[0147] Through the above implementation methods, this system has achieved significant technological advancements in practical applications. Traditional truck scheduling systems can only begin new task allocation after a truck completes its current task and enters an idle state. However, this system, through the introduction of a pre-idle calculation module, enables advance prediction of the truck's working status and pre-allocation of tasks.

[0148] This pre-empty calculation and multi-module collaborative working mechanism effectively shortens the time interval between when a container truck completes its current task and begins a new one. The average turnaround time of container trucks is significantly reduced, the waiting time for critical equipment such as gantry cranes is significantly decreased, and the overall operational efficiency of the terminal is improved. At the same time, by optimizing scheduling and reducing the empty driving distance of container trucks, operating costs are lowered, and the economic benefits of the terminal are improved.

[0149] The system's three core modules each perform their specific functions while working closely together: the pre-emptive calculation module provides forward-looking state predictions, the time matrix module provides accurate environmental perception, and the TPS scheduling algorithm module provides scientific decision support. Together, they constitute an efficient and intelligent truck scheduling system. This system overcomes the limitations of traditional solution methods that only provide feasible solutions, achieving the optimal solution for the current state.

[0150] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A container terminal intelligent truck dispatching system based on pre-empty calculation, characterized in that, It includes a pre-emptive calculation module, a time matrix module, and a TPS scheduling algorithm module. Each module interfaces with the terminal operation system to form a closed-loop scheduling process. The pre-empty calculation module is used to predict the future idle time and idle location of the truck based on the operation stage, task progress and real-time driving time provided by the time matrix module before the truck reports its idle status, and to generate a corresponding pre-empty vehicle pool when the preset time threshold condition is met. The time matrix module is used to calculate the estimated travel time between any two work points using a graph theory-based dynamic shortest path algorithm based on the wharf road network topology, lane direction, traffic conditions and crane operation location, and to feed the updated results back to the pre-empty calculation module and the TPS scheduling algorithm module in real time. The TPS scheduling algorithm module takes the pre-empty vehicle pool and the time matrix output as joint inputs to establish a multi-objective optimization model that includes "task delay penalty" and "driving cost". When solving the problem, it satisfies conditions such as task uniqueness, vehicle uniqueness, load constraints and operation route resource constraints. It completes the allocation of the next task and path planning before the task is completed, thus forming a closed-loop optimization mechanism for advance scheduling.

2. The intelligent truck scheduling system for container terminals based on pre-empty calculation as described in claim 1, characterized in that, The pre-emptive calculation module includes: The status acquisition unit is used to periodically obtain the truck number, operation stage, current location, task start time and estimated completion time from the terminal operation system; The progress analysis unit is used to calculate the remaining operation time based on the operation stage and task progress of the truck, and combine the travel time and waiting time provided by the time matrix module to obtain the expected task completion time. The prediction and judgment unit is used to compare the expected task completion time with the system's preset pre-empty time threshold. If the condition is met, the truck is marked as pre-empty. The vehicle pool generation unit is used to generate a "pre-empty vehicle pool" based on the pre-empty status, and dynamically update the record truck number, expected idle time, idle location and prediction confidence.

3. The intelligent truck scheduling system for container terminals based on pre-empty calculation as described in claim 2, characterized in that, The pre-empty calculation module also includes an adaptive correction unit, which is used to automatically adjust the pre-empty time threshold based on the deviation between the historical task completion time and the current task, and feed the correction result back to the next round of prediction calculation.

4. The intelligent truck scheduling system for container terminals based on pre-empty calculation as described in claim 1, characterized in that, The time matrix module includes: The road network construction unit is used to build a directed graph model G(V,E) with work points and road nodes. Each edge is accompanied by its length, speed limit, lane direction, turning penalty time and congestion coefficient. The weight update unit is used to periodically collect traffic status data and correct the weights of each side in real time; the path calculation unit is used to calculate the shortest travel time and path node sequence between any two work points. The matrix generation unit is used to batch calculate the travel time of all work point pairs to form a time matrix, and triggers matrix refresh when traffic changes exceed a threshold.

5. The intelligent truck scheduling system for container terminals based on pre-empty calculation as described in claim 4, characterized in that, The time matrix module also includes a spatiotemporal feedback unit. This is used to simultaneously feed back the updated time matrix parameters to both the pre-empty calculation module and the TPS scheduling algorithm module after the travel time has been calculated. This allows the pre-empty calculation module to call the latest travel time data when predicting idle times, enabling the TPS scheduling algorithm to achieve spatiotemporally consistent dynamic optimization based on the latest traffic conditions during route planning.

6. The intelligent truck scheduling system for container terminals based on pre-empty calculation as described in claim 1, characterized in that, The TPS scheduling algorithm module constructs a multi-objective optimization model, whose objective function is defined as: ； in: As the task label, , A collection of tasks; The penalty weight for unassigned tasks. This is a penalty variable, set to 1 when no task is assigned. For the task Priority; For truck identification numbers, , A collection of collectible cards; For the task The set of candidate cards; Cost weighting for travel time, As a decision variable, if the task Assigned to truck The value is 1. For trucks To the mission Travel time, For trucks Available time; The model satisfies the following constraints when solving: unique task, unique vehicle, load constraint, operation road resource constraint, and task time window constraint.

7. The intelligent truck scheduling system for container terminals based on pre-empty calculation as described in claim 6, characterized in that, The TPS scheduling algorithm module executes cyclically in units of scheduling cycles, each cycle including: Data acquisition, time matrix retrieval, pre-empty vehicle pool update, scheduling model solution, result write-back and anomaly correction; In each scheduling cycle, the system automatically adjusts the objective function weights based on the number of vehicles in the pre-empty vehicle pool and the task delay rate.

8. The intelligent truck scheduling system for container terminals based on pre-empty calculation according to claim 7, characterized in that, When performing multi-objective optimization, the TPS scheduling algorithm module simultaneously applies the following comprehensive constraints: Task allocation constraints: Each task can be assigned to at most one truck in the same scheduling cycle, and each truck can only execute one task in the same cycle; Task integrity constraint: By introducing a penalty variable into the objective function, it is ensured that all tasks are assigned for execution, and unassigned tasks automatically have an increased penalty weight; Operational route resource constraints: Set the maximum number of empty vehicles and pre-empty vehicles that can be allocated for each operational route to prevent local congestion and waste of resources; Load constraints: When assigning tasks, compare the task weight with the truck's load limit to ensure that the task weight does not exceed the rated load. Joint solution mechanism: The algorithm module inputs the task delay penalty term, travel time cost term and all constraints into the mixed integer programming solver for joint optimization, so as to achieve the minimization of task delay and global path optimization under the conditions of task scheduling, resource constraints and load safety.

9. The intelligent truck scheduling system for container terminals based on pre-empty calculation as described in claim 1, characterized in that, It also includes three independent compute nodes: one active node and two standby nodes. Each node is equipped with the pre-emptive calculation module, time matrix module, and TPS algorithm module, and a distributed lock mechanism is used to achieve master node election and failover. The standby node performs mirroring calculations and verifies the consistency of results in the background. When the primary node's heartbeat times out or the lock is released, the standby node automatically takes over the task, achieving recovery within seconds.

10. A method for intelligent scheduling of container trucks within a container terminal based on pre-empty calculation, used to execute the intelligent scheduling system for container trucks within a container terminal based on pre-empty calculation as described in any one of claims 1-9, characterized in that, Includes the following steps; S1: Obtain real-time status of container trucks, task list, and time window parameters from the terminal operation system; S2: Establish a time matrix based on the wharf road network topology and traffic conditions, and calculate the travel time between any two work points; S3: Based on the truck operation stage, task progress and travel time, predict its future idle time and location, and generate a pre-empty vehicle pool; S4: Based on the pre-empty vehicle pool and time matrix, establish a multi-objective optimization model with the goal of minimizing task delay time and travel distance, and set constraints such as task uniqueness, vehicle uniqueness, work route quota and load. S5: Solve the task—the optimal vehicle matching scheme—using a mixed integer programming algorithm to generate path planning results; S6: Write scheduling instructions into the terminal operation system and update them in real time. When a task delay or vehicle abnormality is detected, perform a local re-solution to achieve dynamic scheduling.

Images