Wharf AGV and TGV joint scheduling optimization system based on air rail transportation

Abstract

The invention belongs to the technical field of equipment operation scheduling, and discloses a wharf AGV and TGV joint scheduling optimization system based on air rail transportation. Comprising an area division module which is used for collecting a wharf operation area and dividing the operation area; the model planning module is used for constructing a linear integer planning model by taking minimization of the maximum completion time and minimization of the total delay time as optimization targets, and setting constraint conditions for the operation process of the AGV and the TGV according to the divided operation areas for constraint; the scheduling optimization module is used for performing multi-objective optimization on the linear integer programming model based on an NSGA-II algorithm, solving the optimized linear integer programming model and obtaining an optimal combined scheduling scheme of the AGV and the TGV; according to the invention, through regional task division and resource constraint, the throughput and operation stability of the system are balanced, and reliable decision support is provided for container transfer operation of a water-iron combined transportation port in a complex engineering environment.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment operation scheduling, and more specifically, to a joint scheduling optimization system for terminal AGVs and TGvs based on aerial rail transportation. Background Art

[0002] Traditional horizontal handling equipment focuses on task assignment and path planning. Although the operation efficiency of gantry crane loading and unloading equipment is relatively stable and controllable, its contribution to the overall efficiency improvement of the terminal is gradually slowing down. The aerial rail collection and distribution operation system has the advantages of small floor area and strong line adaptability because its facilities are elevated in the air, and has become a new solution to solve the connection between the port and the railway.

[0003] With the continuous increase in terminal business volume and the growing demand for multimodal transport, against this background, optimizing horizontal handling equipment, especially the scheduling of automated guided vehicles (AGVs, Automated Guided Vehicle, referring to transportation equipment that can travel automatically without a driver through guidance technologies such as lasers, magnetic strips, and vision) and aerial rail freight vehicles (TGvs, Track Guided Vehicle, referring to transportation equipment that relies on fixed tracks (such as aerial tracks) to travel), has become the key breakthrough point for improving terminal operation efficiency. Task assignment aims to efficiently and reasonably dispatch operation tasks to AGVs and TGvs to ensure the quick and accurate completion of tasks; scheduling optimization strives to plan the optimal usage mechanism for equipment to achieve efficient resource utilization and maximization of operation efficiency.

[0004] Currently, in actual operations, many factors such as the task execution status, operation rules, resource occupancy, operation time constraints of horizontal transportation equipment, as well as the number of container operation tasks, the number of equipment, and operation time within the system are not comprehensively considered. Precise arrangement of the task sequence and execution time for AGVs and TGvs may result in a lack of coordination mechanism for shared resources in the separate scheduling system, leading to problems such as blockages easily occurring due to time conflicts or resource occupancy during the task handover between AGVs and TGvs.

[0005] In view of this, the present invention proposes a joint scheduling optimization system for terminal AGVs and TGvs based on aerial rail transportation to solve the above problems. Summary of the Invention

[0006] To overcome the above-mentioned defects of the prior art and to achieve the above object, the present invention provides the following technical solution: A joint scheduling optimization system for terminal AGVs and TGvs based on aerial rail transportation, including: Area division module: Collect the terminal operation area and divide the operation area to obtain the aerial rail system operation area and the internal operation area of the automated terminal. The aerial rail system operation area includes an aerial rail collection and distribution operation subsystem for controlling the TGV to perform container transfer tasks between both ends of a preset fixed track; Model planning module: Taking minimizing the makespan and minimizing the total tardiness as the optimization objectives, a linear integer programming model is constructed, and constraints are set for the operation processes of AGVs and TGvs according to the divided operation areas; Scheduling optimization module: Based on a multi-objective optimization algorithm, multi-objective optimization is performed on the linear integer programming model, the optimized linear integer programming model is solved, and the optimal scheduling schemes for AGVs and TGvs are obtained.

[0007] Furthermore, the method for dividing the operation areas includes: Collect the dock layout data, define the operation area from the shoreline to the yard as the internal operation area of the automated dock, and define the empty rail transportation network connecting the dock and the railway freight yard as the empty rail system operation area; The internal operation area of the automated dock sequentially includes a horizontal transportation operation subsystem, a shoreline operation subsystem, and a yard operation subsystem from the shoreline to the yard; among them, the shoreline operation subsystem is used to control the quay crane to load and unload containers, the horizontal transportation operation subsystem is used for AGVs to travel between the quay crane and the yard, and between the quay crane and the empty rail dock end, and the yard operation subsystem is used to manage the storage and scheduling of containers in the yard.

[0008] Furthermore, the method for setting constraints on the operation processes of AGVs and TGvs and performing constraints includes: Constraining the task assignment and operation sequence of AGVs and TGvs; Constraining the operation rules of AGVs and TGvs; Constraining the resource capacity; Constraining the relationship between each link of the operation task and the time step; Constraining the time relationship of each operation link: Constraining the time relationship of AGVs and TGvs during the task handover process; Constraining the value range of the decision variables that control the task assignment, execution order, time arrangement, and resource occupancy of AGVs and TGvs.

[0009] Furthermore, the method for obtaining the optimal scheduling schemes for AGVs and TGvs includes: Encoding each AGV, TGV, and task type to obtain chromosome individuals; Randomly generate an initial chromosome population; Preset the ideal points with both the minimum makespan and the minimum total tardiness; Stratify the population according to the dominance relationship to obtain the non-dominated sorting result; Calculate the Euclidean distance of each individual to the ideal point, and sort the individuals in each layer of the non-dominated sorting result in ascending order according to the value of the Euclidean distance; Select individuals with Euclidean distance higher than the preset distance threshold from each layer as parents to enter the next generation; For tasks sharing the same loading and unloading area or executed continuously, perform crossover operations on the associated task blocks to be exchanged, detect and repair the crossover results to obtain repaired crossover individuals; Dynamically mutate the task assignment or timing in combination with the number of iterative updates to obtain mutated individuals; For each non-dominated solution in the parents, identify the longest task chain, and optimize the longest task chain through a tabu search strategy to obtain an optimized longest task chain; Replace the corresponding individuals in the parents with the optimized longest task chain to form an offspring population; Merge the parents, repaired crossover individuals, mutated individuals, and offspring population to obtain a new population, perform non-dominated processing on the individuals in the new population, and obtain the top W optimal individuals of the non-dominated processing results; Use the top W optimal individuals as new parents for the next round of iteration; Through J rounds of iteration, screen out the optimized scheduling plan with the minimum makespan and total tardiness.

[0010] Furthermore, each gene in the chromosome includes the equipment number of the AGV or TGV and a task type identifier; where when the task type identifier is an overhead rail task, the task is assigned to the TGV, otherwise the corresponding gene is an invalid gene.

[0011] Furthermore, the method for obtaining the non-dominated sorting result includes: Step 1: Maintain a domination set and a domination counter for each individual, with the initial values set to an empty set and 0; Step 2: Traverse the domination relationship between any pair of individuals in the population, judge the domination relationship, and count the domination set and the number of dominated individuals of each individual; Step 3: Obtain all individuals with a domination count of 0, mark them as the first non-dominated individuals, summarize the first non-dominated individuals into the first layer of the non-dominated sorting result, and remove the first non-dominated individuals from the population to obtain an updated population; Step 4: For the remaining individuals in the updated population, update the number of dominated individuals of the remaining individuals to obtain the updated number of dominated individuals of the remaining individuals. For each removed first non-dominated individual, traverse the corresponding domination set, and subtract 1 from the number of dominated individuals of each individual in the corresponding domination set to obtain the updated number of dominated individuals of the remaining individuals; Step 5: Repeat Step 3 until all individuals are assigned to the non-dominated sorting result.

[0012] Furthermore, the method for judging the domination relationship includes: If the makespan of individual f is not greater than that of individual g, the total tardiness of individual f is not greater than that of individual g, and any objective value of individual f is less than the corresponding objective value of individual g, then it is determined that individual f dominates individual g. The objective values include makespan and total tardiness.

[0013] Furthermore, the methods for counting the domination set and the number of dominated individuals of each individual include: If individual p dominates individual q, add p to the domination set of q and increment the number of dominated individuals of q by 1; Traverse the domination relationships between individual p and all individuals to obtain the domination set and the number of dominated individuals corresponding to individual p.

[0014] Furthermore, the method for obtaining the longest task chain includes: For each non-dominated solution in the parent generation, traverse its task assignment sequence. For the AGV task chain, accumulate the loading, transportation, unloading times and the no-load travel time of each task; for the TGV task chain, accumulate the loading, transportation, unloading times and the no-load travel time of each task; among them, comprehensively analyze the time constraints between tasks; Select the task chain with the longest cumulative time as the longest task chain.

[0015] Furthermore, the method for obtaining the optimized longest task chain includes: Every L iterations, swap the AGV or TGV assignment tasks of adjacent tasks on the longest task chain and shorten the no-load travel time according to a preset adjustment ratio.

[0016] Furthermore, the method for obtaining the top W optimal individuals of the non-dominated processing result includes: Perform non-dominated sorting on the individuals in the new population, calculate the Euclidean distance from each individual in the new population to the ideal point, sort the individuals in each layer of the non-dominated sorting result in ascending order according to the value of the Euclidean distance to obtain the updated non-dominated sorting result; retain the top W optimal individuals of the updated non-dominated sorting result as the top W optimal individuals of the non-dominated processing result.

[0017] The technical effects and advantages of the joint scheduling optimization system of quay AGV and TGV based on aerial ropeway transportation in the present invention: The present invention realizes fully enclosed transportation of combined rail and water transportation of molten iron through an aerial rail system, avoiding dependence on off-site container trucks and reducing traffic interference around the port; solves the coordination problem caused by the physical property differences between AGVs and TGVs, ensuring the matching of equipment task types and operation rules; combines a multi-objective optimization algorithm to minimize the maximum completion time and total delay time through multi-objective optimization, integrates the constraints of the collaborative scheduling of AGVs and TGVs, and avoids the overall efficiency loss caused by single-objective optimization in separate scheduling; and realizes the efficient collaboration of AGVs and TGVs through task chain optimization and dynamic resource allocation in high-load scenarios, reducing equipment waiting and empty running time; balances the system throughput and operation stability through regional task division and resource constraints, providing reliable decision support for the container transfer operation of combined rail and water transportation ports in complex engineering environments based on aerial rail transportation. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a structural diagram of the combined scheduling optimization system of AGVs and TGVs at a terminal based on aerial rail transportation according to the present invention; Figure 2 is a schematic diagram of the relationship between the operation area and the subsystems according to the present invention; Figure 3 is a schematic diagram of the terminal structure of aerial rail transportation according to the present invention; Figure 4 is a schematic diagram of container task allocation according to the present invention; Figure 5 is a schematic diagram of the relationship between container task operations and time steps according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0020] Embodiment 1 Please refer to Figure 1 as shown, this embodiment provides a combined scheduling optimization system of AGVs and TGVs at a terminal based on aerial rail transportation, including: Area division module: Collect the terminal operation area and divide the operation area; The method for dividing the operation area includes: Refer to Figure 2, collect terminal layout data (such as quay crane positions, yard ranges, railway freight yard coordinates, loading and unloading area coordinates, transfer point coordinates, and track intersection coordinates, etc.), define the operation area from the shoreline to the yard as the internal operation area of the automated terminal. The internal operation area of the automated terminal successively includes a shoreline operation subsystem, a horizontal transportation operation subsystem, and a yard operation subsystem from the shoreline to the yard. Among them, the shoreline operation subsystem is used to control the quay crane to load and unload containers, the horizontal transportation operation subsystem is used for the AGV to travel between the quay crane and the yard, and between the quay crane and the end of the aerial rail terminal. The yard operation subsystem is used to manage the storage and scheduling of containers in the yard. The internal operation area of the automated terminal clarifies the core task scope of the AGV (horizontal transportation between the quay crane and the yard, and between the quay crane and the end of the aerial rail terminal) by dividing the shoreline operation subsystem, the yard operation subsystem, and the horizontal transportation operation subsystem, and provides support for the "last mile" of container collection and distribution through the TGV. Moreover, the time windows of the quay crane and the yard crane determine the loading and unloading sequence of the AGV, and can ensure the coordinated operation of the AGV and the quay crane through constraint conditions in the later stage to avoid waiting delays. The fast loading and unloading rules of the AGV in the exchange area and the constraint conditions of empty running can reduce the idle time of equipment and improve the internal transfer efficiency of the terminal.

[0021] Refer to Figure 3 , define the aerial rail transportation network connecting the terminal front and the railway freight yard as the operation area of the aerial rail system; the operation area of the aerial rail system includes an aerial rail collection and distribution operation subsystem, and the aerial rail collection and distribution operation subsystem is used to control the TGV to perform container transfer tasks between the two ends of the preset fixed track. The operation area of the aerial rail system realizes the seamless connection between the terminal front and the railway freight yard through the aerial rail collection and distribution subsystem, provides a dedicated operation space for the TGV, and solves the problems of long-distance and cross-regional transportation that cannot be covered by the AGV in the combined rail and water transportation. The TGV only performs import and export tasks, which can avoid overlapping with the internal tasks of the AGV in the terminal, and ensure its efficient use of track resources through constraint conditions in the later stage. The TGV travels along the fixed track, and the empty running time can be controlled through constraints, providing deterministic parameters for the scheduling model.

[0022] The area division module can clarify the physical operation boundaries and equipment responsibilities of the AGV and the TGV through area division, and can avoid task overlap; it solves the coordination problems caused by the physical property differences between the AGV and the TGV, and ensures the matching of equipment task types and operation rules.

[0023] Model planning module: With the optimization goals of minimizing the maximum completion time and minimizing the total delay time, construct a linear integer programming model, and set constraint conditions for the operation processes of the AGV and the TGV for constraint; among them, the maximum completion time is the total time required to complete the terminal operation; The model planning module constructs a linear integer programming model based on the following assumptions: The stacking positions of containers in the yard and on the ship are known; all containers have the same size, and an AGV and a TGV can only transport one container at a time; the quay crane and yard crane scheduling plans are predetermined and known, which determine the time window information for all containers; the traveling speeds of AGVs and TGVs are known and fixed, and the container handling speeds of quay cranes and sky-rail terminals are known and fixed; after loading and unloading a container in the exchange area, the AGV immediately departs to make room for the next AGV waiting to enter the exchange area for work; the termination exchange area of the previous operation task and the starting exchange area of the subsequent operation task of AGVs and TGVs are both known.

[0024] The constructed linear integer programming model is as follows: ; Among them, is a continuous variable representing the delay time of task ; is the set of all tasks; is a continuous variable representing the makespan; is a continuous variable representing the completion time of task ; Minimizing the makespan can ensure that the maximum value of the completion times of all tasks is as small as possible, avoid the reduction of the overall efficiency caused by the delay of individual tasks, directly improve the throughput efficiency of the terminal, and especially reduce the impact of "bottleneck" tasks in high-load scenarios; minimizing the total delay time can minimize the sum of the delay times of all tasks, ensure that tasks are completed within the time window as much as possible, improve the punctuality of task execution, and reduce the interruption of the port operation chain caused by delays.

[0025] Referring to Figure 3 and Figure 4 , the methods for setting constraint conditions to constrain the operation processes of AGVs and TGVs include: Constraining the task assignment and operation sequence of AGVs and TGVs; for example: ; ; Among them, is the set of AGVs; is the set of TGVs; is the horizontal transportation equipment index; is the set of all tasks; is the virtual initial task; is the virtual end task; is the set of operation tasks related to the sky-rail system, is the set of import tasks for transporting containers from the railway into the port, is the set of export tasks for transporting containers from the port into the railway; is a 0-1 variable. If the horizontal transportation device is continuously executing tasks and task then it is 1, otherwise it is 0; is a 0-1 variable. If the horizontal transportation device is continuously executing tasks and task then it is 1, otherwise it is 0; Through the above formula, it can be ensured that each task must be executed by a unique AGV or TGV, avoiding duplicate assignment or omission of tasks, and ensuring the rational use of resources.

[0026] For example: ; ; ; ; Among them, is a 0-1 variable. If the horizontal transportation device continuously executes the virtual initial task and task then it is 1, otherwise it is 0; is a 0-1 variable. If the horizontal transportation device continuously executes task and the virtual end task then it is 1, otherwise it is 0; Through the above formula, it can be ensured that the task sequence of the AGV / TGV must be continuously executed from the virtual start task to the virtual end task, forcing the tasks to be arranged in logical order and preventing equipment idleness or task interruption.

[0027] For example: ; ; ; Among them, is a 0-1 variable. If the horizontal transportation device is continuously executing tasks and task then it is 1, otherwise it is 0; is a 0-1 variable, indicating that the horizontal transportation device continuously executes task and task then it is 0, otherwise it is 1, indicating that for any horizontal transportation device and any task , execute task Operations to itself are not allowed; through the above formula, it can be ensured that there must be clear precedence and successor relationships between tasks, guaranteeing that the equipment can immediately connect to the next task after completing the current task and reducing the idle time.

[0028] Constrain the operating rules of AGV and TGV; for example: ; ; Among them, in the value range of , is the loading link of AGV executing task , is the transportation link of AGV executing task , is the unloading link of AGV executing task , is the loading link of TGV executing task , is the set of operation tasks related to the empty rail, is the transportation link of TGV executing task , is the unloading link of TGV executing task ; is the set of all relevant operation links in the TGV heavy-load operation; is a 0-1 variable, indicating that the horizontal transportation equipment executes in the time step, then it is 1, otherwise it is 0; is the set of time steps; is the set of all relevant operation links in the AGV heavy-load operation; through the above formula, it can be ensured that within the same time step, the TGV / AGV processes at most one heavy-load operation, which can avoid the equipment being occupied by multiple tasks at the same time and prevent operation conflicts; For example: ; ; ; Among them, is the set of internal operation tasks at the dock; through the above formula, it can be ensured that the empty rail tasks must be executed by the TGV, and the ordinary tasks must be executed by the AGV, which can clarify the equipment division of labor and ensure the physical rules match between the empty rail system and the ground transportation; For example: ; ; ; ; Among them, is the no-load running operation link of the TGV; is a 0-1 variable, indicating that the horizontal transportation equipment executes within the time step then it is 1, otherwise it is 0; is the no-load running operation link of the AGV; is a 0-1 variable, indicating that the horizontal transportation equipment executes within the time step then it is 1, otherwise it is 0; Through the above formula, it can ensure the vehicle occupancy limit in the no-load running link, prevent resource waste during no-load running, and optimize the equipment utilization rate.

[0029] Constrain the resource capacity; for example: ; Among them, is a 0-1 variable, indicating that if the operation link is executed within the time step then it is 1, otherwise it is 0, is the operation link, is the operation, is the set of all operations, is the heavy-load operation set of the horizontal transportation equipment required to execute the task , for the operation , is the set of TGV heavy-load operations required to execute the task , of the set, is the set of AGV heavy-load operations required to execute the task , of the set, is the set of no-load running operations of the horizontal transportation equipment required to be executed for the conversion process , for the operation , , is the operation network; Through the above formula, it can be ensured that each operation link can only be executed by one device within the same time step, and it can avoid the multi-device competition for key resources such as the loading and unloading area and the track; such as ; Among them, indicates that if the task operation link needs to use the resource then it is 1, otherwise it is 0; indicates the upper limit of the operation links that the resource can serve simultaneously; is a resource set, including the resource set related to the internal operations of the terminal and the resource set related to the aerial ropeway system; through the above formula, it can ensure the service upper limit of each resource (such as loading and unloading equipment) within the same time step, preventing resource overload. For example, the loading and unloading equipment at the terminal end only serves a fixed number of tasks at the same time;

[0030] For example: ; ; ; ; ; ; Among them, is a 0-1 variable. If the horizontal transportation equipment uses the resource in the time step, it is 1, otherwise it is 0; is a 0-1 variable. If the horizontal transportation equipment executes the operation link in the time step, it is 1, otherwise it is 0; is the resource set related to the aerial ropeway system; is a 0-1 variable. If the horizontal transportation equipment in the TGV set executes the operation link in the time step, it is 1, otherwise it is 0, is the transportation link for the TGV to execute the task ; is a 0-1 variable. If the horizontal transportation equipment in the TGV set uses the loading and unloading resources (including parking) at the terminal end in the time step, it is 1, otherwise it is 0; and are both a sufficiently large positive number; is a 0-1 variable. If the horizontal transportation equipment in the TGV set uses the loading and unloading resources (including parking) at the terminal end in the time step, it is 1, otherwise it is 0; is a 0-1 variable. If the horizontal transportation equipment in the TGV set uses the loading and unloading resources (including parking) at the railway freight yard end in the time step, it is 1, otherwise it is 0; and are all a sufficiently large positive number; is a 0-1 variable. If the horizontal transportation equipment in the TGV set uses the loading and unloading resources (including parking) at the railway freight yard terminal within the time step then it is 1, otherwise it is 0; is a 0-1 variable. If the horizontal transportation equipment in the TGV set executes the operation link within the time step then it is 1, otherwise it is 0. is the number of TGV devices; Through the above formula, the resource occupancy limit of TGV during different task conversions can be ensured, which can ensure the coherence of the loading and unloading resources of the suspended monorail system and avoid task stagnation caused by resource conflicts.

[0031] Constraints are imposed on the relationship between each link of the operation task and the time step; For example: ; ; ; ; ; ; ; Among them, is a 0-1 variable. If the link is executed after the time step then it is 1, otherwise it is 0; is a 0-1 variable. If the link is executed before the time step then it is 1, otherwise it is 0; is a 0-1 variable, indicating that if the link is executed within the time step then it is 1, otherwise it is 0; is a 0-1 variable, indicating that if the link is executed after the time step then it is 1, otherwise it is 0; is a 0-1 variable, indicating that if the link is executed before the time step then it is 1, otherwise it is 0; is a 0-1 variable, indicating that if the link is executed within the time step then it is 1, otherwise it is 0; is a 0-1 variable, indicating that if the link is executed within the time step It is 1 if executed later, otherwise 0; is a 0-1 variable, indicating that if the link is executed before the time step it is 1, otherwise 0; is a continuous variable, representing the start time of the link ; is the start time of the time step ; is the next time step of; is a sufficiently large positive number; is a 0-1 variable. If the link is executed within the time step it is 1, otherwise 0; is a continuous variable, representing the end time of the link ; is a 0-1 variable. If the link is executed before the time step it is 1, otherwise 0; is a 0-1 variable. If the link is executed after the time step it is 1, otherwise 0; Constraints are imposed on the time relationships of each operation link: For example: ; ; ; ; ; ; ; ; Among them, is a continuous variable, representing the end time of the loading link when the AGV executes the task ; is a continuous variable, representing the time required for the AGV to complete the loading link of the task ; is the upper limit of the time window of the task ; is a continuous variable, representing the end time of the transportation link when the AGV executes the task ; is a continuous variable, representing the time required for the AGV to complete the transportation link of the task ; is a continuous variable, representing the end time of the unloading link when the AGV executes the task ; Is a continuous variable, indicating that the AGV completes the task The time required for the unloading process; Is a continuous variable, indicating that TGV performs tasks The end time of the loading phase; Is a continuous variable, indicating that TGV completes the task The time required for the loading phase; Is a continuous variable, indicating that TGV performs tasks The end time of the transport link; Is a continuous variable, indicating that TGV completes the task The time required for the transportation link; Is a continuous variable, indicating that TGV performs tasks The end time of the unloading phase; Is a continuous variable, indicating that TGV completes the task The time required for the unloading link; the above formula can ensure that the time sequence of the operation links is strictly defined (such as loading → transportation → unloading), and can force the task to be executed according to the physical process. For example, the AGV must complete loading before starting transportation; it can also define the delay time of the task (the difference between the actual completion time and the time window), quantify the task delay, and provide a calculation basis for the optimization target.

[0032] Constrain the time of AGV and TGV during the task handover process; such as: ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; Among them, is a continuous variable, representing the end time of the no-load running operation link in the task conversion process of the TGV ; is a continuous variable, representing the time required for the no-load running operation link in the task conversion process of the TGV ; is a continuous variable, representing the end time of the loading link of the TGV performing the task ; is a continuous variable, representing the time required for the loading link of the TGV to complete the task ; is a continuous variable, representing the end time of the no-load running operation link in the task conversion process of the AGV ; is a continuous variable, representing the time required for the no-load running operation link in the task conversion process of the AGV ; is the no-load running operation link; is a continuous variable, representing the end time of the loading link of the AGV performing the task ; is a continuous variable, representing the time required for the loading link of the AGV to complete the task ; Through the above formula, the time connection between the AGV and the TGV at the handover point (such as the unloading end time ≤ the no-load running start time) can be clarified, as well as the gantry crane operation time limit during the handover of the export task and the import task, which can ensure seamless cooperation of the equipment during the handover process, avoid waiting or blocking; and coordinate the action timing of the ground equipment and the monorail equipment to reduce the handover delay.

[0033] Constrain the value range of the decision variable: ; ; ; ; ; ; ; ; ; ; ; Among them, is the horizontal transportation equipment when performing tasks and tasks in all task sets; is a 0-1 variable. If the horizontal transportation equipment uses resources in the time step, it is 1, otherwise it is 0; through the above formula, 0-1 variables can be defined to represent task allocation and equipment resource occupancy, and the upper and lower bounds of time variables can be defined (such as task start time ≥ ready time); the scheduling problem can be transformed into a solvable discrete optimization problem, and the reasonable range of variables can be restricted to avoid non-physical solutions in the model (such as tasks starting before being ready).

[0034] The above constraint steps solve the resource competition problem by setting time connection constraints and resource capacity constraints, etc., ensuring seamless cooperation between AGV and TGV at the handover point.

[0035] Scheduling optimization module: Based on the multi-objective optimization algorithm, perform multi-objective optimization on the linear integer programming model, solve the optimized linear integer programming model, and obtain the optimal scheduling plan for AGV and TGV.

[0036] The methods for obtaining the optimal scheduling plan for AGV and TGV include: Encode each AGV, TGV, and task type to obtain chromosome individuals. Each gene in the chromosome contains two parts, including the device number of the AGV or TGV and the task type identifier. Among them, when the task type identifier is an overhead rail task, the task is assigned to the TGV, otherwise the corresponding gene is an invalid gene. The chromosome gene contains the device number and the task type identifier, which can ensure that tasks are allocated according to the job area division, avoiding invalid cross-area allocation; reducing the cross-area travel time of the device (such as the AGV does not need to enter the overhead rail area), shortening the total time-consuming of the task chain; matching the task with the device type, reducing the waiting caused by device incompatibility (such as the TGV does not need to handle the internal tasks of the terminal); setting the invalid gene can make the genes other than the overhead rail tasks invalid, forcing the TGV to only execute the import and export tasks; enabling the TGV to focus on long-distance transportation and improving the efficiency of the overhead rail system; avoiding wasting TGV resources on internal terminal tasks and ensuring the timely completion of import and export tasks.

[0037] Randomly generate the initial chromosome population to ensure that the allocation of each task meets the constraints. The initial chromosome population meets the constraints such as task allocation and resource capacity, which can provide a feasible starting point for optimization.

[0038] Preset the ideal point with the minimum maximum completion time and total delay time. The ideal point is dynamically updated according to the individuals with the minimum values of the maximum completion time and total delay time in each generation. Updating the ideal point according to the optimal individual in each generation can guide the NSGA-II algorithm (Non-dominated Sorting Genetic Algorithm, a multi-objective optimization algorithm based on genetic algorithm) to approach the optimal solution; using the ideal point as a benchmark can promote the population to evolve towards a shorter maximum completion time; the goal of minimizing the total delay time of the ideal point can also prompt the algorithm to give priority to repairing high-delay task chains.

[0039] Stratify the population according to the dominance relationship to obtain the non-dominated sorting result. Stratifying according to the dominance relationship can preferentially select the solutions that optimize both the maximum completion time and the total delay time, avoiding the situation of unilateral deterioration.

[0040] Calculate the Euclidean distance from each individual to the ideal point, and sort the individuals in each layer of the non-dominated sorting result in ascending order according to the value of the Euclidean distance. Preset the distance threshold, and select the individuals with Euclidean distances higher than the distance threshold from each layer as the parents to enter the next generation. By using the distance threshold, individuals with uniform distribution can be selected to maintain the population diversity and avoid local optima.

[0041] For tasks sharing the same loading and unloading area or executed continuously, perform the crossover operation of preferentially swapping associated task blocks, and detect and repair the crossover result to obtain the repaired crossover individuals. Performing crossover on tasks sharing the loading and unloading area or continuous tasks can reduce the waiting between tasks.

[0042] Dynamically mutate the task assignment or timing in combination with the number of iterative updates to obtain mutated individuals; adjust the mutation intensity according to the number of iterations, which can explore new solutions in the early stage and fine-tune in the later stage.

[0043] For each non-dominated solution in the parent generation, identify the longest task chain, and optimize the longest task chain through a tabu search strategy to obtain an optimized longest task chain; the tabu search strategy can shorten the idle time by swapping adjacent task assignments every L iterations.

[0044] Replace the corresponding individuals in the parent generation with the optimized longest task chains to form the offspring population; Merge the parent generation, repaired crossover individuals, mutated individuals and offspring population to obtain a new population, perform non-dominated sorting on the individuals in the new population, calculate the Euclidean distance from each individual in the new population to the ideal point, and sort the individuals in each layer of the non-dominated sorting result in ascending order according to the value of the Euclidean distance to obtain the updated non-dominated sorting result; retain the top W optimal individuals in the updated non-dominated sorting result; screening the top W optimal solutions through non-dominated sorting and Euclidean distance can ensure progress in each generation and accelerate the search for the optimal scheduling plan.

[0045] Use the top W optimal individuals as the new parent generation for the next round of iteration; Through J rounds of iteration, screen out the optimized scheduling plan with the minimum makespan and total tardiness.

[0046] The methods for obtaining the non-dominated sorting result include: Step 1: Maintain a domination set and a domination counter for each individual, with the initial values set to an empty set and 0; Step 2: Traverse the domination relationship between any pair of individuals in the population, judge the domination relationship, and count the domination set and the number of dominated individuals for each individual; Step 3: Obtain all individuals with a domination count of 0, mark them as the first non-dominated individuals, summarize the first non-dominated individuals into the first layer of the non-dominated sorting result, and remove the first non-dominated individuals from the population to obtain an updated population; Step 4: For the remaining individuals in the updated population, update the number of dominated individuals of the remaining individuals to obtain the updated number of dominated individuals of the remaining individuals. For each removed first non-dominated individual, traverse the corresponding domination set, and subtract 1 from the number of dominated individuals of each individual in the corresponding domination set to obtain the updated number of dominated individuals of the remaining individuals; Step 5: Repeat Step 3 until all individuals are assigned to a certain layer of the non-dominated sorting result.

[0047] The methods for judging the domination relationship include: If the makespan of individual f is not greater than that of individual g, the total tardiness of individual f is not greater than that of individual g, and any objective value of individual f is less than the corresponding objective value of individual g, then it is determined that individual f dominates individual g. The objective values include makespan and total tardiness.

[0048] The methods for counting the domination set and the number of dominated individuals of each individual include: If individual p dominates individual q, add p to the domination set of q and increment the number of dominated individuals of q by 1; Traverse the domination relationship between individual p and all individuals to obtain the domination set and the number of dominated individuals corresponding to individual p.

[0049] The methods for obtaining the longest task chain include: For each non-dominated solution in the parent generation, traverse its task assignment sequence. For the AGV task chain, accumulate the loading, transportation, unloading times and the no-load travel time of each task; for the TGV task chain, accumulate the loading, transportation, unloading times and the no-load travel time of each task; among them, comprehensively analyze the time constraints between tasks; Select the task chain with the longest cumulative time as the longest task chain.

[0050] The methods for obtaining the optimized longest task chain include: Every L iterations, swap the AGV or TGV assignment tasks of adjacent tasks on the longest task chain and shorten the no-load travel time according to a preset adjustment ratio; L can be dynamically adjusted according to the population diversity.

[0051] The above multi-objective optimization algorithm, such as the NSGA-II algorithm, shortens the no-load time of the longest task chain through tabu search, and also adjusts the mutation intensity according to the number of iterations, preferentially swapping shared loading and unloading areas or consecutive tasks to reduce waiting time, enhancing the search efficiency of the algorithm in the multi-device collaborative scenario. The genetic algorithm for single scheduling lacks such targeted design.

[0052] Embodiment 2

[0053] This embodiment provides a joint scheduling optimization system for quay AGV and TGV in an aerial rail transportation, including: Second Optimization Module: Taking the makespan and total tardiness as the state vector, and the crossover probability and mutation probability as the actions, the crossover probability and mutation probability are adjusted based on the reinforcement learning algorithm. Through the reinforcement learning framework, the real-time state (system performance metrics) is directly associated with the key parameters of the genetic algorithm (crossover / mutation probability) to achieve the adaptive optimization of the scheduling strategy. Specifically, when the system detects that the makespan is long, the reinforcement learning increases the mutation probability to explore a better task assignment sequence; if the total tardiness is high, the crossover strategy is dynamically adjusted to preferentially exchange task blocks with time window conflicts and reduce tardiness. Through this feedback mechanism, the algorithm can balance global search and local exploitation under complex constraints (such as resource capacity, task handover time), continuously optimize the task assignment, timing planning, and resource utilization of AGV and TGV, and finally achieve the synchronous minimization of the makespan and total tardiness, significantly improving the operation efficiency and stability of the automated terminal.

[0054] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, and all of them should be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

[0055] Finally: The above is only the preferred embodiment of the present invention and is not used to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. The terminal AGV and TGV joint scheduling optimization system based on sky-rail transportation is characterized by: include: Area division module: collects terminal operation areas and divides the operation areas to obtain the sky rail system operation area and the internal operation area of ​​the automated terminal. The sky rail system operation area includes the sky rail collection and distribution operation subsystem, which is used to control the TGV to perform container transfer tasks between the two ends along the preset fixed track; Model planning module: Taking minimizing the maximum completion time and minimizing the total delay time as the optimization objectives, a linear integer programming model is constructed, and constraints are set on the operation process of AGV and TGV according to the divided operation areas; Scheduling optimization module: Based on the multi-objective optimization algorithm, the linear integer programming model is optimized, and the optimized linear integer programming model is solved to obtain the optimal scheduling plan for AGV and TGV.

2. The terminal AGV and TGV joint scheduling optimization system based on sky-rail transportation according to claim 1 is characterized in that: The method for dividing the working area comprises: Collect terminal layout data, define the operating area from the shoreline to the yard as the internal operating area of ​​the automated terminal, and define the sky-rail transportation network connecting the terminal and the railway freight yard as the sky-rail system operating area; The internal operation area of ​​the automated terminal includes a horizontal transport operation subsystem, a shoreline operation subsystem and a yard operation subsystem from the shoreline to the yard; among them, the shoreline operation subsystem is used to control the loading and unloading of containers by the quay crane, the horizontal transport operation subsystem is used for the movement of AGVs between the quay crane and the yard, and between the quay crane and the ASR terminal, and the yard operation subsystem is used to manage the storage and scheduling of containers in the yard.

3. The terminal AGV and TGV joint scheduling optimization system based on sky-rail transportation according to claim 1 is characterized in that: Set constraints on the operation process of AGV and TGV. The methods of setting constraints include: Constrain the task allocation and operation sequence of AGV and TGV; Constrain the operating rules of AGV and TGV; Constrain resource capacity; Constrain the relationship between each link and time step of the task; Constrain the time relationship of each operation link: Constrain the time relationship between AGV and TGV during the task handover process; Constrain the value range of decision variables that control the task allocation, execution sequence, time scheduling and resource occupancy of AGVs and TGVs.

4. The terminal AGV and TGV joint scheduling optimization system based on sky-rail transportation according to claim 1 is characterized in that: Methods for obtaining the optimal scheduling scheme for AGV and TGV include: Encode each AGV, TGV and task type to obtain chromosome individuals; Randomly generate an initial chromosome population; Preset the ideal point where both the maximum completion time and the total delay time are minimized; The population is stratified according to dominance relations to obtain non-dominated sorting results; Calculate the Euclidean distance of each individual to the ideal point, and sort the individuals in each layer of the non-dominated sorting results in ascending order according to the value of the Euclidean distance; From each layer, individuals with Euclidean distance higher than the preset distance threshold are selected as parents to enter the next generation; For tasks that share the same loading and unloading area or are executed continuously, a crossover operation is performed to exchange the associated task blocks, and the crossover results are inspected and repaired to obtain the repaired crossover individuals; Dynamically mutate the task allocation or timing according to the number of iterative updates to obtain mutated individuals; For each non-dominated solution in the parent generation, identify the longest task chain, optimize the longest task chain through the taboo search strategy, and obtain the optimized longest task chain; Replace the corresponding individuals in the parent generation with the optimized longest task chain to form a descendant population; Merge the parent generation, repair crossover individuals, mutant individuals and offspring populations to obtain a new population, perform non-dominated processing on the individuals in the new population, and obtain the top W optimal individuals of the non-dominated processing results; The first W best individuals are used as new parents for the next round of iteration; Through J rounds of iterations, the optimal scheduling plan with the smallest maximum completion time and total delay time is screened out.

5. The terminal AGV and TGV joint scheduling optimization system based on sky-rail transportation according to claim 4 is characterized in that: Each gene in the chromosome includes the equipment number of the AGV or TGV, and a task type identifier; wherein, when the task type identifier is an empty track task, the task is assigned to the TGV, otherwise the corresponding gene is an invalid gene.

6. The terminal AGV and TGV joint scheduling optimization system based on sky-rail transportation according to claim 4 is characterized in that: Methods for obtaining non-dominated sorting results include: Step 1: Maintain the dominating set and dominated counter for each individual, with the initial values ​​set to empty set and 0; Step 2: traverse the dominance relationship of any pair of individuals in the population, determine the dominance relationship, and count the dominance set and the number of dominated sets of each individual; Step 3: Obtain all individuals whose dominance number is 0 and mark them as the first non-dominated individuals, aggregate the first non-dominated individuals into the first layer of the non-dominated sorting result, and remove the first non-dominated individuals from the population to obtain an updated population; Step 4: For the remaining individuals in the updated population, update the dominated number of the remaining individuals to obtain the updated dominated number of the remaining individuals. For each first non-dominated individual that has been removed, traverse the corresponding dominating set, subtract 1 from the dominated value of each individual in the corresponding dominating set to obtain the updated dominated number of the remaining individuals. Step 5: Repeat step 3 until all individuals are assigned to the non-dominated sorting results.

7. The terminal AGV and TGV joint scheduling optimization system based on sky-rail transportation according to claim 6 is characterized in that: Methods for determining dominance relationships include: If the maximum completion time of individual f is not greater than the maximum completion time of individual g, the total delay time of individual f is not greater than the total delay time of individual g, and any target value of individual f is less than the target value corresponding to individual g, then it is judged that individual f dominates individual g, and the target values ​​include the maximum completion time and the total delay time.

8. The terminal AGV and TGV joint scheduling optimization system based on sky-rail transportation according to claim 6 is characterized in that: Methods for counting the dominating set and the number of dominated sets of each individual include: If individual p dominates individual q, add p to the domination set of q and increase the number of q's dominance by 1; Traverse the domination relationship between individual p and all individuals, and obtain the domination set and dominated number corresponding to individual p.

9. The terminal AGV and TGV joint scheduling optimization system based on sky-rail transportation according to claim 4 is characterized in that: Ways to get the longest quest chain include: For each non-dominated solution in the parent generation, traverse its task allocation sequence, accumulate the loading, transportation, unloading time and empty travel time of each task for the AGV task chain; accumulate the loading, transportation, unloading time and empty travel time of each task for the TGV task chain; among them, comprehensively analyze the time constraints between each task; The task chain with the longest cumulative time is selected as the longest task chain.

10. The terminal AGV and TGV joint scheduling optimization system based on sky-rail transportation according to claim 4 is characterized in that: Methods for obtaining the optimized longest task chain include: Every L iterations, the AGVs or TGVs of adjacent tasks on the longest task chain are exchanged for task assignment, and the time of empty travel is shortened according to the preset adjustment ratio.

11. The terminal AGV and TGV joint scheduling optimization system based on sky-rail transportation according to claim 4 is characterized in that: The method for obtaining the top W best individuals of the non-dominated processing results comprises: Perform non-dominated sorting on the individuals in the new population, calculate the Euclidean distance of each individual in the new population to the ideal point, sort the individuals in each layer of the non-dominated sorting result in ascending order according to the value of the Euclidean distance, and obtain the non-dominated sorting update result; retain the first W best individuals in the non-dominated sorting update result as the first W best individuals in the non-dominated processing result.

Images