WO2022021119A1 - 一种集装箱码头间全自主水上运输调度方法及系统 - Google Patents

一种集装箱码头间全自主水上运输调度方法及系统 Download PDF

Info

Publication number
WO2022021119A1
WO2022021119A1 PCT/CN2020/105407 CN2020105407W WO2022021119A1 WO 2022021119 A1 WO2022021119 A1 WO 2022021119A1 CN 2020105407 W CN2020105407 W CN 2020105407W WO 2022021119 A1 WO2022021119 A1 WO 2022021119A1
Authority
WO
WIPO (PCT)
Prior art keywords
path
scheduling
time
wagv
transportation
Prior art date
Application number
PCT/CN2020/105407
Other languages
English (en)
French (fr)
Inventor
郑华荣
徐文
马东方
瞿逢重
Original Assignee
浙江大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 浙江大学 filed Critical 浙江大学
Priority to PCT/CN2020/105407 priority Critical patent/WO2022021119A1/zh
Priority to US17/408,480 priority patent/US11886191B2/en
Publication of WO2022021119A1 publication Critical patent/WO2022021119A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/60—Intended control result
    • G05D1/69—Coordinated control of the position or course of two or more vehicles
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B63—SHIPS OR OTHER WATERBORNE VESSELS; RELATED EQUIPMENT
    • B63B—SHIPS OR OTHER WATERBORNE VESSELS; EQUIPMENT FOR SHIPPING 
    • B63B79/00—Monitoring properties or operating parameters of vessels in operation
    • B63B79/20—Monitoring properties or operating parameters of vessels in operation using models or simulation, e.g. statistical models or stochastic models
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B63—SHIPS OR OTHER WATERBORNE VESSELS; RELATED EQUIPMENT
    • B63B—SHIPS OR OTHER WATERBORNE VESSELS; EQUIPMENT FOR SHIPPING 
    • B63B79/00—Monitoring properties or operating parameters of vessels in operation
    • B63B79/40—Monitoring properties or operating parameters of vessels in operation for controlling the operation of vessels, e.g. monitoring their speed, routing or maintenance schedules
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B17/00—Systems involving the use of models or simulators of said systems
    • G05B17/02—Systems involving the use of models or simulators of said systems electric
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/20—Control system inputs
    • G05D1/22—Command input arrangements
    • G05D1/228—Command input arrangements located on-board unmanned vehicles
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/60—Intended control result
    • G05D1/69—Coordinated control of the position or course of two or more vehicles
    • G05D1/692—Coordinated control of the position or course of two or more vehicles involving a plurality of disparate vehicles
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B63—SHIPS OR OTHER WATERBORNE VESSELS; RELATED EQUIPMENT
    • B63B—SHIPS OR OTHER WATERBORNE VESSELS; EQUIPMENT FOR SHIPPING 
    • B63B35/00—Vessels or similar floating structures specially adapted for specific purposes and not otherwise provided for
    • B63B2035/006—Unmanned surface vessels, e.g. remotely controlled
    • B63B2035/007—Unmanned surface vessels, e.g. remotely controlled autonomously operating
    • G—PHYSICS
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G3/00—Traffic control systems for marine craft

Definitions

  • the invention belongs to the field of transportation, and relates to a method and system for fully autonomous water transportation scheduling between container terminals, in particular, for water container transportation between large port terminals, an unmanned ship is used to perform fully autonomous control and scheduling to realize unmanned operation.
  • AGVs Autonomous Guided Vehicles
  • wAGVs waterborne autonomous container carrier
  • the current scheduling algorithm is mainly based on the multi-vehicle routing problem (VRP), which usually has constraints such as time window and loading capacity.
  • VRP multi-vehicle routing problem
  • the VRP dispatch method has been widely used for food delivery, taxi or ambulance dispatch, etc.
  • One of the common points of this kind of problem is that the transportation task may come online with the execution of the planned route, so the planned route needs to be adjusted according to the new transportation task.
  • Dynamic VRP usually employs an insertion method to add incoming transport tasks to existing routes. Another idea is to use methods such as waiting, buffering, etc. to accumulate a period of time before assigning new transportation tasks to vehicles. According to this idea, the rolling time window method is widely used, that is, re-planning is carried out every time interval considering the current system state and the transportation task in a certain time period in the future. Therefore, dynamic scheduling has higher requirements on algorithm solution efficiency.
  • common scheduling problem solving algorithms include exact solving algorithms such as dynamic programming, column generation method, branch pricing method, etc., heuristic algorithms such as tabu search, domain search, local descent search, evolutionary algorithm or a combination of these methods.
  • the exact solution algorithm can usually give the optimal solution, but the solution is complex and only suitable for small-scale static scheduling problems.
  • the heuristic solution algorithm has high solution efficiency, but can only give approximate optimal solutions.
  • Considering the real-time nature of dynamic scheduling more heuristic solution algorithms are generally used.
  • the existing scheduling models mostly consider static problems and the objects are mostly land transportation systems.
  • the task of container transportation between terminals actually arrives dynamically, and the loading and unloading time of containers at the terminal cannot be ignored, so a new dynamic scheduling model needs to be established.
  • most of the land transportation scheduling problems approximate the distance between nodes as the Euclidean distance, while the distribution of large ports and terminals is complex, and the waterway distance cannot be calculated as the Euclidean distance.
  • the purpose of the present invention is to overcome the deficiencies of the prior art, and to provide a fully autonomous water transportation scheduling method and system between container terminals.
  • the scheduling model of the present invention is based on combinatorial optimization, considers performance indicators such as transportation distance and customer satisfaction, and constraints such as time window, loading capacity, first take-last delivery, container loading and unloading time, etc.
  • the tabu search algorithm is designed to optimize and improve the path, and the approximate optimal multi-wAGVs path is obtained in real time. The system helps to further improve the intelligent operation level of large-scale ports.
  • the invention firstly discloses a method and system for fully autonomous water transportation scheduling between container terminals, which comprises the following steps:
  • the heuristic algorithm based on tabu search is used to further improve the initial path, and the final scheduling scheme under the current time step is obtained by solving, and the final scheduling scheme is the path of all wAGVs;
  • the paths of all wAGVs include: The sequence of terminal nodes that all wAGVs will visit, corresponding to the visit time, dwell time, and the number of loading and unloading goods;
  • the performance indicators of the dynamic scheduling model are:
  • c 1 , c 2 , and c 3 are the weight coefficients of the three performance indicators in the model, where the first term minimizes the total travel distance, and the second and third terms minimize waiting and lateness time respectively;
  • D ij is node i ⁇ j waterway distance between two points (not Euclidean distance between two points);
  • vector in represents the waiting time at node i
  • vector in represents the late arrival time at node i
  • ⁇ 1 represents the 1 -norm
  • Set of nodes for scheduling transportation network between terminals The set of edges for scheduling transportation networks between terminals.
  • the invention also discloses a fully autonomous water transportation scheduling system between container terminals, which comprises a wAGV system, a dynamic scheduling model initial path solving module, an initial path building module, a real-time optimization module, and a scheduling module;
  • the wAGV system includes multiple wAGVs, which are controlled by the scheduling module to perform transportation tasks;
  • the initial path solving module of the dynamic scheduling model is used to obtain the parameters involved in defining the dynamic scheduling model, establish the dynamic scheduling model, determine the constraints of the model, and obtain the initial scheduling path;
  • the initial path building module inserts the newly arrived transportation task into possible positions in the existing paths of all wAGVs, calculates the insertion cost and selects the insertion cost.
  • the smallest path and the corresponding position compare and select the path and position corresponding to the smallest cost value, insert the task into the updated path, process all the newly arrived transportation tasks in turn, and obtain the updated initial path;
  • the real-time optimization module adopts the heuristic algorithm based on tabu search, combines the initial path obtained by the initial path building module and the constraints of the dynamic scheduling model, and obtains the approximate optimal scheduling result under the current moving time window frame, realizing the real-time scheduling system. ;
  • the scheduling module controls the wAGVs to visit a specific terminal at a specific time according to the scheduling result, and take or unload a specific number of containers to approximate the shortest travel path and minimum waiting time. and late arrival time to complete the assigned transportation task.
  • the present invention proposes for the first time a dynamic scheduling model and a real-time solution algorithm for container transportation between multiple terminals within a port by using unmanned ships, so as to realize an autonomous water transportation system between large-scale ports and terminals, thereby further improving the The intelligent operation level of large-scale ports.
  • the scheduling model fully considers the complexity of the port waterway transportation network and the dynamics of wAGVs and transportation tasks. Based on the moving time window, a multi-wAGV pickup-delivery combination optimization model with the shortest transportation distance and the highest customer satisfaction is constructed. A heuristic algorithm is proposed to obtain the approximate optimal solution of the scheduling system in real time. Using the multi-terminal transportation data in the Port of Rotterdam to carry out a simulation experiment, it is confirmed that under the scheduling method of the present invention, all transportation tasks are completed within the time window, which shows the effectiveness of the scheduling method of the present invention.
  • Figure 1 is a schematic diagram of the wAGV dynamic planning path
  • Fig. 2 is a schematic diagram of selection-insertion action neighborhood
  • 3 is a schematic diagram of an exchange action neighborhood
  • Figure 6 is the wAGV distance-time and docking diagram
  • Fig. 7 is the simulation result graph that satisfies the load constraint condition
  • Fig. 8 is a simulation result graph that satisfies the time window constraint condition.
  • the invention considers the dynamic scheduling problem of waterway transportation between large ports and terminals, and decides the nodes and time series for all wAGVs to take a specific number of containers from a specific terminal and transport them to a destination terminal within a certain time window.
  • the overall goal is to complete all transportation tasks between terminals, and at the same time, the wAGV travels the shortest distance, waits at the terminal, and has the shortest delay time.
  • FIG. 1 it is a schematic diagram of the wAGV dynamic planning path. Since the planned transportation tasks may not be executed at the new decision time step, the scheduling decision of the new decision time step needs to consider both the previously uncompleted tasks and the newly arrived transportation tasks.
  • this embodiment provides a fully autonomous water transportation scheduling method between container terminals, which mainly includes the following steps:
  • the heuristic algorithm based on tabu search is used to further improve the initial path, and the final scheduling scheme under the current time step is obtained by solving, and the final scheduling scheme is the path of all wAGVs;
  • the paths of all wAGVs include: The sequence of terminal nodes that all wAGVs will visit, corresponding to the visit time, dwell time, and the number of loading and unloading goods;
  • the present invention first proposes a dynamic scheduling model.
  • the model is a dynamic scheduling mathematical model based on moving time windows and mixed integer programming; by solving the model, the initial optimization of the scheduling system is obtained.
  • path the modeling and model solving of the present invention can occur at any time or time step in the actual transportation scheduling process, the present invention uses the time when the model is solved as the initial time step of the method of the present invention), and the scheduling is considered after obtaining the initial optimized path Due to the real-time nature of the system, for the newly arrived transportation, it needs to be inserted into the existing route, and a new optimized scheduling scheme is obtained by considering the constraints and insertion cost.
  • the present invention further proposes a combined moving time window and the initial path construction method of the insertion method and the path improvement solution algorithm based on tabu search, so that the final scheduling scheme at the current time step can be obtained, and the final scheduling scheme is the path of all wAGVs in the system;
  • the path includes the sequence of terminal nodes that all wAGVs will visit, corresponding to the visit time, dwell time, and the number of loading and unloading goods.
  • the initial path construction method combining the moving time window and the insertion method proposed by the present invention is repeatedly executed according to the path calculated in the previous step of the moving time window frame but not fully executed and the newly arrived transportation task.
  • the path improvement algorithm based on tabu search, so as to complete the fully autonomous water transportation scheduling between container terminals in the entire time range.
  • the container quantity on each wAGV l v (k) ⁇ Q Q is the maximum loading quantity of the wAGV.
  • the real-time position of wAGV is recorded as (x v (k), y v (k)), the corresponding water section is recorded as g v (k) and the remaining time to stay/service at the current point is s v (k). Therefore, the dynamic system state of wAGV can be determined by (x v (k), y v (k), g v (k), s v (k), l v (k), R v (k)), Express.
  • wAGV cruise speed is u.
  • each transport task i ⁇ v ⁇ v R v (k) ⁇ R new (k) it can be determined by (i,pi ,d i ,t i ,min ,t i,max ,q i ,s i ) to represent, respectively, the task number, the pick-up terminal, the delivery terminal, the earliest pick-up time, the latest delivery time, the number of containers and the required terminal service time.
  • Integer variable y i (k) for all Indicates the number of containers loaded on the wAGV arriving at node i;
  • the first term minimizes the total travel distance
  • the second term and the third term minimize waiting time and late arrival time, respectively
  • c 1 , c 2 , and c 3 are weight coefficients between the three.
  • Formulas (2)-(18) represent constraints such as consistency, time, loading, and 0-1 integer variables, respectively. Specifically, formula (2) constrains that each node has one and only one wAGV access, formula (3) defines that for each group of pickup and delivery nodes, the pickup node is visited before the delivery node, and is accessed by the same Ship wAGV access. By formulas (4) and (5), it is ensured that the wAGV will visit the origin node and the destination node with the container.
  • Equation (6) means that wAGV will only enter and exit a node if it visits that node.
  • Equations (7) and (8) constrain wAGV to start and end at the correct node.
  • Equation (9) constrains the temporal consistency of wAGVs that are executing tasks. Inequality (10) ensures delivery before pickup.
  • Equation (11) constrains the time consistency between adjacent visited nodes.
  • Equations (12)-(14) define the time window constraints. Loading constraints are defined by equations (15)-(17). The 0-1 integer variable is defined by equation (18).
  • formulas (1)-(18) are a relatively complex combinatorial optimization problem, they are used in the actual scheduling system to solve the initial path when the number of transportation tasks in the first step is still small.
  • the modeling and model solving of the present invention can occur at any time or time step in the actual transportation dispatching process (preferably in the initial stage when the number of transportation tasks is still small),
  • the present invention uses this time as the initial time step of the method of the present invention, and in the subsequent time steps, the transportation tasks arrive in real time and more and more situations, the present invention further proposes an efficient real-time solution algorithm (heuristic real-time solution algorithm ).
  • the newly arrived task is inserted according to certain rules to obtain the initial planned path.
  • wAGVs It may also be an empty set.
  • the cost of inserting the task into all possible positions in the existing wAGV path will be calculated once. Constraints such as no transfer. For pickup point p i , insert it into the path The edge ⁇ r i ,r j > of , if all constraints are not violated, the insertion cost is calculated as:
  • the tabu search method is used to improve it.
  • Tabu search is a kind of neighborhood search, but its search rules are different from other neighborhood search methods in two main ways. First, during iteration, it allows the searched solution to be temporarily inferior to the current solution, thus avoiding getting stuck in a local optimum. Second, tabu search defines a data structure to memorize the most recently performed actions, so that only actions that are not in this memory structure can be executed in the future, so as to avoid the iteration falling into an infinite loop. This taboo action can also be forgiven when the newly searched solution is better than the global optimal solution.
  • the second neighborhood structure is to exchange a group of pick-up and delivery nodes in the two randomly selected paths, that is, first select the pick-up and delivery node pair from the two paths, and delete them from the original path. Then insert it into another path, as shown in Figure 3. Similarly, the loading constraints need to be strictly satisfied, and the time window constraints are penalized in the cost function.
  • two matrices are defined to store neighborhood search and iteration information.
  • the second matrix stores the relevant information corresponding to the two paths, including the path ID, the position of the task in the path, and the number of tasks.
  • the taboo table including taboo objects and taboo length.
  • the taboo object is defined as the transport task sequence ID and path ID, and the taboo length is positioned as the current number of iteration steps plus a preset taboo release step value.
  • the transport task in the taboo object cannot be inserted. to the corresponding path, so as to avoid iterating into a loop.
  • the transformed cost value is smaller than the currently recorded minimum cost value and the change is not in the taboo table, record the cost value as the minimum cost value and update the path; Note that each iteration only needs to update the matrix information of the path corresponding to the neighborhood transformation in the matrix, and does not need to update the matrix of all paths; repeat the above steps to iterate, and gradually improve the scheduling scheme through continuous neighborhood changes and path updates
  • the quality is approximately optimal; the final scheduling scheme is the path of all wAGVs, including the sequence of terminal nodes to be visited, corresponding to the visiting time, residence time, and the number of loading and unloading goods.
  • the fully autonomous water transportation scheduling system between container terminals of the present invention includes a wAGV system, a dynamic scheduling model initial path solving module, an initial path building module, a real-time optimization module, and a scheduling module;
  • the wAGV system includes multiple wAGVs, which are controlled by the scheduling module to perform transportation tasks;
  • the initial path solving module of the dynamic scheduling model is used to obtain the parameters involved in defining the dynamic scheduling model, establish the dynamic scheduling model, determine the constraints of the model, and obtain the initial scheduling path;
  • the initial path building module inserts the newly arrived transportation task into possible positions in the existing paths of all wAGVs, calculates the insertion cost and selects the insertion cost.
  • the smallest path and the corresponding position compare and select the path and position corresponding to the smallest cost value, insert the task into the updated path, process all the newly arrived transportation tasks in turn, and obtain the updated initial path;
  • the real-time optimization module adopts the heuristic algorithm based on tabu search, and combines the initial path construction module to obtain the initial path and the constraints of the dynamic scheduling model, and obtains the approximate optimal scheduling result under the current moving time window frame, realizing the real-time scheduling system;
  • the scheduling module controls the wAGVs to visit a specific terminal at a specific time according to the scheduling result, and take or unload a specific number of containers to approximate the shortest travel path and minimum waiting time. and late arrival time to complete the assigned transportation task.
  • Figures 4 and 5 contain the planned results of this time step: ID of the terminal to be visited, arrival and departure times, and container loading and unloading. For example, in Fig.
  • Figure 6 shows the accumulation of the distance traveled by one of the wAGVs over time, and the docked docks are also marked in the figure. Whenever the wAGV arrives at a terminal, it will stay for a period of time to complete the loading and unloading operations. The difference in dwell time at different terminals is due to the loading and unloading operations that need to perform multiple tasks at some terminals, or waiting for the start of tasks at that terminal.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Ocean & Marine Engineering (AREA)
  • Automation & Control Theory (AREA)
  • Remote Sensing (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Mechanical Engineering (AREA)
  • Combustion & Propulsion (AREA)
  • Chemical & Material Sciences (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Probability & Statistics with Applications (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Business, Economics & Management (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Game Theory and Decision Science (AREA)
  • Medical Informatics (AREA)

Abstract

一种集装箱码头间全自主水上运输调度方法及系统，属于交通运输领域；方法包括：建立面向水上无人集装箱运输船wAGV的动态调度模型；将动态到来的运输任务快速插入所有wAGV现有路径中，计算插入代价并选择插入代价最小的路径及位置，得到更新后的初始路径；采用基于禁忌搜索的启发式算法对初始路径进行改进得到更优的wAGV路径；对wAGV实施调度。该方法的调度模型基于组合优化，考虑运输距离和客户满意度等性能指标，时间窗、装载量等约束条件，利用基于移动时间窗的插入法构建初始路径，并设计禁忌搜索算法对该路径进行优化改进，实时得到近似最优的多wAGVs路径。该系统有助于提高大型港口智能化作业水平。

Description

一种集装箱码头间全自主水上运输调度方法及系统 技术领域
本发明属于交通运输领域，涉及一种集装箱码头间全自主水上运输调度方法及系统，尤其是针对大型港口码头之间的水上集装箱运输采用无人船进行全自主控制与调度，实现无人作业。
背景技术
大型港口通常涉及多个码头，且地理位置分布较为分散。为了应对日益增长的码头业务，大多数集装箱码头内部作业已依赖自主小车(Automated Guided Vehicles,AGVs)实现全自动化。但港口内不同功能码头间，如空置集装箱码头、多式联运中转码头等，也存在大量集装箱运输。目前，主要采用拖挂车完成这类运输。然而，对于地形复杂的港口，如鹿特丹港，有些码头之间的水上距离远远小于陆地距离，采用水上运输的方式可以进一步降低运输成本和能耗。因此，提出一种水上无人集装箱运输船(waterborne Autonomous guided vessels,wAGVs)实现码头间的自主运输系统。该运输系统需要实现多wAGVs的自主调度。
当前调度算法主要基于多车路径规划问题(Vehicle routing problem,VRP),通常附带时间窗、装载量等约束条件，对码头间集装箱运输还需考虑先取再送、不能转运等约束。VRP调度方法已被广泛用于外卖、出租车或救护车派遣等。这类问题的一个共同点之一就是运输任务可能会随着规划路径的执行在线到来，那么已规划的路径就需要根据新的运输任务作出调整。
动态VRP通常采用插入法将新到来的运输任务加入到现有路径中。另一种思路是采用等待、缓冲等方法在将新运输任务分配给车辆之前先积累一段时间。按照这种思路，应用较多的是滚动时间窗方法，即每间隔一段时间考虑当前系统状态和未来某个时间段上的运输任务进行一次重规划。因此，动态调度对算法求解效率要求也较高。目前，常见的调度问题求解算法包括精确求解算法如动态规划、列生成法、分支定价法等，启发式算法如禁忌搜索、领域搜索、局部下降搜索、进化算法或这些方法的组合。精确求解算法通常能给出最优解，但求解复杂，只 适用于小规模静态调度问题。启发式求解算法求解效率高，但只能给出近似最优解。考虑动态调度的实时性，一般采用启发式求解算法较多。然而，现有调度模型多考虑静态问题且对象多为陆地运输系统。码头间集装箱运输任务实际上动态到来，且集装箱在码头的装卸时间不能忽略，需要建立新的动态调度模型。而且，陆地运输调度问题多将节点间距离近似为欧式距离，而大型港口码头分布复杂，水路距离并不能计算为欧式距离。结合问题的动态性，现有调度模型均不适用。水路运输尤其是港口内部码头之间采用wAGVs这样一类新的动态调度问题还未有研究。此外，动态调度需要实时在线求解，现有求解算法未考虑这类新型调度问题特点，因此也需要提出新的高效求解算法。针对采用wAGVs进行大型港口码头间的集装箱自主运输的动态调度系统还未见已公开研究。
发明内容
本发明的目的在于克服现有技术的不足，提供一种集装箱码头间全自主水上运输调度方法及系统。本发明的调度模型基于组合优化，考虑运输距离和客户满意度等性能指标，时间窗、装载量、先取后送、集装箱装卸时间等约束条件，利用基于移动时间窗的插入法构建初始路径，并设计禁忌搜索算法对该路径进行优化改进，实时得到近似最优的多wAGVs路径。该系统有助于进一步提高大型港口智能化作业水平。
本发明的技术方案如下：
本发明首先公开了一种集装箱码头间全自主水上运输调度方法及系统，其包括如下步骤：
1)对集装箱码头间全自主水上运输调度问题进行建模，建立面向wAGVs的动态调度模型，并确定约束条件，求解该模型确定最初优化路径；
2)根据移动时间窗框架下上一步计算出而未被完全执行的路径，将新到来的运输任务插入所有wAGV现有路径中可能的位置，计算插入代价，比较选取最小代价值对应的路径和位置，将该任务插入得到更新后的路径，依次处理所有新到来的运输任务并得到更新后的初始路径；
3)基于更新后的初始路径，采用基于禁忌搜索的启发式算法进一步改进初始路径，求解得到当前时间步下最终的调度方案，最终的调度方案为所有wAGV 的路径；所述所有wAGV的路径包括所有wAGV将访问的码头节点序列，对应访问的时间、停留时间、装卸货数量；
4)依据求解得到的所有wAGV的路径对wAGV实施调度，对于每个新的时间步，重复执行步骤2)-4)。
在本发明的一个实施例中，所述动态调度模型的性能指标为：
Figure PCTCN2020105407-appb-000001
c 1，c 2，c 3为模型中三项性能指标的权重系数其中，第一项最小化总行驶距离，第二项和第三项分别最小化等待和迟到时间；x ijv(k)为0-1整数变量，对于所有
Figure PCTCN2020105407-appb-000002
v∈v，如果wAGV v从节点i→j,则x ijv(k)＝1，否则x ijv(k)＝0，v为拥有n v艘wAGV的船队集合；D ij为节点i→j间的水路距离(非两点间欧式距离)；向量
Figure PCTCN2020105407-appb-000003
其中
Figure PCTCN2020105407-appb-000004
表示在节点i的等待时间，向量
Figure PCTCN2020105407-appb-000005
Figure PCTCN2020105407-appb-000006
其中
Figure PCTCN2020105407-appb-000007
表示在节点i的迟到时间，‖·‖ 1表示1范数，
Figure PCTCN2020105407-appb-000008
为码头间调度运输网络的节点集，
Figure PCTCN2020105407-appb-000009
为码头间调度运输网络的边集。
本发明还公开了一种集装箱码头间全自主水上运输调度系统，其包括wAGV系统、动态调度模型最初路径求解模块、初始路径构建模块、实时优化模块、调度模块；
所述的wAGV系统包括多艘wAGV，其受调度模块控制执行运输任务；
所述的动态调度模型最初路径求解模块，用于获取定义动态调度模型涉及到的参数，建立动态调度模型，确定模型的约束条件，并获取最初调度路径；
所述的初始路径构建模块，根据移动时间窗框架下上一步计算出而未被完全执行的路径，将新到来的运输任务插入所有wAGV现有路径中可能的位置，计算插入代价并选取插入代价最小的路径及相应位置，比较选取最小代价值对应的路径和位置，将该任务插入得到更新后的路径，依次处理所有新到来的运输任务并得到更新后的初始路径；
实时优化模块，采用基于禁忌搜索的启发式算法，结合初始路径构建模块得的初始路径以及动态调度模型的约束条件，得到当前移动时间窗框架下近似最优的调度结果，实现调度系统的实时化；
调度模块，根据实时优化模块得到的当前移动时间窗框架下所有wAGV的路 径，控制wAGV按照调度结果在特定的时间访问特定的码头，取或卸特定数量的集装箱，以近似最短行驶路径、最小等待和迟到时间的方式完成指派的运输任务。
与现有技术相比，本发明首次提出利用无人集船进行港口内部多码头之间进行集装箱运输的动态调度模型和实时求解算法，实现大型港口码头之间的自主水上运输系统，从而进一步提高大型港口智能化作业水平。该调度模型充分考虑港口水路运输网络的复杂性及wAGV和运输任务的动态性，基于移动时间窗，构建一种运输距离最短、客户满意度最高的多wAGV取货-送货组合优化模型，并提出启发式算法实时求得调度系统的近似最优解。利用鹿特丹港多码头间运输数据开展仿真实验，证实本发明的调度方法下，所有运输任务都在时间窗内完成，表明了本发明调度方法的有效性。
附图说明
图1为wAGV动态规划路径示意图；
图1中，1.集装箱码头，2.水路节点，3.规划的已完成的路径，4.规划的但还未完成的路径，5.1 TEU集装箱，6.wAGV。
图2为选择-插入动作邻域示意图；
图3为交换动作邻域示意图；
图4为时间步k＝7规划路径；
图5为时间步k＝8规划路径；
图6为wAGV距离-时间及停靠码头图；
图7为满足装载量约束条件仿真结果图；
图8为满足时间窗约束条件仿真结果图。
具体实施方式
下面结合具体实施方式对本发明做进一步阐述和说明。本发明中各个实施方式的技术特征在没有相互冲突的前提下，均可进行相应组合。
本发明考虑大型港口码头之间的水路运输动态调度问题，决策出所有wAGV在一定时间窗内，从特定码头取特定数量的集装箱运送至目的地码头的节点和时间序列。整体目标是完成所有码头间的运输任务，同时wAGV行驶距离最短， 在码头等待、延时时间最短。
如附图1所示，为wAGV动态规划路径示意图。由于已规划的运输任务在新决策时间步可能还未执行完，因此新决策时间步的调度决策既需要考虑之前未完成的任务，也需要考虑新到来的运输任务。
基于以上调度决策要求，本实施例提供了一种集装箱码头间全自主水上运输调度方法，其主要包括如下步骤：
1)对集装箱码头间全自主水上运输调度问题进行建模，建立面向wAGV的混合整数规划动态调度模型，求解该混合整数规划问题以得到调度系统最初优化路径；
2)根据移动时间窗框架下上一步计算出而未被完全执行的路径，将新到来的运输任务插入所有wAGV现有路径中可能的位置，计算插入代价，依次处理所有新到来的运输任务并得到更新后的初始路径；
3)基于更新后的初始路径，采用基于禁忌搜索的启发式算法进一步改进初始路径，求解得到当前时间步下最终的调度方案，最终的调度方案为所有wAGV的路径；所述所有wAGV的路径包括所有wAGV将访问的码头节点序列，对应访问的时间、停留时间、装卸货数量；
4)依据求解得到的所有wAGV的路径对wAGV实施调度，对于每个新的时间步，重复执行步骤2)-4)。
从本发明方法的步骤可见，本发明先是提出一种动态调度模型，本实施例中，该模型是基于移动时间窗和混合整数规划的动态调度数学模型；通过模型求解，得到调度系统的最初优化路径(本发明的建模和模型求解可以发生在实际运输调度过程的任一时间或时间步，本发明以该模型求解的时间作为本发明方法的初始时间步)，得到最初优化路径后考虑调度系统的实时性，对于新到来的运输，需将其插入现有路径中，并考虑约束条件和插入代价得到新的优化后的调度方案，为此，本发明进一步提出了一种结合移动时间窗和插入法的初始路径构建方法及基于禁忌搜索的路径改进求解算法，从而可以得到当前时间步下最终的调度方案，最终的调度方案即为系统中所有wAGV的路径；所述系统中所有wAGV的路径包括所有wAGV将访问的码头节点序列，对应访问的时间、停留时间、装卸货数量。而对于下一时间步，则根据移动时间窗框架下上一步计算出而未被完全执行的路径和新到来的运输任务，重复执行本发明提出的结合移动时间窗和插入法 的初始路径构建方法及基于禁忌搜索的路径改进求解算法，从而完成整个时间范围的集装箱码头间全自主水上运输调度。
以下对本发明的各部分进行进一步的说明。
一、基于移动时间窗和混合整数规划的动态调度
首先，定义调度问题数学模型涉及到的参数。考虑一个时间范围t∈[0,∞]，拥有n v艘wAGV船队集合v的动态运输系统，其中正在执行任务的wAGV集合为v w。每间隔时间T s进行一次重新规划，定义离散的规划时间步为k，则t＝kT s。
每个时间步k，考虑前一时刻未完成的任务
Figure PCTCN2020105407-appb-000010
以及[(k+N p-1)T s,(k+N p)T s]新到来的任务R new(k)，在时间域[kT s,(k+N p)T s]上建立重规划问题。其中
Figure PCTCN2020105407-appb-000011
为截止kT s还未开始执行的任务，
Figure PCTCN2020105407-appb-000012
为集装箱已在wAGV v上有待被送达目的地码头的任务。在下一个时间步k+1，重复上述流程并考虑未来N p个时间步上的任务，由此得名“移动时间窗”。同时，可以计算出每艘wAGV上的集装箱量l v(k)≤Q，Q为wAGV的最大装载量。wAGV的实时位置记为(x v(k),y v(k))，对应所在的水路段记为g v(k)及在当前点需停留/服务剩余时间为s v(k)。因此，wAGV的动态系统状态可由(x v(k),y v(k),g v(k),s v(k),l v(k),R v(k)),
Figure PCTCN2020105407-appb-000013
表示。wAGV巡航速度为u。
相应的，对每一个运输任务i∈∪ v∈vR v(k)∪R new(k)，可由(i,p i,d i,t i,min,t i,max,q i,s i)来表示，分别表示任务编号、取货码头、送货码头、最早取货时间、最晚送达时间、集装箱数量和所需码头服务时间。对取货码头p i，q i>0；对送货码头d i，q i<0。
对码头间水路运输网络进行建模。定义v o(k)＝{1,…,n v}为所有wAGVs的起始节点集，v e(k)＝{n v+2n(k)+n d(k)+1,…,2n v+2n(k)+n d(k)}为所有wAGVs的终点节点集。其中，
Figure PCTCN2020105407-appb-000014
为取送运输任务数量，n d(k)为仅送货任务数量。所有的取货节点定义为
Figure PCTCN2020105407-appb-000015
Figure PCTCN2020105407-appb-000016
对应取货节点的送货节点为
Figure PCTCN2020105407-appb-000017
仅送货节点为
Figure PCTCN2020105407-appb-000018
所以，码头间调度问题考虑的运输网络为
Figure PCTCN2020105407-appb-000019
其中节点集
Figure PCTCN2020105407-appb-000020
边集
Figure PCTCN2020105407-appb-000021
定义如下决策变量：
·0-1整数变量x ijv(k)对于所有
Figure PCTCN2020105407-appb-000022
v∈v，如果wAGV v从节点i→j,则x ijv(k)＝1，否则x ijv(k)＝0；
·0-1整数变量z iv(k)对于所有
Figure PCTCN2020105407-appb-000023
v∈v，如果wAGV v访问节点i,则z iv(k)＝1，否则z iv(k)＝0；
·整数变量y i(k)对于所有
Figure PCTCN2020105407-appb-000024
表示到达节点i的wAGV上装载的集装箱数量；
·连续变量A i(k)对于所有
Figure PCTCN2020105407-appb-000025
表示到达节点i的时间；
·连续变量
Figure PCTCN2020105407-appb-000026
对于所有
Figure PCTCN2020105407-appb-000027
表示在节点i的等待时间；
·连续变量
Figure PCTCN2020105407-appb-000028
对于所有
Figure PCTCN2020105407-appb-000029
表示在节点i的延时；
为描述简洁性，以下将省略动态参数中的·(k)。定义了这些参数之后，建立混合整数规划模型对调度问题进行建模。
Figure PCTCN2020105407-appb-000030
受约束于
Figure PCTCN2020105407-appb-000031
Figure PCTCN2020105407-appb-000032
Figure PCTCN2020105407-appb-000033
Figure PCTCN2020105407-appb-000034
Figure PCTCN2020105407-appb-000035
Figure PCTCN2020105407-appb-000036
Figure PCTCN2020105407-appb-000037
Figure PCTCN2020105407-appb-000038
Figure PCTCN2020105407-appb-000039
Figure PCTCN2020105407-appb-000040
Figure PCTCN2020105407-appb-000041
Figure PCTCN2020105407-appb-000042
Figure PCTCN2020105407-appb-000043
Figure PCTCN2020105407-appb-000044
Figure PCTCN2020105407-appb-000045
Figure PCTCN2020105407-appb-000046
x ijv,z iv∈{0,1},  (18)
公式(1)中，第一项最小化总行驶距离，第二项和第三项则分别最小化等待和迟到时间，c 1，c 2，c 3为三者之间的权重系数。公式(2)-(18)分别表示一致性、时间、装载量及0-1整数变量等约束条件。具体而言，公式(2)约束每个节点有且只有一艘wAGV访问，公式(3)定义了对每一组取货和送货节点，取货节点先于送货节点访问，且被同一艘wAGV访问。通过公式(4)和(5)确保wAGV会访问起始节点及装有集装箱的目的地节点。公式(6)意味着wAGV如果访问某个节点才会进出那个节点。公式(7)和(8)约束wAGV在正确的节点出发和结束。公式(9)约束正在执行任务的wAGV的时间一致性。不等式(10)确保先送货后取货。公式(11)约束相邻访问节点间的时间一致性。公式(12)-(14)定义时间窗约束。装载量约束由公式(15)-(17)限定。0-1整数变量由公式(18)限定。
由于公式(1)-(18)是一个较为复杂的组合优化问题，在实际调度系统中用于求解第一步运输任务数量还较少时的最初路径。通过模型求解，可以得到调度系统的最初优化路径，本发明的建模和模型求解可以发生在实际运输调度过程的任一时间或时间步(优选为运输任务数量还较少时的初始阶段)，本发明以该时间作为本发明方法的初始时间步，而对于后续时间步中，运输任务实时到来且越来越多的情况，本发明进一步提出一种高效的实时求解算法(启发式实时求解算法)。
二、启发式实时求解算法
首先，根据移动时间窗框架下上一步计算出而未被完全执行的路径，将新到来的任务按一定规则插入，得到初始规划路径。对所有wAGV v，定义其未完成的路径为
Figure PCTCN2020105407-appb-000047
其中r i,i＝1,…,m为wAGV v上一步规划中将要访问的节点。对于某些wAGV，
Figure PCTCN2020105407-appb-000048
也有可能为空集。如果新到来的运输任务为<p i,d i>，则将该任务插入所有wAGV现有路径中可能位置的代价会被计算一遍，这些插入位置需要满足时间窗、装载量、先取后送及不能转运等约束条件。对取货点p i，将其插入路径
Figure PCTCN2020105407-appb-000049
的边<r i,r j>，如果所有约束条件均没有违反，则插入 代价计算为：
Figure PCTCN2020105407-appb-000050
对p i所有可能插入位置的代价进行升序排序，然后从最小的Δd p开始，同样的将d i插入对应p i点之后。对不违背约束条件的位置计算d i插入代价，计为Δd d。对于d i，如果某个插入点检测出违反了装载量约束条件，则剩余的插入位置可以不用再考虑。找到最小的Δd d值与当前对应的Δd p，则该新任务<p i,d i>插入路径
Figure PCTCN2020105407-appb-000051
的代价即为Δd p+Δd d，依次计算<p i,d i>插入所有wAGV当前路径的代价并比较选取最小代价值对应的路径和位置，将该任务插入得到更新后的路径。依次处理所有新到来的运输任务并得到更新后的初始路径。
基于得到的初始路径，采用禁忌搜索的方法对其进行改进。禁忌搜索是邻域搜索的一种，但其搜索规则与其他邻域搜索方法主要有两个不同之处。第一，在迭代中，它允许搜索到的解暂时劣于当前解，从而避免陷入局部最优解。第二，禁忌搜索定义一种数据结构对最近执行过的动作进行记忆，则未来只有不在此记忆结构中的动作能被执行，以避免迭代陷入死循环。当新搜索到的解优于全局最优解时，此禁忌动作也可被赦免。
首先设计当前解的领域空间。由于存在先取后送和不能转运这样一类约束条件，在设计邻域时必须考虑取货节点和送货节点同时移动。提出两种邻域空间。第一种是对现有解中的任意两条路径，任意选取其中一条路径中的任意一组取货送货节点，先将其从原路径中删除，再将其插入另一条路径，如附图2所示。与之前新任务插入不同的是，此时只需要装载量约束条件严格满足，但时间窗约束类似公式(1)，只对等待时间和迟到时间进行惩罚，而不要求严格满足。
第二种邻域结构是对任意选取的两条路径中的一组取货送货节点进行交换，即先从两条路径中分别选取取货送货节点对，将其从原路径中删除，再插入到另一条路径中，如附图3所示。同样的，装载量约束条件需要严格满足，时间窗约束条件采用在代价函数里惩罚的方式。
基于设计的邻域空间，定义两个矩阵储存邻域搜索和迭代信息。首先，定义n×n(n≤n v)维矩阵存任意两条路径对之间经过邻域变换后得到的最优解信息及该路径对的序号，此序号为定义的第二个矩阵的行序号。第二个矩阵储存对应两条路径的相关信息，包括路径ID、任务在路径中的位置及任务数量等。通过这种数据结构，在每一次迭代中，只需要更新矩阵中有变动的路径相关信息， 而不需要对所有路径进行更新。同时，定义禁忌表，包括禁忌对象和禁忌长度。禁忌对象定义为运输任务序ID和路径ID，禁忌长度定位为当前迭代步数加上一个预设的禁忌释放的步数值，当迭代步数小于禁忌长度时，则禁忌对象中的运输任务不能插入到对应路径中，以此避免迭代陷入循环。在每一次迭代中，选取代价下降最多的邻域变换，如果变换后的代价值比当前记录的最小代价值小且改变换不在禁忌表中，则记录该代价值为最小代价值并更新路径；注意每次迭代仅需要更新矩阵中有进行邻域变换的路径对应的矩阵信息，不需要对所有路径的矩阵更新；重复上述步骤进行迭代，经过连续的邻域变化、路径更新以逐步提高调度方案的质量至近似最优；最终的调度方案为所有wAGV的路径，包括将访问的码头节点序列，对应访问的时间、停留时间、装卸货数量。
本发明的集装箱码头间全自主水上运输调度系统包括wAGV系统、动态调度模型最初路径求解模块、初始路径构建模块、实时优化模块、调度模块；
所述的wAGV系统包括多艘wAGV，其受调度模块控制执行运输任务；
所述的动态调度模型最初路径求解模块，用于获取定义动态调度模型涉及到的参数，建立动态调度模型，确定模型的约束条件，并获取最初调度路径；
所述的初始路径构建模块，根据移动时间窗框架下上一步计算出而未被完全执行的路径，将新到来的运输任务插入所有wAGV现有路径中可能的位置，计算插入代价并选取插入代价最小的路径及相应位置，比较选取最小代价值对应的路径和位置，将该任务插入得到更新后的路径，依次处理所有新到来的运输任务并得到更新后的初始路径；
实时优化模块，采用基于禁忌搜索的启发式算法，结合初始路径构建模块得到初始路径以及动态调度模型的约束条件，得到当前移动时间窗框架下近似最优的调度结果，实现调度系统的实时化；
调度模块，根据实时优化模块得到的当前移动时间窗框架下所有wAGV的路径，控制wAGV按照调度结果在特定的时间访问特定的码头，取或卸特定数量的集装箱，以近似最短行驶路径、最小等待和迟到时间的方式完成指派的运输任务。
为了验证所提算法的有效性，利用鹿特丹港多码头间运输数据，开展了仿真实验。
设置wAGV数量为8，T s＝15min.，N p＝4，即调度时域为1小时。图4和图5显示其中一艘wAGV连续两个时间步k＝7和k＝8的规划路径。路径上的小矩形表示1 TEU，上面的数字表示该集装箱属于的运输任务编号。图4和图5中包含该时间步所规划的结果：将要访问的码头ID，到达、离开时间，集装箱装卸量。例如，图4中，wAGV离开码头6时船上有2个运输任务9的集装箱，有2个任务18的集装箱；接着到达码头8,后先卸下任务18的2个集装箱，再装载任务21的2个集装箱。相同节点的装卸作业通常会被集中在一起以减少总共行驶距离。比较k＝7和k＝8中的规划路径，可以看出新到任务23和26被插入到k＝7的路径中。新任务中与k＝7的路径相同码头节点的作业被整合在一起以减低行驶距离。
图6显示的是其中一艘wAGV行驶距离随时间的积累，图中也标出了停靠的码头。每当wAGV到达一个码头，则会停留一段时间以完成装卸货作业。在不同码头停留时间不一样是由于在某些码头需要执行多个任务的装卸作业，或在那个码头等待任务的开始。
由于wAGV有最大装载量限制，图7中给出了所有8艘wAGV在整个仿真时间段内船上的集装箱数量。可以看到所有wAGV都没有超载，即集装箱数量一直不大于装载量上限4TEU。
对于时间窗约束，图8中画出了所有运输任务的设置时间差和任务时间完成所需要的时间。所有任务都在时间窗内完成，也表明了调度算法的有效性。
以上所述实施例仅表达了本发明的几种实施方式，其描述较为具体和详细，但并不能因此而理解为对本发明专利范围的限制。应当指出的是，对于本领域的普通技术人员来说，在不脱离本发明构思的前提下，还可以做出若干变形和改进，这些都属于本发明的保护范围。因此，本发明专利的保护范围应以所附权利要求为准。

Claims (7)

  1. 一种集装箱码头间全自主水上运输调度方法，其特征在于包括如下步骤：
    1)对集装箱码头间全自主水上运输调度问题进行建模，建立面向wAGV的混合整数规划动态调度模型，求解该混合整数规划问题以得到调度系统最初优化路径；
    2)根据移动时间窗框架下上一步计算出而未被完全执行的路径，将新到来的运输任务插入所有wAGV现有路径中可能的位置，计算插入代价，比较并选取最小代价值对应的路径和位置，将该任务插入得到更新后的路径，依次处理所有新到来的运输任务并得到更新后的初始路径；
    3)基于更新后的初始路径，采用基于禁忌搜索的启发式算法进一步改进初始路径，求解得到当前时间步下最终的调度方案，最终的调度方案为所有wAGV的路径；所述所有wAGV的路径包括所有wAGV将访问的码头节点序列，对应访问的时间、停留时间、装卸货数量；
    4)依据求解得到的所有wAGV的路径对wAGV实施调度，对于每个新的时间步，重复执行步骤2)-4)。
  2. 根据权利要求1所述的集装箱码头间全自主水上运输调度方法，其特征在于：
    所述动态调度模型的性能指标为：
    Figure PCTCN2020105407-appb-100001
    c 1，c 2，c 3为模型中三项性能指标的权重系数，其中，第一项最小化总行驶距离，第二项和第三项分别最小化等待和迟到时间；x ijv(k)为0-1整数变量，对于所有(i,j)
    Figure PCTCN2020105407-appb-100002
    如果wAGV v从节点i→j,则x ijv(k)＝1，否则x ijv(k)＝0，
    Figure PCTCN2020105407-appb-100003
    为拥有n v艘wAGV的船队集合；D ij为节点i→j间的水路距离；向量
    Figure PCTCN2020105407-appb-100004
    其中
    Figure PCTCN2020105407-appb-100005
    表示在节点i的等待时间，向量
    Figure PCTCN2020105407-appb-100006
    Figure PCTCN2020105407-appb-100007
    其中
    Figure PCTCN2020105407-appb-100008
    表示在节点i的迟到时间，‖·‖ 1表示1范数，
    Figure PCTCN2020105407-appb-100009
    为码头间调度运输网络的节点集，
    Figure PCTCN2020105407-appb-100010
    为码头间调度运输网络的边集。
  3. 根据权利要求2所述的集装箱码头间全自主水上运输调度方法，其特征在于：
    所述动态调度模型的约束条件包括：
    Figure PCTCN2020105407-appb-100011
    Figure PCTCN2020105407-appb-100012
    Figure PCTCN2020105407-appb-100013
    Figure PCTCN2020105407-appb-100014
    Figure PCTCN2020105407-appb-100015
    Figure PCTCN2020105407-appb-100016
    Figure PCTCN2020105407-appb-100017
    Figure PCTCN2020105407-appb-100018
    Figure PCTCN2020105407-appb-100019
    Figure PCTCN2020105407-appb-100020
    Figure PCTCN2020105407-appb-100021
    Figure PCTCN2020105407-appb-100022
    Figure PCTCN2020105407-appb-100023
    Figure PCTCN2020105407-appb-100024
    Figure PCTCN2020105407-appb-100025
    Figure PCTCN2020105407-appb-100026
    x ijv,z iv∈{0,1}, (18)
    其中，z iv(k)为0-1整数变量，对于所有
    Figure PCTCN2020105407-appb-100027
    如果wAGV v访问节点i,则z iv(k)＝1，否则z iv(k)＝0；定义
    Figure PCTCN2020105407-appb-100028
    为所有wAGVs的起始节点集，
    Figure PCTCN2020105407-appb-100029
    为所有wAGVs的终点节点集；其中，
    Figure PCTCN2020105407-appb-100030
    为取送运输任务数量，n d为仅送货任务数量；所有的取货节点定义为
    Figure PCTCN2020105407-appb-100031
    Figure PCTCN2020105407-appb-100032
    对应取货节点的送货节点为
    Figure PCTCN2020105407-appb-100033
    仅送货节点为
    Figure PCTCN2020105407-appb-100034
    u为wAGV巡航速度，T s为时间步的步长，k为离散的时间步序号，A i表示到达节点i的时间，y i表示到达节点i的wAGV上装载的集装箱数量，s i为所需码头服务时间，t i,min,t i,max分别为最早取货时间、最晚送达时间；Q为wAGV的最大装载量，下标max表示对应物理量的最大允许值，下标min表示对应物理量的最小允许值。
  4. 根据权利要求1所述的集装箱码头间全自主水上运输调度方法，其特征 在于：所述的根据移动时间窗框架下上一步计算出而未被完全执行的路径，将新到来的运输任务插入所有wAGV现有路径中可能的位置，计算插入代价，包括：
    对所有wAGV，定义其未完成的路径为
    Figure PCTCN2020105407-appb-100035
    其中r i,i＝1,…,m为wAGV v上一步规划中将要访问的节点，
    Figure PCTCN2020105407-appb-100036
    可能为空集；设新到来的运输任务为<p i,d i>，则将该任务插入所有wAGV现有路径中可能的位置，计算插入代价；对取货节点p i，将其插入某条路径
    Figure PCTCN2020105407-appb-100037
    的边<r i,r j>，如果装载量和时间窗约束条件均没有违反，则插入代价计算为：
    Figure PCTCN2020105407-appb-100038
    其中，
    Figure PCTCN2020105407-appb-100039
    表示节点r i到p i间的距离；对取货节点p i所有可能插入位置的代价进行升序排序，然后从最小的Δd p开始，同样的将送货节点d i插入对应取货节点p i点之后不违反装载量和时间窗约束条件的可能的位置，计算送货节点d i的插入代价计为Δd d；选取最小的Δd d值与当前对应的Δd p，则该新任务<p i,d i>插入路径
    Figure PCTCN2020105407-appb-100040
    的代价即为Δd p+Δd d，依次计算<p i,d i>插入所有wAGV当前路径的代价并比较选取最小代价值对应的路径和位置，将该任务插入得到更新后的路径；依次按到达时间顺序处理所有新到来的运输任务。
  5. 根据权利要求1所述的集装箱码头间全自主水上运输调度方法，其特征在于：所述的步骤3)包括：
    3.1)设计当前解的邻域空间；
    3.2)定义两个矩阵储存邻域搜索和迭代信息；首先，定义n×n维矩阵，其中n≤n v，其储存任意两条路径对之间经过邻域变换后得到的最优解信息及该路径对的序号，此序号为定义的第二个矩阵的行序号；第二个矩阵存储相应路径对的邻域变换信息，包括变换所涉及的两条路径ID、两条路径中变换的任务数量及其在路径中的位置；
    3.3)定义禁忌表，禁忌表包括禁忌对象和禁忌长度，禁忌对象定义为运输任务ID和路径ID，禁忌长度定位为当前迭代步数与一个预设的禁忌释放的步数值之和，当迭代步数小于禁忌长度时，则禁忌对象中的运输任务不能插入到对应路径中，以此避免迭代陷入循环；
    3.4)在每一次迭代中，选取代价下降最多的邻域变换，如果变换后的代价值比当前记录的最小代价值小且该变换不在禁忌表中，则记录该代价值为最小代价值并更新路径；注意每次迭代仅需要更新矩阵中有进行邻域变换的路径对应的 矩阵信息，不需要对所有路径的矩阵更新；
    3.5)重复上述3.1)-3.4)的步骤进行迭代，经过连续的邻域变化、路径更新以逐步提高调度方案的质量至近似最优；最终的调度方案为所有wAGV的路径，包括将访问的码头节点序列，对应访问的时间、停留时间、装卸货数量。
  6. 根据权利要求5所述的集装箱码头间全自主水上运输调度方法，其特征在于：所述的步骤3.1)为：
    设计两种邻域空间，第一种是对现有解中的任意两条路径，任意选取其中一条路径中的任意一组取货送货节点，先将其从原路径中删除，再将其插入另一条路径；插入另一条路径前需检测是否满足装载量约束条件，由于时间窗约束而产生的等待或迟到时间则按照动态调度模型中的权重与行驶距离一起计算插入代价；
    第二种邻域结构是对任意选取的两条路径中的一组取货送货节点进行交换，即先从两条路径中分别选取取货送货节点对，将其从原路径中删除，再插入到另一条路径中；插入另一条路径前需检测是否满足装载量约束条件，由于时间窗约束而产生的等待或迟到时间则按照动态调度模型中的权重与行驶距离一起计算插入代价。
  7. 一种集装箱码头间全自主水上运输调度系统，其特征在于包括wAGV系统、动态调度模型最初路径求解模块、初始路径构建模块、实时优化模块、调度模块；
    所述的wAGV系统包括多艘wAGV，其受调度模块控制执行运输任务；
    所述的动态调度模型最初路径求解模块，用于获取定义动态调度模型涉及到的参数，建立动态调度模型，确定模型的约束条件，并获取调度系统的第一步最初调度路径；
    所述的初始路径构建模块，根据移动时间窗框架下上一步计算出而未被完全执行的路径，将新到来的运输任务插入所有wAGV现有路径中可能的位置，计算插入代价并选取插入代价最小的路径及相应位置，比较选取最小代价值对应的路径和位置，将该任务插入得到更新后的路径，依次处理所有新到来的运输任务并得到更新后的初始路径；
    实时优化模块，采用基于禁忌搜索的启发式算法，结合初始路径构建模块得的初始路径以及动态调度模型的约束条件，得到当前移动时间窗框架下近似最优的调度结果，实现调度系统的实时化；
    调度模块，根据实时优化模块得到的当前移动时间窗框架下所有wAGV的路径，控制wAGV按照调度结果在特定的时间访问特定的码头，取或卸特定数量的集装箱，以近似最短行驶路径、最小等待和迟到时间的方式完成指派的运输任务。
PCT/CN2020/105407 2020-07-29 2020-07-29 一种集装箱码头间全自主水上运输调度方法及系统 WO2022021119A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/CN2020/105407 WO2022021119A1 (zh) 2020-07-29 2020-07-29 一种集装箱码头间全自主水上运输调度方法及系统
US17/408,480 US11886191B2 (en) 2020-07-29 2021-08-22 Scheduling method and system for fully autonomous waterborne inter terminal transportation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2020/105407 WO2022021119A1 (zh) 2020-07-29 2020-07-29 一种集装箱码头间全自主水上运输调度方法及系统

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US17/408,480 Continuation US11886191B2 (en) 2020-07-29 2021-08-22 Scheduling method and system for fully autonomous waterborne inter terminal transportation

Publications (1)

Publication Number Publication Date
WO2022021119A1 true WO2022021119A1 (zh) 2022-02-03

Family

ID=80003036

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/105407 WO2022021119A1 (zh) 2020-07-29 2020-07-29 一种集装箱码头间全自主水上运输调度方法及系统

Country Status (2)

Country Link
US (1) US11886191B2 (zh)
WO (1) WO2022021119A1 (zh)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114781706A (zh) * 2022-04-11 2022-07-22 东南大学 物流车辆配送调度方法、车辆、电子设备及存储介质
CN115423393A (zh) * 2022-08-12 2022-12-02 北京邮电大学 一种基于lstm的动态自适应调度周期的订单调度方法及装置
CN116736819A (zh) * 2023-08-14 2023-09-12 聊城大学 一种基于混合整数规划模型的fjsp-agv的建模方法
CN116777064A (zh) * 2023-06-20 2023-09-19 广东工业大学 基于非一刀切约束和分支定价算法的二维装箱方法
CN117361164A (zh) * 2023-12-07 2024-01-09 福建科盛智能物流装备有限公司 一种自动识别卸车方法

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114565195B (zh) * 2022-04-27 2022-08-19 苏州美集供应链管理股份有限公司 本地局部反馈与云端全局优化结合的调配系统与方法
CN115081674B (zh) * 2022-05-09 2023-04-18 香港理工大学深圳研究院 新型卡车排队行驶模式下的本地集装箱运输排版优化方法
CN115049334B (zh) * 2022-06-16 2023-04-07 四川省交通勘察设计研究院有限公司 一种水路运输需求统计分析方法、设备及存储介质
CN115865778B (zh) * 2023-02-17 2023-06-06 中国西安卫星测控中心 一种改进的Floyd路径搜索方法
CN116245257B (zh) * 2023-05-06 2023-09-12 季华实验室 一种多机器人调度方法及装置
CN116382099A (zh) * 2023-06-02 2023-07-04 上海数字大脑科技研究院有限公司 一种机器人路径调度规划方法及系统
CN117829713B (zh) * 2024-03-05 2024-05-10 深圳市久通物联科技股份有限公司 一种基于混合整数线性规划的成品油智慧运输方法

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103186879A (zh) * 2013-01-30 2013-07-03 广州智盈网络科技有限公司 公路运输调度方法
CN103413209A (zh) * 2013-07-17 2013-11-27 西南交通大学 多客户多仓库物流配送路径选择方法
CN104036380A (zh) * 2014-06-26 2014-09-10 广东工业大学 基于混沌免疫算法的多车型流水线物流运输调度方法
CN108665179A (zh) * 2018-05-17 2018-10-16 北京顺丰同城科技有限公司 物流调度方法、装置、电子设备以及计算机可读存储介质
CN108985597A (zh) * 2018-06-29 2018-12-11 华南理工大学 一种动态物流调度方法
US20190080271A1 (en) * 2017-09-11 2019-03-14 Hefei University Of Technology Coordinated Production and Transportation Scheduling Method and System Based on Improved Tabu Search Algorithm
CN111242368A (zh) * 2020-01-09 2020-06-05 中央财经大学 一种基于场景推演的物流配送调度算法

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11091211B1 (en) * 2017-12-22 2021-08-17 AI Incorporated Autonomous versatile vehicle system
GB2585680B (en) * 2019-07-10 2021-08-04 Maersk Line As Reefer power control
US11675324B2 (en) * 2019-09-19 2023-06-13 Bao Tran Air transportation systems and methods
US11334794B2 (en) * 2020-04-28 2022-05-17 Trabus Artificial-intelligence-based waterway information system

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103186879A (zh) * 2013-01-30 2013-07-03 广州智盈网络科技有限公司 公路运输调度方法
CN103413209A (zh) * 2013-07-17 2013-11-27 西南交通大学 多客户多仓库物流配送路径选择方法
CN104036380A (zh) * 2014-06-26 2014-09-10 广东工业大学 基于混沌免疫算法的多车型流水线物流运输调度方法
US20190080271A1 (en) * 2017-09-11 2019-03-14 Hefei University Of Technology Coordinated Production and Transportation Scheduling Method and System Based on Improved Tabu Search Algorithm
CN108665179A (zh) * 2018-05-17 2018-10-16 北京顺丰同城科技有限公司 物流调度方法、装置、电子设备以及计算机可读存储介质
CN108985597A (zh) * 2018-06-29 2018-12-11 华南理工大学 一种动态物流调度方法
CN111242368A (zh) * 2020-01-09 2020-06-05 中央财经大学 一种基于场景推演的物流配送调度算法

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114781706A (zh) * 2022-04-11 2022-07-22 东南大学 物流车辆配送调度方法、车辆、电子设备及存储介质
CN114781706B (zh) * 2022-04-11 2024-05-03 东南大学 物流车辆配送调度方法、车辆、电子设备及存储介质
CN115423393A (zh) * 2022-08-12 2022-12-02 北京邮电大学 一种基于lstm的动态自适应调度周期的订单调度方法及装置
CN116777064A (zh) * 2023-06-20 2023-09-19 广东工业大学 基于非一刀切约束和分支定价算法的二维装箱方法
CN116777064B (zh) * 2023-06-20 2024-03-01 广东工业大学 基于非一刀切约束和分支定价算法的二维装箱方法
CN116736819A (zh) * 2023-08-14 2023-09-12 聊城大学 一种基于混合整数规划模型的fjsp-agv的建模方法
CN116736819B (zh) * 2023-08-14 2023-10-20 聊城大学 一种基于混合整数规划模型的fjsp-agv的建模方法
CN117361164A (zh) * 2023-12-07 2024-01-09 福建科盛智能物流装备有限公司 一种自动识别卸车方法

Also Published As

Publication number Publication date
US11886191B2 (en) 2024-01-30
US20220035374A1 (en) 2022-02-03

Similar Documents

Publication Publication Date Title
WO2022021119A1 (zh) 一种集装箱码头间全自主水上运输调度方法及系统
CN112001064B (zh) 一种集装箱码头间全自主水上运输调度方法及系统
Lim et al. Airport gate scheduling with time windows
CN112418497B (zh) 一种面向制造物联的物料配送路径优化方法
Benantar et al. On the integration of container availability constraints into daily drayage operations arising in France: Modelling and optimization
CN113848970B (zh) 一种车辆-无人机多目标协同路径规划方法
CN113240234A (zh) 煤炭港口装船设备分配与船舶交通组织协调优化方法
Ibrahim et al. An improved genetic algorithm for vehicle routing problem pick-up and delivery with time windows
Siswanto et al. Maritime inventory routing problem with multiple time windows
CN114611794A (zh) 基于亚启发式算法的车机协同取送货路径优化方法和系统
Alaia et al. Optimization of the multi-depot & multi-vehicle pickup and delivery problem with time windows using genetic algorithm
Farahani et al. Online multimodal transportation planning using deep reinforcement learning
Wang et al. Optimization of ride-sharing with passenger transfer via deep reinforcement learning
CN113592248B (zh) 航空货运网络的规划方法、系统、电子设备及存储介质
Pal et al. SmartPorter: A combined perishable food and people transport architecture in smart urban areas
Zheng et al. Dynamic rolling horizon scheduling of waterborne AGVs for inter terminal transportation: Mathematical modeling and heuristic solution
CN113610453A (zh) 一种多运输方式联合的集装箱运输路径选择方法
JP2008168966A (ja) 運搬ルート生成装置、運搬ルート生成方法及び物流ネットワーク最適化装置
CN114663011A (zh) 一种自动化码头agv的调度方法
CN113741418A (zh) 异构车机编队协同路径的生成方法和装置
CN114706386A (zh) 车机协同取送货路径优化方法和系统
Chow et al. Adaptive scheduling of mixed bus services with flexible fleet size assignment under demand uncertainty
Van Heeswijk et al. Scalability and performance of decentralized planning in flexible transport networks
Teimoury et al. The sustainable hybrid truck-drone delivery model with stochastic customer existence
Cai et al. No-wait Drone Scheduling Traveling Salesman Problem Based on HAPGA

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20946979

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20946979

Country of ref document: EP

Kind code of ref document: A1