A method and system for intelligently scheduling container transportation routes considering time windows

Abstract

The present invention discloses an intelligent scheduling method and system for container transportation paths considering time windows, including: subdividing scheduling tasks into subtasks, forming scheduling groups based on transportation time clustering, determining the optimal multimodal transportation combination for each group, constructing a transportation network, and setting a dynamic time window based on priority. And evaluate uncertain factors through Monte Carlo simulation, simulate the transportation process, identify high-probability stagnation nodes with the probability of no waiting within the time window as an indicator, adjust the transportation route and resource allocation according to the situation that the transportation time in the high-probability stagnation node exceeds the time window range and the transportation priority, form a scheduling plan, predict changes in transportation demand, and continuously optimize the scheduling plan. Through the combination of intelligent scheduling algorithm and multimodal transportation, this method can achieve the optimal allocation of resources, improve the transportation efficiency of containers, reduce unnecessary waiting and transit time, reduce logistics costs, and has good interpretability.

Description

Technical Field

[0001] The present invention relates to the field of transport scheduling, and in particular to a method and system for intelligently scheduling container transport routes considering time windows. Background Art

[0002] With the continuous development of global trade, the container shipping industry is facing increasing transportation demand and complex transportation environment. Traditional container transportation scheduling methods often rely on manual experience and are difficult to cope with various uncertainties that arise during transportation, such as traffic congestion, weather changes, and delays in loading and unloading operations.

[0003] In the process of container transportation, the reasonable arrangement of time windows is of great significance to improving transportation efficiency and reducing transportation costs. However, traditional scheduling methods often ignore the dynamic adjustment of time windows and cannot flexibly adjust time windows according to actual operation progress and resource changes. In addition, when faced with complex transportation modes such as multimodal transport, traditional scheduling methods are difficult to achieve effective connection and coordination between different transportation modes, and cannot meet customers' requirements for transportation timeliness and reliability.

[0004] An intelligent scheduling method and system for container transportation routes taking time windows into consideration, which realizes dynamic optimization of container transportation routes and intelligent adjustment of time windows through optimization algorithms and intelligent decision-making models, realizes real-time monitoring and dynamic adjustment of the transportation process, optimizes the transportation process in all aspects, and improves logistics efficiency. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for intelligently scheduling container transportation routes taking time windows into consideration.

[0006] To achieve the above object, the present invention is implemented according to the following technical solutions:

[0007] A first aspect of the present invention provides a method for intelligently scheduling container transportation paths considering time windows, comprising:

[0008] S100 divides the scheduling task into subtasks, clusters each subtask based on the transportation time, and forms a scheduling group;

[0009] S200 determines a multi-modal transport combination of the optimal transport cost for each dispatch group in different road sections;

[0010] S300 constructs a transportation network according to the multimodal transportation combination, determines the time window of each node according to the transportation priority, and introduces a dynamic time window model to dynamically adjust the time window according to the operation progress of the node;

[0011] S400 uses Monte Carlo simulation to simulate the randomness of uncertain factors in transportation, and further simulates container transportation, taking the probability of no container waiting in the time window as an evaluation indicator to obtain high-probability stagnation nodes;

[0012] S500 adjusts subsequent transportation routes and scheduling resource allocation based on the situation that the transportation time in the high probability stagnation node exceeds the time window range and the transportation priority, and obtains a scheduling plan;

[0013] S600 predicts container transportation demand and continuously optimizes the scheduling plan based on the prediction result.

[0014] As a further method, the method of dividing the scheduling task into subtasks and clustering each subtask based on the transportation time to form scheduling groups includes:

[0015] According to the target geographical location and cargo type of the transportation task, the entire scheduling task is divided into multiple subtasks, each of which contains transportation tasks with the same target geographical location and cargo type characteristics;

[0016] For each subtask, the cluster center is set based on the average transportation time of historical subtasks, and the threshold of the transportation time interval where the deviation between each subtask and the average transportation time of historical subtasks is 30% and 60% is set. Clustering is performed to obtain scheduling groups.

[0017] As a further method, the method for determining the multimodal transport combination of the optimal transport cost of each dispatch group in different road sections includes:

[0018] Obtain the available transportation modes and transportation transfer points in each dispatch group;

[0019] For the road sections between each transport transfer point, the objective function is constructed by considering the optimal transport cost, and the expression is:

[0020]

[0021] in, and are the total number of transport transfer points and the total number of transport modes, For transfer point from transportation arrive Choose your shipping method The unit transportation cost, For transportation The transport volume, For transport transfer points From the mode of transportation Switch to transport mode The conversion time, For transportation The unit time cost, For transport transfer points From the mode of transportation Switch to transport mode conversion costs;

[0022] Under the premise of meeting the transportation time requirement, obtain the multimodal transportation combination that minimizes the objective function of transportation cost for each scheduling grouping.

[0023] As a further method, the method of constructing a transportation network according to the multimodal transportation combination and determining the time window of each node according to the transportation priority includes:

[0024] The transport starting point, transport transfer point and transport end point in the multimodal transport combination are taken as nodes, and the edges are the transport paths between nodes. The topological structure of the transport network is constructed, and the dynamic changes of the transport network are considered to update the status of the transport network in real time.

[0025] Determine the transportation priority of each type of container, the operating capacity of each node and resource limitation information;

[0026] According to the order of container priority from high to low, node time windows are allocated to high-priority containers first. The goal is to minimize the operation time of containers at the node. Under the constraints of each node, the time window of each node for each container is solved. The expression of the constraint condition of each node is:

[0027]

[0028] in, , , , and They are respectively represented as time window constraint, transportation time continuity constraint, node operation capacity constraint, waiting time constraint and operation cost constraint. and Respectively Container in the The start and end time of each node, and are the set of containers and the set of nodes in the transportation network, respectively. For the Container in the The start time of the node, The node is The next transport node of nodes, For Node to The transportation time of each node, For in time At that time, Container in the The job progress of each node, For the Container in the The workload of each node, For the The operating capacity of each node, is the operation time range of the node, For the Container in the The waiting time of each node, For the The dwell time of a container, For the Container in the The operation cost of each node, For the The fixed cost of each node is For the Unit activity variable cost of each node;

[0029] The mixed integer programming method is used to transform the time window solving problem into a mixed integer programming problem to solve the time window of each node.

[0030] As a further method, the method of introducing a dynamic time window model and dynamically adjusting the time window according to the job progress of the node includes:

[0031] Obtain the operation progress data of each node, including the completion percentage of loading and unloading operations, the operating status of equipment, the work efficiency of operators and the container handling volume at the current time point;

[0032] Predict future work progress and resource requirements based on the acquired work progress data;

[0033] The node states in the transportation network are defined as state space, the node time window adjustment strategy is used as action space, the state transition probability matrix is ​​constructed based on the prediction of the operation progress, and the Markov decision algorithm is selected to construct the dynamic time window model;

[0034] According to the changes in the predicted job progress, the time window of each node is updated through the dynamic time window model under the condition of meeting resource requirements.

[0035] As a further method, the method of simulating the randomness of uncertain factors in transportation using Monte Carlo simulation includes:

[0036] Conduct sensitivity analysis on project data during transportation to identify uncertainties;

[0037] Construct a probability distribution model for each uncertain factor, the expression is:

[0038]

[0039] in, is the value of the uncertain factor, is the normalization factor, is the number of Gaussian distributions in the uncertain factors, For the The weight coefficient of the Gaussian distribution, and Respectively The standard deviation and mean of a Gaussian distribution, is the number of Poisson distribution correlations in the uncertain factors, For the The weight coefficient of the Poisson distribution related terms, For the The frequency coefficient of the Poisson distribution related terms, For the The offset of the Poisson distribution related terms;

[0040] According to the probability model of uncertain factors, Monte Carlo simulation is carried out to generate a large number of random samples to simulate various possible situations of uncertain factors in the transportation process.

[0041] As a further method, the method of further performing container transportation simulation and taking the probability of no container waiting in the time window as an evaluation index to obtain a high probability stagnation node includes:

[0042] The various possible situations of uncertain factors in the transportation process are used as simulation scenarios. In each simulation scenario, container transportation simulation is carried out according to the set transportation process and scheduling rules;

[0043] For each node in the transportation network, count the proportion of the number of times there is no container waiting in the time window in all simulation scenarios to the total number of simulation scenarios;

[0044] Nodes with a probability of no container waiting in the time window lower than the average are identified as high-probability stagnation nodes.

[0045] As a further method, the method for adjusting subsequent transportation routes and scheduling resource allocation based on the situation that the transportation time in the high probability stagnation node exceeds the time window range and the transportation priority includes:

[0046] For all containers whose transportation time exceeds the time window range in the high-probability stagnation node, the time window is re-planned with the goal of minimizing the transportation time. If the re-planned time window causes transportation timeout, the time windows of the overdue containers are re-sorted in order of transportation priority from high to low;

[0047] For containers that have not yet arrived at a high-probability stagnation node, the transportation route is replanned at the previous node in the transportation route to skip the high-probability stagnation node;

[0048] Obtain currently available scheduling resources, and obtain a strategy set for allocating scheduling resources under the conditions of satisfying the time window and transportation path after replanning;

[0049] In order to minimize the overall transportation cost, the optimal strategy is selected from the strategy set as the scheduling solution.

[0050] As a further method, the method of predicting container transportation demand and continuously optimizing the scheduling scheme based on the prediction result includes:

[0051] Predict the growth of container transportation demand, obtain the scheduling plan based on the forecast results, calculate the deviation between the scheduling plan based on the forecast results and the current scheduling plan, and determine the adjustment trend of the current scheduling plan;

[0052] Each node in the transportation network is regarded as an intelligent agent, and a data communication protocol is built for each node. Through the multi-agent collaborative algorithm, autonomous decisions are made based on the data of node interaction and the scheduling trend of the current scheduling plan to obtain the scheduling strategy of each node.

[0053] A second aspect of the present invention provides an intelligent scheduling system for container transportation paths considering time windows, comprising:

[0054] The subtask division module is used to divide the scheduling task into subtasks, cluster each subtask based on the transportation time, and form a scheduling group;

[0055] A transport mode combination module, used to determine a multi-modal transport combination of the optimal transport cost of each dispatch group in different road sections;

[0056] A time window determination module is used to construct a transportation network according to the multimodal transportation combination, determine the time window of each node according to the transportation priority, and introduce a dynamic time window model to dynamically adjust the time window according to the operation progress of the node;

[0057] A simulation evaluation module is used to simulate the randomness of uncertain factors in transportation by Monte Carlo simulation, and further simulate container transportation, taking the probability of no container waiting in the time window as an evaluation index to obtain high probability stagnation nodes;

[0058] A path resource adjustment module, used to adjust subsequent transportation routes and scheduling resource allocation to obtain a scheduling solution based on the situation that the transportation time in the high probability stagnation node exceeds the time window range and the transportation priority;

[0059] The prediction and optimization module is used to predict the container transportation demand and continuously optimize the scheduling plan based on the prediction results.

[0060] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects:

[0061] (1) The present invention introduces a dynamic time window model by allocating node time windows. By collecting node operation progress data, including the percentage of completion of loading and unloading, equipment operating status, personnel work efficiency and container handling volume, the future operation trend and resource demand are predicted, and the time window is dynamically adjusted. This dynamic adaptability can ensure the smoothness of the transportation process, ensure transportation timeliness, and reduce transportation delay rate;

[0062] (2) The present invention simulates uncertain factors, constructs a probability model including Gaussian and Poisson distributions to simulate their randomness, and then simulates container transportation. Subsequently, the transportation routes and scheduling resources are adjusted according to the transportation time exceeding the window and priority of the high-probability stagnation node, which can reduce transportation risks and ensure transportation stability.

[0063] (3) The present invention predicts container transportation demand and continuously optimizes the scheduling plan based on the prediction results, so as to quickly respond to market demand fluctuations, reasonably allocate resources during peak demand periods, and improve transportation resource utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 The present invention is a flowchart of the steps of an intelligent method for scheduling container transportation routes taking into account time windows in an embodiment of the present invention. DETAILED DESCRIPTION

[0065] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0066] Reference Figure 1 As shown, the present invention provides an intelligent scheduling method for container transportation paths considering time windows, comprising:

[0067] S100 divides the scheduling task into subtasks, clusters each subtask based on the transportation time, and forms a scheduling group;

[0068] In actual evaluation, the invention is applied to the logistics transportation of a certain port. The port received 63 container transportation instructions on working days. According to the cargo destination information and cargo type information, these transportation tasks were divided into 3 subtasks, among which subtask 1 was 12 containers shipped to A and the cargo type was electronic products; subtask 2 was 20 containers shipped to B and the cargo type was clothing and textiles; subtask 3 was 31 containers shipped to C and the cargo type was mechanical parts. The historical transportation data accumulated by the port was used, among which for subtask 1, the average transportation time was 12 days. A clustering algorithm was used, and the transportation time was used as the clustering dimension to process subtask 1 to form 4 scheduling groups. Group 1 contained 4 containers with a transportation time of 8.4-10.5 days; Group 2 contained 3 containers with a transportation time of 10.5-13.5 days; Group 3 contained 1 container with a transportation time of 13.5-16.5 days; Group 4 contained 6 containers with a transportation time of 16.5-19.2 days.

[0069] S200 determines a multi-modal transport combination of the optimal transport cost for each dispatch group in different road sections;

[0070] In the actual evaluation, for the scheduling group 1 of subtask 1, the feasible transportation modes are determined to include sea transportation, rail transportation and road transportation. The transportation transfer points are five hub cities D, E, F, G, and H. The objective function is constructed to determine the combination of transportation modes. The multimodal transportation combination of scheduling group 1 is specifically as follows: depart from the port and arrive at the transfer point D by sea, switch to rail transportation at the transfer point D to the transfer point G, and finally arrive at the final destination by road transportation from the transfer point G.

[0071] S300 constructs a transportation network according to the multimodal transportation combination, determines the time window of each node according to the transportation priority, and introduces a dynamic time window model to dynamically adjust the time window according to the operation progress of the node;

[0072] It should be explained that by taking the transportation starting point, transit point and end point as nodes and the transportation paths between nodes as edges, a dynamic topological structure can be formed to better simulate and predict what may happen during the transportation process. In addition, the introduction of the dynamic time window model is to make the scheduling scheme more flexible and to adjust the time window according to the real-time operation progress, thereby reducing the waiting time of containers at the nodes and improving the overall transportation efficiency. The creativity of this method lies in its ability to respond to changes in the transportation process in real time, rather than relying on a static schedule, thereby improving the adaptability and efficiency of the transportation system;

[0073] In the actual evaluation, based on the transportation path of multimodal transport combination planning, the port of departure, transit point and terminal destination are taken as network nodes, and the transportation routes connecting the nodes are defined as network edges. The transportation network topology architecture is constructed, and the priority order of each type of container is determined. For the D transit point, the time window is specifically as follows: for scheduling group 1, start time: 09:00 on the 8th day, end time: 16:00 on the 9th day; the dynamic time window model is introduced, and a specific time window of scheduling group 1 is optimized as follows: start time: 07:00 on the 7th day, end time: 18:00 on the 9th day.

[0074] S400 uses Monte Carlo simulation to simulate the randomness of uncertain factors in transportation, and further simulates container transportation, taking the probability of no container waiting in the time window as an evaluation indicator to obtain high-probability stagnation nodes;

[0075] It should be explained that the Monte Carlo simulation is used to simulate the randomness of uncertainties in transportation in order to quantify and evaluate the impact of these uncertainties on the container transportation process. Through simulation, a large number of random samples can be generated, which represent various situations that may occur during transportation, such as traffic congestion, weather changes, and fluctuations in loading and unloading time. These simulation results help predict the probability of a container arriving within the time window, thereby evaluating stagnation points in the transportation network. Through further container transportation simulation, the probability of no container waiting within the time window is used as an evaluation indicator to identify high-probability stagnation nodes, that is, those nodes that frequently experience delays in the simulation. The purpose of this step is to identify and alleviate possible transportation bottlenecks in advance, optimize scheduling plans, and reduce uncertainties and delays in the actual transportation process;

[0076] In the actual evaluation, the uncertain factors are determined to be weather conditions, failure rate of transportation vehicles, genetic fluctuations in port loading and unloading efficiency, etc. For weather conditions, there are three Gaussian distributions with weight coefficients of 0.5, 0.3, and 0.2, respectively. The corresponding standard deviations are 2.5, 3.2, and 1.8, and the means are 6, 8, and 4, respectively. There are two Poisson distribution correlations with weight coefficients of 0.4 and 0.6, and frequency coefficients of 0.25 and 0.35. Further, based on the Monte Carlo simulation of the uncertainties, a total of 500 simulation scenarios were used, and then the transportation network simulation was carried out. For the D transfer point, the frequency of no container waiting within the statistical time window accounted for 28% of the total number of simulation scenarios, which was lower than the node average of 43%. The D transfer point was marked as a high-probability stagnation node.

[0077] S500 adjusts subsequent transportation routes and scheduling resource allocation based on the situation that the transportation time in the high probability stagnation node exceeds the time window range and the transportation priority, and obtains a scheduling plan;

[0078] It should be explained that the purpose of adjusting subsequent transportation routes and scheduling resource allocation is to solve the problem of transportation time exceeding the time window caused by high-probability stagnation nodes, and to take into account transportation priority. The creativity of this step lies in its ability to dynamically adjust the transportation plan based on real-time data and forecast results to adapt to changing transportation conditions and needs. By re-planning the time window for containers whose transportation time exceeds the time window, and re-planning the transportation route for containers that have not yet arrived at high-probability stagnation nodes, transportation delays can be effectively reduced and transportation efficiency can be improved;

[0079] In the actual evaluation, for the D transfer point, the number of containers whose transportation time exceeds the established time window is 7. While meeting the transportation time requirements, the time window is re-planned, and 3 containers are still overdue, namely R1, R2 and R3. These three containers are sorted in order of transportation priority, and the order is: R1, R3, R2; for the 5 containers that did not arrive at the D transfer point subsequently, the path is re-planned at the previous node in the transportation path, the starting node port position, skipping the D transfer point, and finding the lowest cost and best timeliness alternative path while meeting the transportation time requirements; under the dual constraints of the re-planned time window and the transportation path, a multi-element scheduling resource allocation strategy set is generated through an intelligent optimization algorithm, and the strategy set with the best overall transportation cost is selected from the strategy set as the scheduling method.

[0080] S600 predicts container transportation demand and continuously optimizes the scheduling plan based on the prediction result.

[0081] In the actual evaluation, the forecasting model was used to predict that the demand for container transportation would increase by 12% in the next week, and then the scheduling plan of the forecast result was obtained. The deviation value between the scheduling plan of the forecast result and the current scheduling plan was calculated, and the adjustment trend of the current scheduling plan was determined. Based on the scheduling strategy, the decision was made through the multi-agent optimization algorithm to obtain the scheduling optimization strategy for each node.

[0082] Specifically, after three months of operation, it was found that the average container transportation time was significantly shortened by 2.4%, the transportation cost was reduced by 5%, and the frequency of high-probability stagnation nodes was reduced by 4.6%.

[0083] In this implementation, the method of dividing the scheduling task into subtasks and clustering each subtask based on the transportation time to form scheduling groups includes:

[0084] According to the target geographical location and cargo type of the transportation task, the entire scheduling task is divided into multiple subtasks, each of which contains transportation tasks with the same target geographical location and cargo type characteristics;

[0085] For each subtask, the cluster center is set based on the average transportation time of historical subtasks, and the threshold of the transportation time interval where the deviation between each subtask and the average transportation time of historical subtasks is 30% and 60% is set. Clustering is performed to obtain scheduling groups.

[0086] In this embodiment, the method for determining the multimodal transport combination of the optimal transport cost of each dispatch group in different road sections includes:

[0087] Obtain the available transportation modes and transportation transfer points in each dispatch group;

[0088] For the road sections between each transport transfer point, the objective function is constructed by considering the optimal transport cost, and the expression is:

[0089]

[0090] in, and are the total number of transport transfer points and the total number of transport modes, For transfer point from transportation arrive Choose your shipping method The unit transportation cost, For transportation The transport volume, For transport transfer points From the mode of transportation Switch to transport mode The conversion time, For transportation The unit time cost, For transport transfer points From the mode of transportation Switch to transport mode conversion costs;

[0091] Under the premise of meeting the transportation time requirement, obtain the multimodal transportation combination that minimizes the objective function of transportation cost for each scheduling grouping.

[0092] In this implementation, the method of constructing a transportation network according to the multimodal transportation combination and determining the time window of each node according to the transportation priority includes:

[0093] The transport starting point, transport transfer point and transport end point in the multimodal transport combination are taken as nodes, and the edges are the transport paths between nodes. The topological structure of the transport network is constructed, and the dynamic changes of the transport network are considered to update the status of the transport network in real time.

[0094] Determine the transportation priority of each type of container, the operating capacity of each node and resource limitation information;

[0095] According to the order of container priority from high to low, node time windows are allocated to high-priority containers first. The goal is to minimize the operation time of containers at the node. Under the constraints of each node, the time window of each node for each container is solved. The expression of the constraint condition of each node is:

[0096]

[0097] in, , , , and They are respectively represented as time window constraint, transportation time continuity constraint, node operation capacity constraint, waiting time constraint and operation cost constraint. and Respectively Container in the The start and end time of each node, and are the set of containers and the set of nodes in the transportation network, respectively. For the Container in the The start time of the node, The node is The next transport node of nodes, For Node to The transportation time of each node, For in time At that time, Container in the The job progress of each node, For the Container in the The workload of each node, For the The operating capacity of each node, is the operation time range of the node, For the Container in the The waiting time of each node, For the The dwell time of a container, For the Container in the The operation cost of each node, For the The fixed cost of each node is For the Unit activity variable cost of each node;

[0098] The mixed integer programming method is used to transform the time window solving problem into a mixed integer programming problem to solve the time window of each node.

[0099] In this implementation, the dynamic time window model is introduced, and the method for dynamically adjusting the time window according to the job progress of the node includes:

[0100] Obtain the operation progress data of each node, including the completion percentage of loading and unloading operations, the operating status of equipment, the work efficiency of operators and the container handling volume at the current time point;

[0101] Predict future work progress and resource requirements based on the acquired work progress data;

[0102] The node states in the transportation network are defined as state space, the node time window adjustment strategy is used as action space, the state transition probability matrix is ​​constructed based on the prediction of the operation progress, and the Markov decision algorithm is selected to construct the dynamic time window model;

[0103] According to the changes in the predicted job progress, the time window of each node is updated through the dynamic time window model under the condition of meeting resource requirements.

[0104] In this embodiment, the method of simulating the randomness of uncertain factors in transportation by using Monte Carlo simulation includes:

[0105] Conduct sensitivity analysis on project data during transportation to identify uncertainties;

[0106] Construct a probability distribution model for each uncertain factor, the expression is:

[0107]

[0108] in, is the value of the uncertain factor, is the normalization factor, is the number of Gaussian distributions in the uncertain factors, For the The weight coefficient of the Gaussian distribution, and Respectively The standard deviation and mean of a Gaussian distribution, is the number of Poisson distribution correlations in the uncertain factors, For the The weight coefficient of the Poisson distribution related terms, For the The frequency coefficient of the Poisson distribution related terms, For the The offset of the Poisson distribution related terms;

[0109] According to the probability model of uncertain factors, Monte Carlo simulation is carried out to generate a large number of random samples to simulate various possible situations of uncertain factors in the transportation process.

[0110] In this embodiment, the method of further performing container transportation simulation and taking the probability of no container waiting in the time window as an evaluation index to obtain a high probability stagnation node includes:

[0111] The various possible situations of uncertain factors in the transportation process are used as simulation scenarios. In each simulation scenario, container transportation simulation is carried out according to the set transportation process and scheduling rules;

[0112] For each node in the transportation network, count the proportion of the number of times there is no container waiting in the time window in all simulation scenarios to the total number of simulation scenarios;

[0113] Nodes with a probability of no container waiting in the time window lower than the average are identified as high-probability stagnation nodes.

[0114] In this embodiment, the method for adjusting the subsequent transportation route and scheduling resource allocation based on the situation that the transportation time in the high probability stagnation node exceeds the time window range and the transportation priority includes:

[0115] For all containers whose transportation time exceeds the time window range in the high-probability stagnation node, the time window is re-planned with the goal of minimizing the transportation time. If the re-planned time window causes transportation timeout, the time windows of the overdue containers are re-sorted in order of transportation priority from high to low;

[0116] For containers that have not yet arrived at a high-probability stagnation node, the transportation route is replanned at the previous node in the transportation route to skip the high-probability stagnation node;

[0117] Obtain currently available scheduling resources, and obtain a strategy set for allocating scheduling resources under the conditions of satisfying the time window and transportation path after replanning;

[0118] In order to minimize the overall transportation cost, the optimal strategy is selected from the strategy set as the scheduling solution.

[0119] In this embodiment, the method of predicting container transportation demand and continuously optimizing the scheduling plan based on the prediction result includes:

[0120] Predict the growth of container transportation demand, obtain the scheduling plan based on the forecast results, calculate the deviation between the scheduling plan based on the forecast results and the current scheduling plan, and determine the adjustment trend of the current scheduling plan;

[0121] Each node in the transportation network is regarded as an intelligent agent, and a data communication protocol is built for each node. Through the multi-agent collaborative algorithm, autonomous decisions are made based on the data of node interaction and the scheduling trend of the current scheduling plan to obtain the scheduling strategy of each node.

[0122] The second aspect of the present invention further provides an intelligent scheduling system for container transportation paths considering time windows, comprising:

[0123] The subtask division module is used to divide the scheduling task into subtasks, cluster each subtask based on the transportation time, and form a scheduling group;

[0124] A transport mode combination module, used to determine a multi-modal transport combination of the optimal transport cost of each dispatch group in different road sections;

[0125] A time window determination module is used to construct a transportation network according to the multimodal transportation combination, determine the time window of each node according to the transportation priority, and introduce a dynamic time window model to dynamically adjust the time window according to the operation progress of the node;

[0126] A simulation evaluation module is used to simulate the randomness of uncertain factors in transportation by Monte Carlo simulation, and further simulate container transportation, taking the probability of no container waiting in the time window as an evaluation index to obtain high probability stagnation nodes;

[0127] A path resource adjustment module, used to adjust subsequent transportation routes and scheduling resource allocation to obtain a scheduling solution based on the situation that the transportation time in the high probability stagnation node exceeds the time window range and the transportation priority;

[0128] The prediction and optimization module is used to predict the container transportation demand and continuously optimize the scheduling plan based on the prediction results.

[0129] The above contents are merely examples and explanations of the structure of the present invention. The technicians in this technical field may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the claims, they should all fall within the protection scope of the present invention.

Claims

1. An intelligent scheduling method for container transportation routes considering time windows, characterized in that: The following steps are involved: Divide the scheduling task into subtasks, cluster each subtask based on the transportation time, and form scheduling groups; Determine a multi-modal transport combination of optimal transport costs for each of the dispatch groups on different road sections; Constructing a transportation network according to the multimodal transportation combination, determining the time window of each node according to the transportation priority, and introducing a dynamic time window model to dynamically adjust the time window according to the operation progress of the node; Monte Carlo simulation is used to simulate the randomness of uncertain factors in transportation, and container transportation simulation is further performed. The probability of no container waiting in the time window is used as an evaluation index to obtain high-probability stagnation nodes; Based on the situation that the transportation time in the high-probability stagnation node exceeds the time window range and the transportation priority, the subsequent transportation route and scheduling resource allocation are adjusted to obtain a scheduling plan; Predicting container transportation demand and continuously optimizing the scheduling plan based on the prediction results; The method of constructing a transportation network according to the multimodal transportation combination and determining the time window of each node according to the transportation priority includes: The transport starting point, transport transfer point and transport end point in the multimodal transport combination are taken as nodes, and the edges are the transport paths between nodes. The topological structure of the transport network is constructed, and the dynamic changes of the transport network are considered to update the status of the transport network in real time. Determine the transportation priority of each type of container, the operating capacity of each node and resource limitation information; According to the order of container priority from high to low, node time windows are allocated to high-priority containers first. The goal is to minimize the operation time of containers at the node. Under the constraints of each node, the time window of each node for each container is solved. The expression of the constraint condition of each node is: in, , , , and They are respectively represented as time window constraint, transportation time continuity constraint, node operation capacity constraint, waiting time constraint and operation cost constraint. and Respectively Container in the The start and end time of each node, and are the set of containers and the set of nodes in the transportation network, respectively. For the Container in the The start time of the node, The node is The next transport node of nodes, For Node to The transportation time of each node, For in time At that time, Container in the The job progress of each node, For the Container in the The workload of each node, For the The operating capacity of each node, is the operation time range of the node, For the Container in the The waiting time of each node, For the The dwell time of a container, For the Container in the The operation cost of each node, For the The fixed cost of each node is For the Unit activity variable cost of each node; The mixed integer programming method is used to transform the time window solving problem into a mixed integer programming problem to solve the time window of each node.

2. The method for intelligently scheduling container transportation routes considering time windows according to claim 1 is characterized in that: The method of dividing the scheduling task into subtasks and clustering each subtask based on the transportation time to form scheduling groups includes: According to the target geographical location and cargo type of the transportation task, the entire scheduling task is divided into multiple subtasks, each of which contains transportation tasks with the same target geographical location and cargo type characteristics; For each subtask, the cluster center is set based on the average transportation time of historical subtasks, and the threshold of the transportation time interval where the deviation between each subtask and the average transportation time of historical subtasks is 30% and 60% is set. Clustering is performed to obtain scheduling groups.

3. The method for intelligently scheduling container transportation routes considering time windows according to claim 1 is characterized in that: The method for determining the multimodal transport combination of the optimal transport cost of each dispatch group in different road sections comprises: Obtain the available transportation modes and transportation transfer points in each dispatch group; For the road sections between each transport transfer point, the objective function is constructed by considering the optimal transport cost, and the expression is: in, and are the total number of transport transfer points and the total number of transport modes, For transfer point from transportation arrive Choose your shipping method The unit transportation cost, For transportation The transport volume, For transport transfer points From the mode of transportation Switch to transport mode The conversion time, For transportation The unit time cost, For transport transfer points From the mode of transportation Switch to transport mode conversion costs; Under the premise of meeting the transportation time requirement, obtain the multimodal transportation combination that minimizes the objective function of transportation cost for each scheduling grouping.

4. The method for intelligently scheduling container transportation routes considering time windows according to claim 1 is characterized in that: The method of introducing a dynamic time window model and dynamically adjusting the time window according to the job progress of the node includes: Obtain the operation progress data of each node, including the completion percentage of loading and unloading operations, the operating status of equipment, the work efficiency of operators and the container handling volume at the current time point; Predict future work progress and resource requirements based on the acquired work progress data; The node states in the transportation network are defined as state space, the node time window adjustment strategy is used as action space, the state transition probability matrix is ​​constructed based on the prediction of the operation progress, and the Markov decision algorithm is selected to construct the dynamic time window model; According to the changes in the predicted job progress, the time window of each node is updated through the dynamic time window model under the condition of meeting resource requirements.

5. The method for intelligently scheduling container transportation routes considering time windows according to claim 1, characterized in that: The method of simulating the randomness of uncertain factors in transportation by using Monte Carlo simulation includes: Conduct sensitivity analysis on project data during transportation to identify uncertainties; Construct a probability distribution model for each uncertain factor, the expression is: in, is the value of the uncertain factor, is the normalization factor, is the number of Gaussian distributions in the uncertain factors, For the The weight coefficient of the Gaussian distribution, and Respectively The standard deviation and mean of a Gaussian distribution, is the number of Poisson distribution correlations in the uncertain factors, For the The weight coefficient of the Poisson distribution related terms, For the The frequency coefficient of the Poisson distribution related terms, For the The offset of the Poisson distribution related terms; According to the probability model of uncertain factors, Monte Carlo simulation is carried out to generate a large number of random samples to simulate various possible situations of uncertain factors in the transportation process.

6. The method for intelligently scheduling container transportation routes considering time windows according to claim 1, characterized in that: The method of further performing container transportation simulation and taking the probability of no container waiting in the time window as an evaluation index to obtain a high probability stagnation node includes: The various possible situations of uncertain factors in the transportation process are used as simulation scenarios. In each simulation scenario, container transportation simulation is carried out according to the set transportation process and scheduling rules; For each node in the transportation network, count the proportion of the number of times there is no container waiting in the time window in all simulation scenarios to the total number of simulation scenarios; Nodes with a probability of no container waiting in the time window lower than the average are identified as high-probability stagnation nodes.

7. The method for intelligently scheduling container transportation routes considering time windows according to claim 1, characterized in that: The method for adjusting subsequent transportation routes and scheduling resource allocation based on the situation that the transportation time in the high probability stagnation node exceeds the time window range and the transportation priority includes: For all containers whose transportation time exceeds the time window range in the high-probability stagnation node, the time window is re-planned with the goal of minimizing the transportation time. If the re-planned time window causes transportation timeout, the time windows of the overdue containers are re-sorted in order of transportation priority from high to low; For containers that have not yet arrived at a high-probability stagnation node, the transportation route is replanned at the previous node in the transportation route to skip the high-probability stagnation node; Obtain currently available scheduling resources, and obtain a strategy set for allocating scheduling resources under the conditions of satisfying the time window and transportation path after replanning; In order to minimize the overall transportation cost, the optimal strategy is selected from the strategy set as the scheduling solution.

8. The method for intelligently scheduling container transportation routes considering time windows according to claim 1, characterized in that: The method of predicting container transportation demand and continuously optimizing the scheduling scheme based on the prediction result includes: Predict the growth of container transportation demand, obtain the scheduling plan based on the forecast results, calculate the deviation between the scheduling plan based on the forecast results and the current scheduling plan, and determine the adjustment trend of the current scheduling plan; Each node in the transportation network is regarded as an intelligent agent, and a data communication protocol is built for each node. Through the multi-agent collaborative algorithm, autonomous decisions are made based on the data of node interaction and the scheduling trend of the current scheduling plan to obtain the scheduling strategy of each node.

9. An intelligent scheduling system for container transportation routes considering time windows, used to execute an intelligent scheduling method for container transportation routes considering time windows as claimed in any one of claims 1 to 8, characterized in that: The system comprises: The subtask division module is used to divide the scheduling task into subtasks, cluster each subtask based on the transportation time, and form a scheduling group; A transport mode combination module, used to determine a multi-modal transport combination of the optimal transport cost of each dispatch group in different road sections; A time window determination module is used to construct a transportation network according to the multimodal transportation combination, determine the time window of each node according to the transportation priority, and introduce a dynamic time window model to dynamically adjust the time window according to the operation progress of the node; A simulation evaluation module is used to simulate the randomness of uncertain factors in transportation by Monte Carlo simulation, and further simulate container transportation, taking the probability of no container waiting in the time window as an evaluation index to obtain high probability stagnation nodes; A path resource adjustment module, used to adjust subsequent transportation routes and scheduling resource allocation to obtain a scheduling solution based on the situation that the transportation time in the high probability stagnation node exceeds the time window range and the transportation priority; The prediction and optimization module is used to predict the container transportation demand and continuously optimize the scheduling plan based on the prediction results.

Images