International logistics chain transportation method and system

Abstract

The invention relates to the technical field of international logistics, and discloses an international logistics chain transportation method and system, and the method comprises the following steps: S1, collecting logistics environment information based on real-time environment monitoring data, including weather conditions, traffic flow, flight time, port conditions, current positions of goods, and states of transportation tools; s2, cleaning, standardizing and fusing the acquired environment data to obtain available real-time data; and S3, based on the multi-objective optimization model, comprehensively considering timeliness, cost, resource utilization rate and environmental influence, and carrying out link splitting and path selection optimization to obtain an optimal path selection scheme. Link splitting and path selection optimization are carried out based on the multi-objective optimization model, and then global balance among timeliness, transportation cost, resource utilization rate and environmental influence is realized, so that the problem that a path scheme does not have overall optimality due to the fact that most of traditional path design methods only pay attention to a single objective is solved.

Description

Technical Field

[0001] This invention relates to the field of international logistics technology, specifically to a chain transportation method and system for international logistics. Background Technology

[0002] With the development of cross-border e-commerce, global supply chains, and multimodal transport, international logistics and transportation face challenges such as complex routes, volatile environments, and limited transportation resources. Traditional international logistics route planning methods rely heavily on static rules and experience-based decisions, lacking the ability to respond in real time to dynamic environmental changes. Furthermore, they cannot simultaneously consider transportation timeliness, cost control, resource utilization, and environmental protection requirements, resulting in low route execution efficiency, unreasonable resource scheduling, and seriously affecting the stability and controllability of the overall transportation system.

[0003] Currently, most traditional path design methods only focus on a single objective. Because they fail to consider the trade-offs between multiple objectives, the path schemes do not have overall optimality. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a chain transportation method and system for international logistics, which improves upon the problem that traditional route design methods mostly focus on a single objective and fail to consider the trade-offs between multiple objectives, resulting in route schemes that lack overall optimality.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a chain transportation method for international logistics, comprising the following steps: S1. Collect logistics environment information based on real-time environmental monitoring data, including weather conditions, traffic flow, flight schedules, port conditions, current location of goods, and status of transport vehicles; S2. Clean, standardize, and fuse the collected environmental data to obtain usable real-time data; S3. Based on the multi-objective optimization model, taking into account timeliness, cost, resource utilization and environmental impact, the chain splitting and path selection optimization are carried out to obtain the optimal path selection scheme. S4. Adaptive reinforcement learning algorithm is adopted to adjust the path selection and chain splitting in real time based on environmental feedback and historical data, thereby optimizing transportation timeliness and resource consumption. S5. Perform constraint optimization on the path selection and chain splitting based on the Lagrange multiplier method to ensure that the constraint conditions are met; S6. Based on the optimal route selection and chain splitting scheme, perform automated scheduling of transportation vehicles, personnel and resources; S7. Use IoT technology to monitor the transportation process in real time to ensure that the transportation route is executed in accordance with the predetermined plan; S8. Based on feedback information from the execution layer, adjust and optimize the scheduling scheme in real time.

[0006] By adopting the above technical solution, chain splitting and route selection optimization are performed based on a multi-objective optimization model, thereby achieving a global balance between timeliness, transportation cost, resource utilization and environmental impact. This improves the problem that traditional route design methods mostly focus on a single objective and fail to consider the trade-offs between multiple objectives, resulting in route solutions that do not have overall optimality.

[0007] Preferably, in step S1, the logistics environment information is obtained in real time through interface with an external platform, which includes a traffic management platform, a weather forecast system, a flight management system, and a port management system.

[0008] Preferably, step S2 includes removing redundant data, filling in missing values, and merging data from different sources to ensure data consistency and integrity. Furthermore, data standardization technology is used to integrate data from different sources with the same format and units.

[0009] Preferably, in step S3, the Pareto optimal solution algorithm is used to perform objective trade-offs, ensuring that the best balance is found among multiple objectives such as timeliness, cost, resource utilization and environmental impact, and outputting the optimal path selection scheme.

[0010] Preferably, in step S4, the Q-learning algorithm is used to learn strategies for path selection and chain splitting in a dynamic environment, optimize the reward function based on real-time feedback, and gradually improve the decision-making accuracy of the system through multiple rounds of trials.

[0011] Preferably, in step S5, the Lagrange multiplier method is used to take transportation timeliness, resource utilization rate, and environmental impact as constraints, and the optimal route selection scheme is solved according to the constraints to ensure that the scheme meets all preset resource and environmental limitations in actual execution.

[0012] Preferably, in step S6, a rule-based decision tree algorithm is used to automatically allocate transportation vehicles, arrange personnel and resources according to the optimal path selection and chain splitting scheme, and dynamically adjust each link in the logistics scheduling process to optimize transportation efficiency.

[0013] Preferably, in step S7, IoT technology, including RFID tags and GPS positioning technology, is used to monitor the current location and transportation status of the transport vehicle in real time, ensuring the traceability of the transportation process and sending alarms to the system in a timely manner in case of abnormalities.

[0014] Preferably, in step S8, based on the execution layer feedback mechanism, by receiving real-time feedback information during the transportation process, a rolling optimization strategy is adopted to dynamically adjust the path selection, chain splitting, and resource scheduling to ensure continuous optimization of the transportation process.

[0015] An international logistics chain transportation system includes the following modules: The data acquisition module is used to collect logistics environment information in real time; The data processing module is used to clean, standardize, and integrate the collected environmental data; A multi-objective optimization module is used to optimize chain splitting and path selection based on timeliness, cost, resource utilization and environmental impact; The adaptive reinforcement learning module is used to adjust path selection and chain splitting in real time based on environmental feedback and historical data; The constraint optimization module is used to perform constraint optimization on path selection and chain splitting based on the Lagrange multiplier method. The scheduling module is used to automatically schedule transportation vehicles, personnel, and resources according to the optimal plan; The real-time monitoring module is used to monitor the transportation process in real time through Internet of Things (IoT) technology. The feedback module is used to receive feedback information from the execution layer and adjust and optimize the scheduling scheme in real time.

[0016] This invention provides a chain transportation method and system for international logistics. It has the following beneficial effects: 1. In this invention, chain splitting and path selection optimization are performed based on a multi-objective optimization model, thereby achieving a global balance between timeliness, transportation cost, resource utilization and environmental impact. This improves the problem that traditional path design methods mostly focus on a single objective and fail to consider the trade-offs between multiple objectives, resulting in path schemes that do not have overall optimality.

[0017] 2. In this invention, by cleaning, standardizing and fusion processing the collected environmental data, a unified and high-quality data input foundation is constructed, thereby improving the problem that traditional data processing processes mostly adopt a decentralized manual integration method, which causes the optimization algorithm to have distorted input and unstable output results due to inconsistent data formats and many missing values.

[0018] 3. In this invention, by adopting an adaptive reinforcement learning algorithm, the path selection and chain splitting are adjusted in real time based on environmental feedback and historical data, thereby enabling the system to have self-learning and dynamic adaptation capabilities. This improves the problem that traditional path decision-making models are mostly rule-driven or static models, which lack closed-loop feedback learning of actual execution results, resulting in rigid scheduling strategies and a lack of adaptability.

[0019] 4. In this invention, the path selection and chain splitting are constrained and optimized by using the Lagrange multiplier method, thereby ensuring that the solution meets the resource boundary and environmental constraints in actual execution. This improves the problem that traditional optimization methods mostly ignore the executability constraints. Although the generated solution is mathematically optimal, it lacks practical operability, which leads to execution conflicts or resource insufficiency. Attached Figure Description

[0020] Figure 1 This is a flowchart of the method proposed in this invention; Figure 2 This is a diagram of the system module architecture proposed in this invention; Figure 3 This is the IoT monitoring and feedback closed-loop diagram proposed in this invention; Figure 4 This is a decision loop diagram of the learning mechanism proposed in this invention; Figure 5 This is a diagram of the decision tree structure for the scheduling rules proposed in this invention. Detailed Implementation

[0021] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Example 1 In a first embodiment of the present invention, the present invention provides a chain-link transportation method for international logistics, such as... Figures 1-5 As shown, it includes the following steps: S1. Collect logistics environment information based on real-time environmental monitoring data, including weather conditions, traffic flow, flight schedules, port conditions, current location of goods, and status of transport vehicles; Furthermore, in S1, logistics environment information is obtained in real time through interfaces with external platforms, including traffic management platforms, weather forecasting systems, flight management systems, and port management systems.

[0023] Specifically, logistics environment information is collected based on real-time environmental monitoring data, including weather conditions, traffic flow, flight schedules, port conditions, current cargo location, and vehicle status. The purpose is to provide environmental input parameters under the current state for subsequent chain segmentation and route scheduling, so as to achieve dynamic optimization and control of the entire logistics transportation process.

[0024] By interfaceing with external platforms, the system obtains real-time logistics environment information. These platforms include traffic management platforms, weather forecasting systems, flight management systems, and port management systems. Access to these platforms ensures the timeliness and completeness of the collected environmental data, enabling the system to dynamically update route selection and resource scheduling based on real-time status.

[0025] The set of states formed by collecting environmental data is: E = {e1, e2, ..., e} n}, where e1 represents current weather conditions; e2 represents current traffic flow data; e3 represents current flight schedule data; e4 represents current port operation status; e5 represents the current location of goods; e6 represents the real-time operating status of transportation vehicles; and n represents the number of environmental variables. The environmental state set E is used for the path selection function f. route With the chain splitting function f chain One of the inputs.

[0026] The various types of data returned by external platform interfaces can be modeled as: D i (t) = API i (t), where D i (t) represents the value of the i-th data type at time t; API i This provides the data interface function for the i-th type of data platform; t represents the current sampling time. All collected data, after standardization and structuring, forms a set E, which serves as the environment input variable for subsequent model optimization: Input. env =E, this input variable will be used in the model calculations for dynamic path planning, resource allocation, and chain determination.

[0027] S2. The collected environmental data is cleaned, standardized, and fused to obtain usable real-time data. Further, S2 includes removing redundant data, filling in missing values, and fusing data from different sources to ensure data consistency and integrity. Data standardization technology is further adopted to integrate data from different sources with the same format and units.

[0028] Specifically, the environmental data collected in step S1 is cleaned, standardized, and fused. The purpose is to transform the raw environmental data into structured and unified input data that can be directly used for chain splitting and path scheduling optimization, so as to ensure that the data received by the subsequent optimization module is accurate, consistent, and comparable.

[0029] Redundant data removal removes duplicates, invalid records, and outliers to avoid interfering with subsequent models; missing value imputation fills in or replaces missing fields to ensure dataset integrity; data fusion and standardization integrate data fields from different data platforms and standardizes their dimensions and formats. This ensures semantic and structural consistency of the collected multi-source heterogeneous data, preventing inconsistencies in data format, units, or precision from affecting the accuracy of optimization results.

[0030] Let the collected raw data set be: E raw ={e ij |1≤i≤n,1≤j≤m i}, where i represents the data category; j represents the j-th record in the i-th data category, and m i This represents the number of records in the i-th data category.

[0031] First, consider set E. raw Redundancy removal and denoising are performed to obtain the denoised dataset: E clean ={e ij ′∈E raw |valid(e ij )=1}, where the function valid(·) is the record validity determination function, and the output value is 1 to indicate validity.

[0032] For records with missing information, a fill function φ is defined to fill in the missing information using historical averages or adjacent interpolation. Where φ(e) ij ) represents the filler value for the missing item.

[0033] Data from different platforms may have different field definitions. To unify the data format, each data entry will be represented by an attribute vector as follows: in Let represent the l-th field of the k-th data entry, and d be the total number of fields. A field mapping table M maps data from different sources to a unified set of fields: x k ′=M(x k ), such that all x′ k With consistent field definitions and semantic structure, they constitute a structured fusion dataset: E fused ={x′1,x′2,…,x′ p}

[0034] To avoid the impact of different data units and scales on the performance of the optimization model, the fused data needs to be standardized. A linear standardization method is used to map each field value to a uniform interval: Where x is the original field value; x min ,xmax The minimum and maximum values ​​of this field in the sample data; x″ the standardized field value. After standardization, a standardized dataset is formed: E std ={x″1,x″2,…,x″ p}

[0035] Input: Raw environmental data E raw ; Processing flow: Redundant data removal → Missing value imputation → Field fusion → Format unification → Unit standardization; Output: Standardized dataset E, which can be directly used by subsequent models. std .

[0036] S3. Based on the multi-objective optimization model, taking into account timeliness, cost, resource utilization and environmental impact, the chain splitting and path selection optimization are carried out to obtain the optimal path selection scheme. Furthermore, in S3, the Pareto optimal solution algorithm is used to make objective trade-offs, ensuring that the best balance is found among multiple objectives such as timeliness, cost, resource utilization and environmental impact, and outputting the optimal path selection scheme.

[0037] Specifically, based on a multi-objective optimization model, considering factors such as timeliness, cost, resource utilization, and environmental impact, chain segmentation and route selection optimization are performed to obtain the optimal route selection scheme. The core purpose of this step is to ensure the output of the best logistics route scheme under various constraints by comprehensively considering the trade-offs between multiple objectives.

[0038] By using the Pareto optimal solution algorithm, step S3 achieves a balance among multiple objectives, resolving potential conflicts or contradictions between them. For example, optimizing timeliness might increase costs, while optimizing resource utilization might affect environmental impact. The Pareto optimal solution algorithm finds a balance point among these objectives, ensuring that the performance of each objective is optimized to the maximum extent without sacrificing the performance of other objectives.

[0039] The multi-objective optimization model considers four main objectives: Timeliness (T): Optimizing transportation time to ensure goods arrive on time; Cost (C): Minimizing transportation costs, including fuel costs, vehicle costs, etc.; Resource Utilization (R): Improving the utilization rate of transportation vehicles, personnel, and warehousing; Environmental Impact (E): Reducing environmental burdens such as carbon emissions and energy consumption. Each objective is defined in the model as follows: Where x i Let t represent the selection variable for the i-th path. i c i r i ei These represent the timeliness, cost, resource utilization, and environmental impact of the corresponding routes, respectively.

[0040] In multi-objective optimization, Pareto optimal solutions are used to handle trade-offs between objectives. A Pareto optimal solution is one that cannot improve other objectives without sacrificing one. Consider a set of solutions {x1, x2, ..., x...} n}, where x i For possible path selection solutions, then the solution x k It is Pareto optimal, and there is no other solution that is better than it on all objectives.

[0041] The following criteria are used to determine whether a solution is Pareto optimal: The algorithm iterates through all possible path selection schemes and gradually selects the Pareto optimal solution set, ensuring the balance and optimization among the objectives.

[0042] The Pareto optimal solution algorithm outputs a final optimal path selection scheme that balances timeliness, cost, resource utilization, and environmental impact. The final optimal path selection scheme x * The solution satisfies the following conditions: Where X is the set of all possible path choices, and T(x), C(x), R(x), and E(x) represent the timeliness, cost, resource utilization, and environmental impact of path x, respectively. The optimal path choice is x. * It is the optimal solution under multi-objective constraints.

[0043] Input: Environmental data E std (Obtained through the aforementioned data cleaning and standardization steps). Objective function coefficients t i c i r i e i , representing the timeliness, cost, resource utilization, and environmental impact of each path, respectively. Output: Optimal path selection scheme x * A set of optimal path selection decisions can simultaneously optimize timeliness, cost, resource utilization, and environmental impact.

[0044] By applying multi-objective optimization and Pareto optimality, the contradictions and constraints between multiple objectives can be effectively handled, resulting in the optimal route selection scheme. This optimization process can improve resource utilization efficiency and reduce environmental impact while ensuring transportation timeliness and reducing costs, thereby enhancing overall logistics efficiency and ensuring sustainable development.

[0045] S4. Adaptive reinforcement learning algorithm is adopted to adjust the path selection and chain splitting in real time based on environmental feedback and historical data, thereby optimizing transportation timeliness and resource consumption. Furthermore, in S4, the Q-learning algorithm is used to learn strategies for path selection and chain splitting in a dynamic environment, optimize the reward function based on real-time feedback, and gradually improve the decision accuracy of the system through multiple rounds of trials.

[0046] Specifically, by employing an adaptive reinforcement learning algorithm, the path selection and chain splitting strategies are dynamically adjusted based on environmental feedback and historical data. This aims to improve the system's responsiveness and adaptability in uncertain transportation environments, thereby optimizing transportation timeliness and resource consumption.

[0047] A Q-learning algorithm is introduced to establish a state-action decision model for the path selection and chain splitting problem. In each state, the system selects an action based on the current strategy and receives feedback from the environment after the action is executed, updating the strategy accordingly. Through continuous learning, the accuracy of path decisions and the system's adaptability to dynamic environments are gradually improved.

[0048] The state space of the system is: S = {s1, s2, ..., s} n}, where any state s t ∈S represents the overall state of the logistics system under the current environment, which includes a combination of multiple factors such as the smoothness of the route, the availability of transportation tools, the current time period, and the weather level.

[0049] The system in any state s t The set of possible actions is: A = {a1, a2, ..., a...} m}, where any action a t ∈A represents the state s t The decision to make a certain route selection or chain splitting strategy, such as changing the mode of transportation, switching routes, or delaying shipment.

[0050] The system is in state s t Perform action a t And transition to the new state s t+1 Afterwards, receive the immediate reward R returned by the environment. t The reward is used to evaluate the impact of this action on overall transportation timeliness and resource consumption. The reward function is defined as: R(s) t ,a t )=-α·T t -β·C t -γ·R t T t C represents the transport time corresponding to the current chain segment. t R represents the transportation cost incurred in the current chain.t The resource consumption of the current stage is represented by α, β, and γ, which are objective weighting coefficients used to balance multiple optimization objectives.

[0051] The Q-learning algorithm updates the expected reward (Q-value) for each state-action pair using the following formula: Where Q(s) t ,a t ) represents the Q-value of the current state-action pair; η is the learning rate, controlling the learning speed; δ is the discount factor, used to evaluate the weight of future returns; max a ′Q(s t+1 R(s,a′) represents the maximum Q-value of the next state; t ,a t () is the immediate reward after the current action is performed.

[0052] The system uses an ε-greedy strategy to select actions in each iteration: Through multiple rounds of training, the Q-value will gradually converge to the optimal strategy, and the system's path selection and chain splitting decision-making capabilities will be gradually improved.

[0053] Input: Current environment state s t The input data is processed by the preceding steps (S1 and S2); the historical state-action-Q value table Q(s,a); and the reward function parameter T. t C t R t And weighting coefficients α, β, γ; training parameters η, δ, ε. Output: The optimal path selection and chain splitting strategy after training, i.e., the mapping relationship π. * (s)=a * , representing the optimal action selection in each state; the updated Q-value table Q′(s,a).

[0054] By introducing the Q-learning algorithm, the decision-making process for logistics routes and links acquires self-learning and adaptive capabilities. The system can make optimal decisions based on the current state in a dynamic environment, and gradually optimize its strategy as data accumulates. This mechanism effectively solves the limitation of traditional methods that rely on static rules for route selection, providing a decision-making basis for dynamic adjustment and real-time response in subsequent steps, and enhancing the robustness and real-time performance of the overall logistics scheduling system.

[0055] S5. Based on the Lagrange multiplier method, the route selection and chain splitting are constrained and optimized to ensure that the constraints are met. Furthermore, in S5, the Lagrange multiplier method is used to take transportation timeliness, resource utilization rate and environmental impact as constraints, and the optimal route selection scheme is solved according to the constraints to ensure that the scheme meets all preset resource and environmental limitations in actual implementation.

[0056] Specifically, the Lagrange multiplier method is used to perform constrained optimization of path selection and chain splitting. The purpose is to introduce system constraints such as transportation timeliness, resource utilization and environmental impact on the basis of multi-objective optimization, so as to ensure that the obtained path selection scheme not only has optimal performance, but also meets the feasibility requirements in actual operation.

[0057] The Lagrange multiplier method is used to introduce multiple inequality constraints into the objective function of the path selection optimization problem, so as to solve the extreme value solution under the constraints and make the final output scheme satisfy the boundary conditions of the transportation system in terms of timeliness, resources, environment, etc.

[0058] Let the path selection scheme be a decision vector: x = (x1, x2, ..., xn) n ), where x i Indicates whether the i-th path is selected, x i ∈{0,1}. The objective function is the total transportation cost, denoted as: Where c i Let represent the transportation cost of the i-th route.

[0059] The constraint definition introduces the following three constraints: Transportation timeliness constraints: Where t i Let T be the transport time for the i-th route. max This represents the upper limit of acceptable delivery time.

[0060] Resource utilization constraints: Where r i R represents the resource utilization level of the i-th path. min This represents the minimum resource utilization efficiency threshold.

[0061] Environmental impact constraints: Where e i E represents the environmental emissions corresponding to the i-th path. max This represents the upper limit of permissible environmental impact.

[0062] The Lagrange function is constructed as follows: Where λ1, λ2, and λ3 are Lagrange multipliers, corresponding to the three constraints respectively. Through... Regarding x and λ i The first-order partial derivatives of x are zero, forming the KKT conditions. Solve for x that satisfies the following conditions. - : Therefore, the optimal path selection scheme that satisfies the constraints can be obtained.

[0063] Input: Cost parameter c i Time parameter ti Time limit T max Resource parameter r i Resource minimum R min Environmental parameter e i Environmental limit E max Output: The path selection scheme x that satisfies the constraints. - ; the corresponding Lagrange multipliers

[0064] By introducing the Lagrange multiplier method, systematic constraints can be explicitly added when optimizing path schemes, ensuring the feasibility of the optimal path at the execution level and meeting time constraints, resource efficiency, and environmental protection requirements. This method solves the problem of theoretically optimal but practically unfeasible solutions caused by the lack of consideration for executability constraints in traditional multi-objective optimization models. It corrects and compensates for the constraints of the model solutions in previous steps S3 and S4, further enhancing the practical application value and overall robustness of the system's path planning decision-making.

[0065] S6. Based on the optimal route selection and chain splitting scheme, the transportation vehicles, personnel and resources are automatically scheduled. Furthermore, in S6, a rule-based decision tree algorithm is used to automatically allocate transportation vehicles, arrange personnel and resources according to the optimal route selection and chain splitting scheme, and dynamically adjust each link in the logistics scheduling process to optimize transportation efficiency.

[0066] Specifically, based on the optimal route selection and chain splitting scheme determined in the preceding steps, automated scheduling of transportation vehicles, personnel, and resources is implemented. The technical objective is to achieve optimal resource allocation in the route execution phase, ensuring the operability of route decisions and the efficiency of logistics execution.

[0067] A rule-based decision tree algorithm is introduced to model and execute the allocation process of transportation resources. By analyzing path and link schemes, and combining the current status of transportation resources and scheduling rules, the system automatically completes the allocation of transportation vehicles, personnel arrangements, and material resource distribution. During execution, it can make scheduling adjustments based on real-time status to ensure that transportation efficiency remains at an optimal level in a dynamic environment.

[0068] Let the set of path selection and link schemes be: P = {p1, p2, ..., p...} n}, where p i This represents the i-th determined chain path, which includes information such as the starting point, ending point, estimated transit time, type of goods being transported, and priority.

[0069] The set of transportation resources includes: the set of transportation vehicles: V = {v1, v2, ..., v m}; Personnel set: H={h1,h2,...,h k}; Set of material resources: R = {r1, r2, ..., r l Each resource unit has an attribute vector representation, such as a means of transportation v. j The state vector is: a vj ={capacity,location,available_time,vehicle_type}; The rule-based decision tree is constructed as a decision structure, with the path scheme p as the input. i The resource status is considered, and the output is the scheduling decision. Let the rule function be: D(p i ) = Match(p i The function Match (V,H,R) matches the path p. i The transportation requirements are matched with the current resource status, and a resource combination that meets the scheduling constraints is returned.

[0070] Decision logic nodes are defined as Boolean decision conditions, for example: if capacity(v j )≥load(p i And location(v) j ) = start(p i If the vehicle v is allocated, then the allocation will be made. j If available_time(h k )≤start_time(p i If so, then assign personnel h k If resource type r m If the characteristics of the goods can be met, then the dispatched materials r m The system traverses the decision tree to bind resources and ultimately generates a scheduling decision output.

[0071] Define the scheduling result as: S = {p i ,v j ,h k ,r m Let ) | i = 1, ..., n}, where each tuple represents the path p of the linked list. i Assigned to transport vehicle v j Operator h k and required resources r m .

[0072] Input: Set of chain path solutions P; Set of current transportation vehicle states V; Set of personnel availability H; Set of resource information R. Output: Set of resource scheduling solutions S.

[0073] During transportation execution, the system periodically monitors the execution status of each link node. When any of the following states occur, a scheduling reconfiguration is triggered: vehicle delay or malfunction; temporary unavailability of personnel; or external environmental factors affecting the accessibility of the transportation path. At this time, the decision tree model is reactivated to rematch the affected paths and output a new scheduling combination S′, thereby ensuring the continuity and real-time performance of the scheduling.

[0074] By introducing a rule-based decision tree algorithm, the resource scheduling in the path execution phase is fully automated, significantly reducing the proportion of manual intervention. A dynamic update mechanism ensures the adaptability and continuous effectiveness of resource allocation. This module serves as a bridge between the upper-level path optimization model and the lower-level transportation execution, making it a key control node in forming a closed-loop, responsive international logistics chain system.

[0075] S7. Real-time monitoring of the transportation process is carried out through Internet of Things (IoT) technology to ensure that the transportation route is executed in accordance with the predetermined plan. Furthermore, in S7, IoT technology, including RFID tags and GPS positioning technology, is used to monitor the current location and transportation status of the transportation vehicle in real time to ensure the traceability of the transportation process and to send alarms to the system in a timely manner in case of abnormalities.

[0076] Specifically, the transportation process is monitored in real time using IoT technology to ensure that the execution of the transportation route is consistent with the predetermined route selection and chain splitting plan. The purpose of this step is to achieve full-process visualization, traceability, and controllability of the transportation process, providing data support for timely intervention and scheduling adjustments in abnormal situations.

[0077] The IoT technologies employed in the S7 primarily include RFID tags and GPS positioning modules. RFID tags are embedded in the transport carrier or cargo to identify and confirm the identity and status of the transport unit; the GPS module is embedded in the transport vehicle to obtain its real-time geographical location information. The combination of these two technologies enables multi-dimensional monitoring of the transport unit in both spatial and status dimensions.

[0078] Real-time location tracking: Let the number of a certain means of transport be v j The position at time t is: Where λ t φ is the longitude of the vehicle at time t. t Let be the latitude of the vehicle at time t; this coordinate data is reported by the embedded GPS module every Δt seconds. The designed path contains link nodes N = {n1, n2, ..., n}. k The set of positions corresponding to} is: Define the current position deviation as: Where Dist(·) is the Euclidean distance function for geographic coordinates. When δ(t) > θ (preset path tolerance threshold), the system determines that the current path has deviated and generates an alert: Alert position =1.

[0079] The RFID module is used to record the status of goods or transport units. Let the current goods number be h. i The status data is as follows: Each s i This represents specific status parameters, such as seal integrity, abnormal temperature and humidity markings, and loading status. The RFID reader periodically polls these parameters and uploads the results to the system. The anomaly detection function is defined as follows: When there exists any s i Satisfying Anomaly(s) i =1, the system triggers an abnormal status alarm: Alert status =1.

[0080] Input: Real-time GPS location data Path node coordinate set L plan RFID status parameters Output: Alert for path deviation assessment result position Transportation status anomaly alert status Status record data stream, used for decision system backtracking and analysis.

[0081] When the system detects a route deviation or abnormal status, it immediately sends a feedback signal to the control center. Based on the alarm type, the control system automatically performs one of the following actions: reactivates the route optimization and scheduling module; pushes operational instructions to the transportation manager; marks the node task as "failed" and transfers it to the abnormal chain handling process. This closed-loop mechanism ensures that the execution of the transportation route is not only planned but also has real-time control and traceability capabilities, meeting the operational requirements of a high-safety, high-reliability logistics system.

[0082] By introducing Internet of Things (IoT) technology, especially the integration of RFID and GPS, step S7 achieves continuous sensing, anomaly identification, and execution monitoring of transportation links in the international logistics chain. Compared to traditional static task monitoring methods, this step provides real-time and refined path status control capabilities, which is a key technical support for ensuring the implementation of chain path execution and scheduling strategies. This step forms a closed-loop structure of "data-driven - path planning - resource scheduling - path execution - execution monitoring" in the overall system operation.

[0083] S8. Based on feedback information from the execution layer, adjust and optimize the scheduling scheme in real time; Furthermore, in S8, based on the execution layer feedback mechanism, by receiving real-time feedback information during the transportation process, a rolling optimization strategy is adopted to dynamically adjust the path selection, chain splitting, and resource scheduling to ensure continuous optimization of the transportation process.

[0084] Specifically, based on real-time feedback information from the execution layer, the path selection, chain segmentation, and resource scheduling schemes are dynamically adjusted to achieve continuous optimization of the entire logistics chain transportation process. The core of this step lies in building a feedback closed-loop mechanism, so that optimization decisions are no longer single static executions, but have dynamic response capabilities and rolling correction characteristics.

[0085] By using a rolling optimization strategy, real-time status data acquired during execution is used as new input to the optimization model to correct the original path and scheduling decisions in real time, ensuring that the system continues to approach the optimal operating state and enhancing its adaptability to uncertainties in the transportation process.

[0086] Execution feedback data modeling: Let the current time be t. The data fed back by the system execution layer includes transportation status, resource status, and path execution deviation. Real-time path position error: δ p (t)=Dist(L actual (t),L plan (t)); The current set of available resource states: R t ={r1(t),r2(t),…,r m (t)}; Link execution completion state vector: χ(t)={χ1(t),χ2(t),…,χ k (t)}; Where, χ i (t)∈{0,1} indicates whether the i-th link is completed as planned within time t.

[0087] The system performs a rolling optimization process for execution paths and resource scheduling at fixed time intervals Δt. Within each rolling window, the optimization objective function is reconstructed: Where x is the path and scheduling decision vector at the current moment; T t (x) represents the predicted total lead time; C t (x) represents the predicted total cost; E t (x) represents the predicted environmental impact; α, β, γ are the target weight coefficients; X(t) is the feasible solution space dynamically updated based on feedback data. Constraints are introduced, including current resource state constraints: Current execution state constraints of the chain: This means that completed chain processes will not be optimized, and allocated resources will not be rescheduled.

[0088] After rolling optimization, a new path and resource configuration scheme is output: This is used to replace or modify the current plan and push it to the scheduling system for execution. Then, the next round of sampling and optimization continues.

[0089] Input: Real-time execution layer feedback data: δ p (t),R t ,χ(t); Optimize the objective function and weight coefficients f t (x), α, β, γ; Current rolling window parameters Δt. Output: Updated path and resource scheduling scheme x * (t).

[0090] This step constitutes the feedback adjustment loop of the control system. Its operation process is as follows: 1. Receive feedback data from the execution layer; 2. Correct the current system state model; 3. Enter the optimization module to recalculate the path and resource configuration; 4. Generate a new scheme and replace the old schedule; 5. Start the next rolling window and repeat the process.

[0091] By constructing a rolling optimization and feedback closed-loop mechanism, S8 achieves continuous dynamic optimization of transportation routes and resource allocation. This mechanism not only enhances the system's adaptability to dynamic transportation environments but also significantly improves resource utilization efficiency and task completion stability. Compared to single-time scheduling and static execution methods, this solution possesses higher robustness and stronger engineering feasibility, making it one of the core technologies enabling continuous optimization of chain-link transportation methods and systems.

[0092] Example 2 In the context of large-scale order consolidation during peak cross-border e-commerce promotional periods, international logistics transportation needs to achieve high-intensity, high-density cross-border transportation operations within a very short time. Routes encompass multiple modes including air, rail, truck, and sea freight, involving multiple countries and regulatory nodes, resulting in a complex and volatile logistics environment. In this scenario, traditional route planning and scheduling methods struggle to cope with unforeseen events such as real-time weather changes, port congestion, flight cancellations, and vehicle malfunctions. These methods suffer from problems such as the inability to dynamically split chain paths, poor real-time performance of route optimization under multi-objective conditions, inefficient allocation of transportation resources, rigid scheduling strategies lacking adaptability, and a lack of real-time monitoring and anomaly response mechanisms for transportation execution. To address these issues, this invention provides a chain transportation system for international logistics, the structure of which is as follows: Figure 2 As shown. The specific implementation process of this system is as follows: The data acquisition module is used to collect logistics environment information in real time; The data processing module is used to clean, standardize, and integrate the collected environmental data; A multi-objective optimization module is used to optimize chain splitting and path selection based on timeliness, cost, resource utilization and environmental impact; The adaptive reinforcement learning module is used to adjust path selection and chain splitting in real time based on environmental feedback and historical data; The constraint optimization module is used to perform constraint optimization on path selection and chain splitting based on the Lagrange multiplier method. The scheduling module is used to automatically schedule transportation vehicles, personnel, and resources according to the optimal plan; The real-time monitoring module is used to monitor the transportation process in real time through Internet of Things (IoT) technology. The feedback module is used to receive feedback information from the execution layer and adjust and optimize the scheduling scheme in real time.

[0093] Specifically, the data acquisition module interfaces with external platforms such as traffic management systems, weather forecasting systems, flight scheduling systems, and port operation systems to collect data related to transportation routes, including traffic flow, weather conditions, port operation status, flight times, cargo location information, and vehicle status. The collected data is in the form of a structured time-series data stream and is automatically updated at a set frequency to ensure real-time performance and completeness. The data acquisition results constitute an environmental status set, serving as a key input for multi-objective optimization and dynamic route decision-making.

[0094] The data processing module first addresses redundant and missing data, including outlier removal, data interpolation, and null value imputation, ensuring data quality. It then performs format conversion, unit standardization, and field alignment on data from different sources, employing standardization methods to normalize data with different dimensions. Finally, it generates a data matrix that can be directly input into the optimization model, achieving unified processing of input data in terms of structure, dimension, and format.

[0095] The multi-objective optimization module constructs a multi-objective model with path timeliness, transportation cost, resource utilization, and environmental impact as optimization objectives. The Pareto optimal solution method is used to achieve non-dominated solution trade-offs among the objectives. The model outputs a set of optimal path solutions that satisfy the constraints under the current environmental conditions, and performs feasibility assessment and ranking for each link node, providing an optimization basis for subsequent scheduling strategies.

[0096] The adaptive reinforcement learning module uses the Q-learning algorithm to construct a state-action space. In each state, the system selects an action according to the learning strategy and adjusts the Q-value function through environmental feedback, gradually approaching the optimal strategy. Through continuous iteration and experience accumulation, the system can proactively correct the path and split the strategy according to the current state, thus possessing dynamic adaptability in actual execution and solving the problem of rigid and invariant traditional path strategies.

[0097] The constraint optimization module constructs a joint optimization model based on the Lagrange multiplier method, which integrates the objective function with multiple inequality constraints. Constraint variables include maximum acceptable time, minimum resource utilization, and maximum environmental impact limits. By solving the Lagrange function, the module obtains the optimal solution that satisfies all constraints, ensuring the scheduling scheme is feasible and resource-efficient.

[0098] The scheduling module incorporates a rule-based decision tree algorithm to match route task requirements with current resource status, enabling automatic allocation of transportation resources. Resource status includes the capacity, availability, location, and type of transportation vehicles, personnel scheduling, and loading / unloading capabilities for materials. The system automatically generates scheduling plans based on task priority and resource constraints, and can perform resource replacement and task reallocation during actual execution.

[0099] The real-time monitoring module combines RFID tags and GPS positioning technology to collect information such as the geographical location of the transport vehicle, cargo status, and mission progress in real time. By comparing the current location with the predetermined route nodes, the degree of deviation can be calculated and abnormal states can be identified. When location deviation, timeout, or abnormal transport status occurs, the system triggers an anomaly flag and transmits the status information to the feedback module for further processing.

[0100] The feedback module periodically receives path execution results, task completion status, and resource change information, constructs a feedback state vector, and uses it as rolling input to the optimization module. A rolling optimization strategy is employed to restart the path solving process, enabling dynamic optimization updates based on the current system state. This feedback mechanism effectively improves the system's responsiveness to unplanned factors, ensuring the continuous optimal execution of link transportation.

[0101] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A chain-link transportation method for international logistics, characterized in that, Includes the following steps: S1. Collect logistics environment information based on real-time environmental monitoring data, including weather conditions, traffic flow, flight schedules, port conditions, current location of goods, and status of transport vehicles; S2. Clean, standardize, and fuse the collected environmental data to obtain usable real-time data; S3. Based on the multi-objective optimization model, taking into account timeliness, cost, resource utilization and environmental impact, the chain splitting and path selection optimization are carried out to obtain the optimal path selection scheme. S4. Adaptive reinforcement learning algorithm is adopted to adjust the path selection and chain splitting in real time based on environmental feedback and historical data, thereby optimizing transportation timeliness and resource consumption. S5. Perform constraint optimization on the path selection and chain splitting based on the Lagrange multiplier method to ensure that the constraint conditions are met; S6. Based on the optimal route selection and chain splitting scheme, perform automated scheduling of transportation vehicles, personnel and resources; S7. Use IoT technology to monitor the transportation process in real time to ensure that the transportation route is executed in accordance with the predetermined plan; S8. Based on feedback information from the execution layer, adjust and optimize the scheduling scheme in real time.

2. The chain transportation method for international logistics according to claim 1, characterized in that, In step S1, the logistics environment information is obtained in real time through interface with external platforms, including traffic management platforms, weather forecasting systems, flight management systems, and port management systems.

3. The chain transportation method for international logistics according to claim 1, characterized in that, S2 includes removing redundant data, filling in missing values, and merging data from different sources to ensure data consistency and integrity. Furthermore, data standardization technology is used to integrate data from different sources with the same format and units.

4. The chain transportation method for international logistics according to claim 1, characterized in that, In S3, the Pareto optimal solution algorithm is used to perform objective trade-offs, ensuring that the best balance is found among multiple objectives such as timeliness, cost, resource utilization and environmental impact, and outputting the optimal path selection scheme.

5. The chain-link transportation method for international logistics according to claim 1, characterized in that, In S4, the Q-learning algorithm is used to learn strategies for path selection and chain splitting in a dynamic environment, optimize the reward function based on real-time feedback, and gradually improve the decision accuracy of the system through multiple rounds of trials.

6. The chain transportation method for international logistics according to claim 1, characterized in that, In S5, the Lagrange multiplier method is used to take transportation timeliness, resource utilization rate and environmental impact as constraints, and the optimal route selection scheme is solved according to the constraints to ensure that the scheme meets all preset resource and environmental limitations in actual implementation.

7. The chain transportation method for international logistics according to claim 1, characterized in that, In S6, a rule-based decision tree algorithm is used to automatically allocate transportation vehicles, arrange personnel and resources according to the optimal path selection and chain splitting scheme, and dynamically adjust each link in the logistics scheduling process to optimize transportation efficiency.

8. The chain transportation method for international logistics according to claim 1, characterized in that, In S7, IoT technology, including RFID tags and GPS positioning technology, is used to monitor the current location and transportation status of the transportation vehicle in real time, ensuring the traceability of the transportation process and sending alarms to the system in a timely manner in case of abnormalities.

9. The chain transportation method for international logistics according to claim 1, characterized in that, In S8, based on the execution layer feedback mechanism, by receiving real-time feedback information during the transportation process, a rolling optimization strategy is adopted to dynamically adjust the path selection, chain splitting, and resource scheduling to ensure continuous optimization of the transportation process.

10. A chain transportation system for international logistics, characterized in that, The chain-link transportation method for international logistics as described in any one of claims 1-9 Includes the following modules: The data acquisition module is used to collect logistics environment information in real time; The data processing module is used to clean, standardize, and integrate the collected environmental data; A multi-objective optimization module is used to optimize chain splitting and path selection based on timeliness, cost, resource utilization and environmental impact; The adaptive reinforcement learning module is used to adjust path selection and chain splitting in real time based on environmental feedback and historical data; The constraint optimization module is used to perform constraint optimization on path selection and chain splitting based on the Lagrange multiplier method. The scheduling module is used to automatically schedule transportation vehicles, personnel, and resources according to the optimal plan; The real-time monitoring module is used to monitor the transportation process in real time through Internet of Things (IoT) technology. The feedback module is used to receive feedback information from the execution layer and adjust and optimize the scheduling scheme in real time.

Images