Cross-border multimodal transport path dynamic planning method and device, equipment and storage medium

Abstract

The invention is suitable for the technical field of intelligent logistics scheduling, and provides a cross-border multimodal transport path dynamic planning method, device and equipment and a storage medium. Multi-source real-time data is acquired and fused, policy information in the multi-source real-time data is analyzed and quantified, and the real-time data is obtained; according to the method, an unstructured policy text is converted into a dynamic constraint condition which can be understood by a multi-objective optimization model, so that dynamic optimization of path planning is realized, changes of an external policy environment can be responded in real time, a generated path scheme achieves a better balance in multiple objectives such as compliance, cost and timeliness, and the path planning efficiency is improved. And the accuracy, timeliness and anti-risk capability of the planning result are improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of intelligent logistics scheduling, and particularly relates to a cross-border multimodal transport path dynamic planning method and device, equipment and a storage medium. BACKGROUND

[0002] In the global supply chain management, cross-border e-commerce logistics and other industries, path planning is a core link. The traditional path planning system usually assigns fixed weights to transportation cost, time and other targets for static optimization based on historical operation data. However, trade rules such as customs policies and tariff regulations of various countries often change suddenly (for example, a country suddenly announces the imposition of import duties on certain goods), and the traditional method is difficult to identify, analyze and consider the impact in real time, which may lead to the fact that the planned path solution cannot adapt to the latest policy environment, and a route that faces high tariffs or clearance obstacles may be selected, thereby causing serious problems such as a sharp increase in logistics costs and transportation delays. That is, the prior art is difficult to plan a path in a dynamic policy environment in real time in response to and integrate the impact of policy changes. SUMMARY

[0003] In view of this, the embodiments of the present application provide a cross-border multimodal transport path dynamic planning method, device, equipment and storage medium, which can solve the problem that the related art is difficult to plan a path in a dynamic policy environment in real time in response to and integrate the impact of policy changes.

[0004] In a first aspect, the embodiments of the present application provide a cross-border multimodal transport path dynamic planning method, comprising: Obtaining multi-source real-time data and fusing the multi-source real-time data to obtain fused data; Performing policy analysis on the fused data to obtain policy evaluation information; Quantifying the policy evaluation information into dynamic constraint conditions; Inputting the dynamic constraint conditions and the fused data into a multi-objective optimization model to obtain a Pareto optimal solution set, the Pareto optimal solution set containing at least one cross-border multimodal transport path solution; Determining a target path solution from the Pareto optimal solution set.

[0005] In a second aspect, the embodiments of the present application provide a cross-border multimodal transport path dynamic planning device, the device comprising: An acquisition module configured to obtain multi-source real-time data and fuse the multi-source real-time data to obtain fused data; An analysis module configured to perform policy analysis on the fused data to obtain policy evaluation information; A quantification module configured to quantify the policy evaluation information into dynamic constraint conditions; An input module is configured to input the dynamic constraint condition and the fused data into the multi-objective optimization model to obtain a set of Pareto optimal solutions, and the set of Pareto optimal solutions comprises at least one cross-border multimodal transport path scheme. A determination module is configured to determine a target path scheme from the set of Pareto optimal solutions.

[0006] In a third aspect, an embodiment of the present application provides a terminal device, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the cross-border multimodal transport path dynamic programming method when executing the computer program.

[0007] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, which stores a computer program, and the computer program implements the steps of the cross-border multimodal transport path dynamic programming method when executed by a processor.

[0008] In a fifth aspect, an embodiment of the present application provides a computer program product, which, when executed on a terminal device, causes the terminal device to perform the cross-border multimodal transport path dynamic programming method.

[0009] Compared with the prior art, the embodiment of the present application has the beneficial effects that: the embodiment of the present application obtains multi-source real-time data, fuses the multi-source real-time data to obtain fused data, performs policy analysis on the fused data to obtain policy evaluation information, quantifies the policy evaluation information into a dynamic constraint condition, inputs the dynamic constraint condition and the fused data into a multi-objective optimization model to obtain a set of Pareto optimal solutions, and determines a target path scheme from the set of Pareto optimal solutions. The embodiment of the present application obtains and fuses multi-source real-time data, analyzes and quantifies policy information, converts unstructured policy text into a dynamic constraint condition understandable by a multi-objective optimization model, thereby realizing dynamic optimization of path planning, responding to changes in the external policy environment in real time, making the generated path scheme achieve a better balance in compliance, cost, timeliness and other multiple targets, and improving the accuracy, timeliness and anti-risk ability of the planning result. BRIEF DESCRIPTION OF DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0011] Figure 1 is an implementation flow diagram of the cross-border multimodal transport path dynamic programming method provided by the embodiment of the present application.

[0012] Figure 2 FIG. 1 is a structural schematic diagram of a cross-border intermodal path dynamic programming device provided by an embodiment of the present application.

[0013] Figure 3 FIG. 2 is a structural schematic diagram of a terminal device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0014] In order to make the objects, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not intended to limit the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of the present application.

[0015] It should be noted that the terms "comprise", "contain", and "have" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover the inclusions without exclusivity. For example, a process, method, terminal, product or device comprising a series of steps or units is not limited to the listed steps or units, but optionally further comprises steps or units not listed, or optionally further comprises other steps or units inherent to the process, method, product or device. In the claims, specification and drawings of the present application, the relationship terms such as "first" and "second" and the like are only used to distinguish one entity / operation / object from another entity / operation / object, and do not necessarily require or imply any such real-time relationship or sequence between the entities / operations / objects.

[0016] Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment can be included in at least one embodiment of the present application. The appearance of the phrase in various places in the specification does not necessarily all refer to the same embodiment, nor is it necessarily independent or alternative embodiments to other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0017] In global supply chain management, cross-border e-commerce logistics, and other industries expanding overseas, route planning is a core element. Traditional route planning systems typically rely on historical operational data, assigning fixed weights to objectives such as transportation costs and time for static optimization. However, customs policies and tariff regulations in various countries often undergo sudden changes (e.g., a country suddenly announces additional import tariffs on specific goods), and traditional methods struggle to identify, analyze, and incorporate their impact in real time. This can easily lead to planned routes that are incompatible with the latest policy environment, potentially choosing routes that are already facing high tariffs or customs clearance obstacles, resulting in soaring logistics costs, transportation delays, and other serious problems. In other words, existing technologies struggle to respond in real time and integrate the impact of policy changes when planning routes in a dynamic policy environment.

[0018] In view of this, this application provides a method for dynamic planning of cross-border multimodal transport routes. By acquiring and integrating real-time data from multiple sources, and parsing and quantifying the policy information therein, unstructured policy texts are transformed into dynamic constraints that can be understood by a multi-objective optimization model. This enables dynamic optimization of route planning, allowing for real-time responses to changes in the external policy environment. The generated route plan achieves a better balance among multiple objectives such as compliance, cost, and timeliness, improving the accuracy, timeliness, and risk resistance of the planning results.

[0019] To illustrate the technical solution of this application, specific embodiments are described below.

[0020] Figure 1 This illustration shows a flowchart of a dynamic planning method for cross-border multimodal transport routes provided in an embodiment of this application. This method can be applied to terminal devices. Terminal devices can be servers, service clusters, mobile phones, tablets, laptops, ultra-mobile personal computers (UMPCs), netbooks, etc.

[0021] Specifically, the above-mentioned dynamic planning method for cross-border multimodal transport routes may include the following steps S101 to S105.

[0022] Step S101: Acquire multi-source real-time data and fuse the multi-source real-time data to obtain fused data.

[0023] Multi-source real-time data refers to the latest data obtained from different sources, which may include policy documents, freight rates, port status, weather information, etc.

[0024] In the embodiments of the present application, the terminal device can collect information from multiple heterogeneous data sources such as the customs offices of various governments, shipping exchanges, satellite remote sensing service providers, etc. in real time through a data interface. The specific multi-source real-time data can include unstructured policy announcement texts, structured shipping rate tables, satellite images reflecting port congestion, etc. Then the terminal device can clean, format convert and spatio-temporally align the data of different formats and standards (for example, associate a certain policy with the specific port it affects and the effective time point), and finally generate a fusion data that is internally unified and contains multi-dimensional information, thereby breaking the data silos.

[0025] Step S102, policy analysis is performed on the fusion data to obtain policy evaluation information.

[0026] The policy evaluation information is used to judge the influence of the policy on the transportation path.

[0027] In the embodiments of the present application, the terminal device can use natural language processing technology to automatically analyze the multi-language policy texts in the fusion data, identify key provisions (such as prohibition of passage, change of customs duty, special document requirements, etc.), applicable objects, effective regions and effective periods, and evaluate the potential influence of the policy in combination with other information (such as port congestion shown by satellite images), and finally generate structured policy evaluation information, such as "a country imposes a 30% customs duty on A type of goods, which is expected to significantly increase the cost of paths passing through the country". The role of this step is to convert unstructured texts that are difficult to calculate directly into preliminary judgments about the influence of the policy that can be understood by subsequent models.

[0028] Step S103, quantifying the policy evaluation information into dynamic constraint conditions.

[0029] The dynamic constraint condition refers to a limiting condition or weight parameter that is adjusted in real time in the optimization model as the external environment (such as the policy) changes.

[0030] In the embodiments of the present application, the terminal device can convert the specific influence of the policy evaluation information into mathematical parameters. For example, "imposing a 30% customs duty" can be quantified as a "cost weight increase coefficient" of the relevant path in the cost optimization target, or "a port is closed due to policy restrictions" can be quantified as a hard "path feasibility constraint". Then these parameters are input into the optimization model as dynamic constraint conditions. The embodiments of the present application can convert qualitative policy influences into quantified indicators that can be recognized and processed by the optimization algorithm, so that the mathematical model can respond to policy changes in real time.

[0031] Step S104, inputting the dynamic constraint condition and the fused data into a multi-objective optimization model to obtain a Pareto optimal solution set, the Pareto optimal solution set containing at least one cross-border multimodal transport path scheme.

[0032] The multi-objective optimization model refers to a mathematical model that needs to optimize multiple objectives (such as the lowest cost, the shortest time, and the best compliance) at the same time.

[0033] The Pareto optimal solution set refers to a set of optimal solutions in a multi-objective optimization problem, and any solution in the set cannot become better in a certain objective without deteriorating in other objectives.

[0034] In the embodiments of the present application, the terminal device can run a pre-constructed multi-objective optimization algorithm (such as a genetic algorithm), and the optimization objectives of the model include at least total cost, total time length, and compliance. The terminal device can input the dynamic constraint condition (such as the adjusted cost weight) and the real-time market data (such as the latest freight rate) in the fused data into the model for solving. The model then outputs a Pareto optimal solution set, each solution in the set representing a path scheme that achieves the best balance between cost, time efficiency, and compliance, but each solution has its own emphasis. The embodiments of the present application can automatically and quickly find a series of scientific optimal candidate schemes in the complex multi-objective trade-off, replacing the inefficient manual planning that relies on experience.

[0035] Step S105, determining a target path scheme from the Pareto optimal solution set.

[0036] In the embodiments of the present application, the terminal device can select a scheme with the highest comprehensive score or the most suitable for the current business demand from the solution set according to the user's preset preference (for example, whether the cost or the time efficiency is more important for this transportation) or through the received instructions of human-computer interaction, and determine the scheme as the target path scheme, thereby introducing the user's preference or final decision on the basis of the multiple possibilities calculated by the machine, ensuring that the output scheme is not only scientifically optimal, but also meets the actual business demand.

[0037] Compared with the prior art, the embodiment of the present application has the beneficial effects that: the embodiment of the present application obtains multi-source real-time data, fuses the multi-source real-time data to obtain fused data, performs policy analysis on the fused data to obtain policy evaluation information, quantifies the policy evaluation information into a dynamic constraint condition, inputs the dynamic constraint condition and the fused data into a multi-objective optimization model to obtain a Pareto optimal solution set, and determines a target path scheme from the Pareto optimal solution set. The embodiment of the present application obtains and fuses multi-source real-time data, analyzes and quantifies policy information, converts unstructured policy text into a dynamic constraint condition understandable by a multi-objective optimization model, thereby realizing dynamic optimization of path planning, responding to changes in an external policy environment in real time, making the generated path scheme achieve a more optimal balance in compliance, cost and timeliness and the like, and improving the accuracy, timeliness and anti-risk ability of the planning result.

[0038] In some embodiments of the present application, the multi-source real-time data includes multi-language unstructured policy text data, structured transport market data, and satellite remote sensing image data, and the obtaining of the multi-source real-time data and the fusion of the multi-source real-time data to obtain fused data can specifically include steps S401 to S406.

[0039] Step S401: The multi-source real-time data from multiple heterogeneous data sources is collected in parallel.

[0040] The heterogeneous data sources refer to original data providers with different technical architectures, data formats and communication protocols.

[0041] In the embodiments of the present application, the terminal device can configure multiple data interface clients to simultaneously asynchronously capture the latest data streams from different types of data such as official websites of customs in various countries, commercial satellite data providers, and databases of shipping exchanges.

[0042] Step S402: The policy text data is cleaned and segmented to obtain processed policy text data.

[0043] In the embodiments of the present application, the terminal device can first remove noise such as webpage formats and irrelevant symbols through a preset rule and algorithm, and then can use a segmentation tool in natural language processing to cut continuous policy text strings into independent and semantic word sequences, thereby converting unstructured original text into standardized text data for deep semantic analysis.

[0044] Step S403: The satellite remote sensing image data is denoised and key region cropped to obtain processed satellite remote sensing image data.

[0045] In the embodiments of the present application, the terminal device can first apply an image filtering algorithm to reduce interference caused by cloud cover and sensor errors, and then use a target detection algorithm to locate and intercept the area containing key logistics nodes such as ports and warehouses in the image, which can greatly reduce the data volume on the premise of retaining valuable information, thereby improving the subsequent processing efficiency.

[0046] In step S404, the abnormal value correction and normalization are performed on the transport market data to obtain processed transport market data.

[0047] In the embodiments of the present application, the terminal device can identify and correct abnormal values caused by input errors or transient fluctuations through statistical methods, and scale the price and capacity data of different sources and dimensions to a unified numerical interval, thereby eliminating data bias and establishing a fair benchmark for subsequent multi-source data fusion and comparison.

[0048] In step S405, a text feature vector is extracted from the processed policy text data, a visual feature vector is extracted from the processed satellite remote sensing image data, and a numerical feature vector is extracted from the processed transport market data.

[0049] In the embodiments of the present application, the terminal device can use a word embedding model and a convolutional neural network feature extractor to respectively map the cleaned text, image and numerical data to numerical vectors in a high-dimensional space, so that the terminal device can perform unified mathematical operations thereon.

[0050] In step S406, the text feature vector, the visual feature vector and the numerical feature vector are spatio-temporally aligned and mapped to the same feature space to obtain the fusion data.

[0051] In the embodiments of the present application, the terminal device can associate the feature vectors of different modalities to the same spatio-temporal point according to the time stamp and geographic coordinates, and then project them into a shared low-dimensional feature space (in which the data of different sources are comparable and can complement each other) through a deep learning model (such as a cross-modal encoder), and finally generate a deep integrated and internally consistent comprehensive data representation.

[0052] The embodiments of the present application can improve the data quality and internal consistency from the root by performing a complete technical process from parallel collection, respective preprocessing, feature extraction to deep fusion in a unified feature space, and realize high concentration and computability of information through deep feature extraction of unstructured data, and finally generate a fusion data that can be directly utilized.

[0053] In some more specific embodiments of the present application, the policy analysis on the fused data to obtain policy evaluation information can specifically include steps S501-S504.

[0054] In step S501, a pre-trained multilingual large language model is called to perform semantic understanding on the fused data, and key policy clauses, their constraint objects and effective conditions are extracted to obtain structured policy evaluation information.

[0055] In embodiments of the present application, the terminal device can call a multilingual large language model (such as XLM-RoBERTa or GPT) through an application program interface, input the preprocessed policy text data in the fused data into the model, and then use the semantic understanding capability built in the model to automatically identify and extract the core clause content of the policy (such as prohibiting or restricting the passage of certain goods), the specific object to which the clause applies (such as commodity category, transportation tool), and the time and geographical range in which the clause takes effect, and finally output a neat, field-specific structured information table or JSON object, thereby converting the natural language description that is difficult to directly calculate into logical data that can be accurately processed.

[0056] In step S502, image recognition is performed on the satellite remote sensing image data to obtain traffic hub state information.

[0057] In embodiments of the present application, the terminal device can analyze the satellite remote sensing image in the fused data using a computer vision model to identify visual features such as the density of containers in the yard, the number of ships parked in the port, and the traffic flow on the approach road, and determine the current operation state level of the traffic hub (such as smooth, slight congestion, and severe congestion) according to these features, thereby evaluating the implementation impact of the policy in the actual environment.

[0058] In step S503, a quantitative evaluation result is generated according to the structured policy evaluation information and the traffic hub state information.

[0059] In embodiments of the present application, the terminal device can perform correlation analysis on the structured policy clause (such as “a certain port implements temporary embargo on A type goods”) and the traffic hub state (such as “the port is currently severely congested”) identified from the satellite remote sensing image data according to pre-defined rules or small prediction models, thereby calculating specific numerical evaluation results such as expected delay days, additional cost percentage, or compliance risk level that the policy may cause in the current actual environment, for example, if the policy causes a port to close and the satellite image shows that the port is congested, the time delay risk is evaluated, thereby quantifying the policy impact.

[0060] In step S504, the quantitative evaluation results of all policy clauses are integrated to generate the policy evaluation information.

[0061] In the embodiments of the present application, the terminal device can combine the quantitative evaluation results (such as cost increase, time delay, risk level, etc.) of all identified relevant policy provisions into a policy evaluation report containing multi-dimensional scores through weighted averaging, aggregation or other aggregation algorithms, thereby clearly reflecting the potential impact of the current policy environment on the overall path planning in each optimization target, and providing a direct basis for subsequent quantitative constraints.

[0062] The embodiments of the present application realize deep semantic understanding and automatic information extraction of unstructured policy text by using a multi-language large language model, and combine image recognition technology to analyze the physical environment state of policy implementation, and finally generate quantitative policy evaluation information, which greatly improves the analysis accuracy and efficiency of multi-language and unstructured policy information. Combining text semantics with real situation, the policy evaluation result is more objective and accurate, which provides high-quality and quantifiable decision input for subsequent optimization models, and fundamentally solves the path planning error caused by the inability of traditional methods to effectively analyze policies.

[0063] In some specific embodiments of the present application, the step of quantifying the policy evaluation information into dynamic constraint conditions can specifically include steps S601 to S603.

[0064] Step S601: Analyzing the policy evaluation information to identify the affected optimization dimension corresponding to the policy provision.

[0065] Among them, the optimization dimension includes at least one of compliance, cost and timeliness.

[0066] In the embodiments of the present application, the terminal device can analyze the provision content contained in the policy evaluation information to determine the influence of the provision on the path scheme, for example, if the policy content is to impose additional tariffs, it mainly affects the "cost" dimension, if the policy content is to add quarantine procedures, it mainly affects the "timeliness" dimension, and if the policy content is to prohibit passage, it directly affects the "compliance" dimension.

[0067] Step S602: Calculating the cost weight adjustment factor and the timeliness weight adjustment factor corresponding to the affected optimization dimension.

[0068] Among them, the weight adjustment factor is used to dynamically modify the importance or priority of a certain optimization target (such as cost or timeliness) in the overall objective function in the optimization algorithm.

[0069] In the embodiments of the present application, the terminal device can calculate a corresponding weight adjustment factor according to the quantified impact degree (such as the percentage of the predicted increase in cost, the number of days of the predicted delay in time) in the policy evaluation information through a predefined mapping rule or function. For example, if it is identified that the “cost” dimension is affected and the evaluation result is that the cost will increase by 30%, the weight adjustment factor of 0.5 can be mapped to the weight increase in the cost objective function. In addition, if the time delay is 20%, the time weight adjustment factor can be set to 1.2, so as to accurately translate the external influence of the policy into a parameter instruction that can be understood and responded by the internal optimization model.

[0070] In step S603, the dynamic constraint condition is generated according to the cost weight adjustment factor and the time weight adjustment factor.

[0071] In the embodiments of the present application, the terminal device can encapsulate one or more calculated weight adjustment factors to combine a standard parameter configuration instruction set, which can be used to guide the optimization algorithm to adjust the weight distribution of different parts of the objective function in the next solution, for example, to increase the weight of the cost term to avoid high-cost routes. The embodiments of the present application can complete the final conversion from policy information to model executable instructions, so as to effectively inject the dynamic policy influence into the static optimization model.

[0072] The embodiments of the present application can accurately quantify the abstract policy influence into parameters that can be directly used by the optimization model by analyzing the policy evaluation information into specific optimization dimensions and calculating the corresponding weight adjustment factor to generate the dynamic constraint condition, so as to effectively connect the external dynamic policy to the internal mathematical model, overcome the problem that the traditional static model cannot respond to policy changes, and ensure that the planned path can be real-time adapted to the latest policy environment in terms of cost, time and other key targets, significantly improving the practicality and economy of the planning result.

[0073] In some specific embodiments of the present application, the dynamic constraint condition and the fused data are input into a multi-objective optimization model to obtain a Pareto optimal solution set, which can specifically include steps S701 to S703.

[0074] In step S701, a multi-objective optimization model is constructed, and a non-dominated sorting genetic algorithm with an elite strategy is used as a solver of the multi-objective optimization model.

[0075] The non-dominated sorting genetic algorithm with an elite strategy (NSGA-II) is a commonly used multi-objective optimization algorithm, which can search for optimal solutions by simulating “survival of the fittest” and “gene inheritance” in natural selection.

[0076] In the embodiments of the present application, the terminal device can first define a mathematical function with compliance, total cost and transportation timeliness as the core optimization objectives, and explicitly define the decision variable as path selection, and then select the NSGA-II algorithm as the engine for solving the model. The NSGA-II algorithm can effectively handle the conflict relationship between multiple objectives and ensure the search efficiency and solution quality through its built-in elite reservation mechanism. Specifically, the objective function of the multi-objective optimization model can be represented as: Minimize [w1 * Cost + w2 * Time + w3 * (1 - Compliance)]. Wherein, Cost is the total cost, Time is the transportation timeliness, Compliance is the compliance score (defined as the probability value based on the risk of policy violation, ranging from 0 to 1, 1 indicating full compliance), w1, w2, w3 are dynamic weights that can be dynamically adjusted according to policy impact.

[0077] Step S702, input the dynamic constraint conditions and the transport capacity market data in the fusion data as input parameters into the multi-objective optimization model.

[0078] In the embodiments of the present application, the terminal device can set the dynamic constraint conditions (such as the adjusted cost weight) and the transport capacity market data (such as the real-time updated freight and cabin price) reflecting the current market situation in the fusion data as specific parameters for this run of the model, so that the rules and cost information based on each path planning are always up-to-date, thereby ensuring that the model can be solved under conditions that truly reflect the current policy and market environment.

[0079] Step S703, run the multi-objective optimization model to search and iterate, and generate a set of Pareto optimal solutions representing the trade-off relationship between different objectives, wherein each solution corresponds to a complete cross-border multimodal transport path scheme.

[0080] In the embodiments of the present application, the terminal device can start the NSGA-II solver to perform multiple rounds of iterative search in the huge solution space composed of all possible paths, and in each generation, new schemes are generated through genetic operations such as selection, crossover and mutation, and the scheme set that achieves the best balance in compliance, cost and timeliness is selected according to the non-dominated sorting, and finally a set of Pareto optimal solutions is output, each solution in the set explicitly corresponds to a specific complete transport path sequence with different emphases among multiple objectives, providing users with multiple scientific optimal choices.

[0081] The multi-objective optimization solver of the non-dominated sorting genetic algorithm with elitist strategy is adopted in the embodiments of the present application, and dynamic constraint conditions and real-time transport market data are taken as model inputs, so that the search in the complex multi-objective trade-off space can be efficiently and automatically performed, and a series of Pareto optimal path schemes achieving the best balance in compliance, cost and timeliness can be generated at one time, thereby providing the user with diversified and high-quality decision options based on real-time data and accurate mathematical calculation, and overcoming the problem of low efficiency and difficulty in considering multiple objectives in manual planning.

[0082] In some embodiments of the present application, the determining of the target path scheme from the Pareto optimal solution set can include steps S801 to S803.

[0083] Step S801 is to calculate the trade-off score of each path scheme in the Pareto optimal solution set under different optimization objectives.

[0084] The optimization objectives include compliance, total cost and transportation timeliness.

[0085] In the embodiments of the present application, the terminal device can convert the original numerical values (for example, specific cost amount and transportation days) of each path scheme in compliance, total cost and transportation timeliness into dimensionless scores by using a predefined scoring function, so as to generate an objective evaluation vector including multi-dimensional scores for each scheme.

[0086] Step S802 is to obtain the preference weights of different optimization objectives, and to generate a comprehensive priority score corresponding to each path scheme by weighting and integrating the trade-off scores of each path scheme according to the preference weights.

[0087] In the embodiments of the present application, the terminal device can receive the importance weight set by the user for the cost, timeliness and other objectives through a human-computer interaction interface, and then calculate a comprehensive priority score of each path scheme by using a linear weighted summation method to multiply and add the trade-off scores of the path scheme in each objective and the corresponding preference weight, so as to convert the mathematical optimal solution into a candidate scheme meeting the actual business requirements.

[0088] Step S803 is to determine the path scheme with the highest comprehensive priority score as the target path scheme.

[0089] In the embodiments of the present application, the terminal device can compare the comprehensive priority scores calculated for all path schemes, and select the scheme with the highest score as the final target path scheme, without the need for manual repeated comparison.

[0090] The implementation of the present application first quantitatively scores the schemes in the Pareto optimal solution set, then introduces user preference weights for weighted decision, and finally automatically selects the scheme with the highest comprehensive score as the execution target, which not only ensures that the finally selected path is a good solution based on multi-objective optimization in mathematics, but also ensures that the solution meets the user's current actual business priority to the greatest extent, effectively improving the decision efficiency and user satisfaction.

[0091] In some embodiments of the present application, after determining the target path scheme from the Pareto optimal solution set, steps S901 to S903 can also be included.

[0092] Step S901, according to the regulations, language and cultural customs of the destination country involved in the target path scheme, generate a customs clearance strategy.

[0093] In the embodiments of the present application, the terminal device can call a large language model, taking the customs regulations, official language requirements and potential cultural and religious taboos of the specific destination country determined by the target path scheme as input, to generate a draft of customs clearance documents and a suggestion of declaration process that meet the specific requirements of the destination country, that is, a customs clearance strategy, thereby extending intelligent planning to the key customs clearance operation level and avoiding risks caused by non-compliant documents in advance.

[0094] Step S902, execute the target path scheme and monitor the status of key nodes in real time during execution to collect actual transportation data.

[0095] In the embodiments of the present application, the terminal device can continuously track the status and timestamp of the goods at key nodes (such as departure from port, arrival at port, start of customs declaration, and completion of customs declaration) of the path through integrated GPS positioning, port information system, and electronic lock data interface, and record the actual fees and other information, thereby collecting a complete set of full-process data reflecting the actual situation of this transportation.

[0096] Step S903, compare the actual transportation data with the expected data to generate feedback information, and dynamically adjust the parameters of the multi-objective optimization model according to the feedback information.

[0097] In the embodiments of the present application, the terminal device can compare the actual transportation data (such as actual customs clearance time and actual fuel cost) with the expected data when the model is initially planned, thereby calculating the deviation value, and then using a reinforcement learning algorithm to use the deviation value as a training signal to backpropagate and fine-tune the related parameters in the optimization model (such as the coefficients of the time estimation function and the weights of the cost calculation model), so that the model can make more accurate predictions in the next planning.

[0098] The embodiments of the present application can significantly improve the passing efficiency of the clearance link, continuously calibrate and optimize the model prediction accuracy by using real operation data, thereby improving the reliability, adaptability and overall efficiency of the whole cross-border multimodal transport path planning.

[0099] Figure 3 A structure schematic diagram of a cross-border multimodal transport path dynamic planning device provided by an embodiment of the present application is shown. The cross-border multimodal transport path dynamic planning device 2 can be configured on a terminal device. Specifically, the cross-border multimodal transport path dynamic planning device 2 can include: The acquisition module 201 is configured to acquire multi-source real-time data, fuse the multi-source real-time data, and obtain fused data. The analysis module 202 is configured to analyze the fused data according to policies, and obtain policy evaluation information. The quantification module 203 is configured to quantize the policy evaluation information into dynamic constraint conditions. The input module 204 is configured to input the dynamic constraint conditions and the fused data into a multi-objective optimization model, obtain a Pareto optimal solution set, and the Pareto optimal solution set contains at least one cross-border multimodal transport path scheme. The determination module 205 is configured to determine a target path scheme from the Pareto optimal solution set.

[0100] Compared with the prior art, the embodiments of the present application have the beneficial effects that: the embodiments of the present application acquire multi-source real-time data, fuse the multi-source real-time data, obtain fused data, analyze the fused data according to policies, obtain policy evaluation information, quantize the policy evaluation information into dynamic constraint conditions, input the dynamic constraint conditions and the fused data into a multi-objective optimization model, obtain a Pareto optimal solution set, and determine a target path scheme from the Pareto optimal solution set. The embodiments of the present application acquire and fuse multi-source real-time data, analyze and quantize policy information, convert unstructured policy text into dynamic constraint conditions that can be understood by a multi-objective optimization model, thereby realizing dynamic optimization of path planning, responding to changes in the external policy environment in real time, making the generated path scheme achieve a better balance in compliance, cost and timeliness, and improving the accuracy, timeliness and anti-risk ability of the planning result.

[0101] In some embodiments of the present application, the acquisition module 201 is further configured to: acquire the multi-source real-time data from multiple heterogeneous data sources in parallel; clean and tokenize the policy text data to obtain processed policy text data; The satellite remote sensing image data is denoised and key region cropped to obtain processed satellite remote sensing image data; The capacity market data is subjected to outlier correction and normalization to obtain processed capacity market data; Text feature vectors are extracted from the processed policy text data, visual feature vectors are extracted from the processed satellite remote sensing image data, and numerical feature vectors are extracted from the processed capacity market data; The text feature vectors, the visual feature vectors and the numerical feature vectors are spatio-temporally aligned and mapped to the same feature space to obtain the fusion data.

[0102] In some embodiments of the present application, the above-mentioned analysis module 202 is further used for: A pre-trained multilingual large language model is called to perform semantic understanding on the fusion data, and key policy provisions, their constraint objects and effective conditions are extracted to obtain structured policy evaluation information; Image recognition is performed on the satellite remote sensing image data to obtain transportation hub state information; Quantitative evaluation results are generated according to the structured policy evaluation information and the transportation hub state information; The quantitative evaluation results of all policy provisions are integrated to generate the policy evaluation information.

[0103] In some embodiments of the present application, the above-mentioned quantitative module 203 is further used for: The policy evaluation information is analyzed to identify the affected optimization dimensions corresponding to the policy provisions, the optimization dimensions including at least one of compliance, cost and timeliness; Cost weight adjustment factors and timeliness weight adjustment factors corresponding to the affected optimization dimensions are calculated; The dynamic constraint conditions are generated according to the cost weight adjustment factors and the timeliness weight adjustment factors.

[0104] In some embodiments of the present application, the above-mentioned input module 204 is further used for: A multi-objective optimization model is constructed, and a non-dominated sorting genetic algorithm with an elitist strategy is used as a solver of the multi-objective optimization model; The dynamic constraint conditions and the capacity market data in the fusion data are input as input parameters into the multi-objective optimization model; The multi-objective optimization model is run to search and iterate, and a set of Pareto optimal solutions representing trade-off relationships between different objectives are generated, wherein each solution corresponds to a complete cross-border multimodal transport path scheme.

[0105] In some embodiments of the present application, the determination module 205 is further configured to: calculate the trade-off score of each path scheme in the set of Pareto optimal solutions under different optimization objectives, including compliance, total cost, and transportation timeliness; obtain the preference weights of different optimization objectives, and weight and integrate the trade-off score of each path scheme according to the preference weights to generate a comprehensive priority score corresponding to each path scheme; determine the path scheme with the highest comprehensive priority score as the target path scheme.

[0106] In some embodiments of the present application, the cross-border multimodal transport path dynamic programming device 2 can further include an adjustment module configured to: generate a customs clearance strategy according to the regulations, language, and cultural customs of the destination country involved in the target path scheme; execute the target path scheme and monitor the status of key nodes in real time during execution to collect actual transportation data; compare the actual transportation data with the expected data to generate feedback information, and dynamically adjust the parameters of the multi-objective optimization model according to the feedback information.

[0107] As shown in Figure 3 , it is a schematic diagram of a terminal device provided by an embodiment of the present application. The terminal device 3 can include a processor 301, a memory 302, and a computer program 303 stored in the memory 302 and executable on the processor 301, such as a cross-border multimodal transport path dynamic programming program. The processor 301 executes the computer program 303 to implement the steps in each of the above cross-border multimodal transport path dynamic programming embodiments, such as Figure 1 the steps S101 to S105 shown in the figure.

[0108] The computer program can be divided into one or more modules / units, one or more modules / units are stored in the memory 302 and executed by the processor 301 to complete the present application. One or more modules / units can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the terminal device.

[0109] The terminal device can include, but is not limited to, the processor 301, the memory 302. Those skilled in the art can understand that Figure 3 is only an example of the terminal device and does not constitute a limitation on the terminal device, and can include more or fewer components than the illustration, or combine certain components, or different components, for example, the terminal device can also include an input / output device, a network access device, a bus, etc.

[0110] The processor 301 can be a central processing unit (CPU), and can also be other general-purpose processors, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware component, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0111] The memory 302 can be an internal storage unit of the terminal device, for example, a hard disk or a memory of the terminal device. The memory 302 can also be an external storage device of the terminal device, for example, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the terminal device. Further, the memory 302 can also include both the internal storage unit and the external storage device of the terminal device. The memory 302 is used to store computer programs and other programs and data required by the terminal device. The memory 302 can also be used to temporarily store data that has been output or will be output.

[0112] It should be noted that, for the convenience and brevity of description, the structure of the terminal device can also be referred to the specific description of the structure in the method embodiments, which will not be repeated here.

[0113] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above functional units and modules is exemplified, and in actual application, the above functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of software function unit. In addition, the specific name of each functional unit and module is only for easy distinction, and does not limit the protection scope of the present application. The specific working process of the unit and module in the system can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

[0114] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can implement the steps in the above-described cross-border multimodal transport route dynamic planning method.

[0115] This application provides a computer program product that, when run on a mobile terminal, enables the mobile terminal to implement the steps in the above-mentioned dynamic planning method for cross-border multimodal transport routes.

[0116] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0117] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for various specific applications, but such implementations should not be considered beyond the scope of this application.

[0118] In the embodiments provided in this application, it should be understood that the disclosed apparatus / terminal devices and methods can be implemented in other ways. For example, the apparatus / terminal device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0119] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0120] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0121] The integrated module / unit, if implemented in the form of a software functional unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. When the computer program is executed by a processor, the steps of the above-mentioned various method embodiments can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable files or some intermediate forms. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content included in the computer readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer readable medium does not include electric carrier signals and telecommunication signals.

[0122] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for some technical features. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A dynamic planning method for cross-border multimodal transport routes, characterized in that, include: Acquire real-time data from multiple sources and fuse the real-time data from multiple sources to obtain fused data; The fused data is analyzed to obtain policy evaluation information; The policy evaluation information is quantified into dynamic constraints. The dynamic constraints and the fused data are input into a multi-objective optimization model to obtain a Pareto optimal solution set, which includes at least one cross-border multimodal transport route scheme. The target path scheme is determined from the Pareto optimal solution set.

2. The dynamic planning method for cross-border multimodal transport routes as described in claim 1, characterized in that, The multi-source real-time data includes multilingual unstructured policy text data, structured transportation capacity market data, and satellite remote sensing image data. The process of acquiring multi-source real-time data and fusing it to obtain fused data includes: The multi-source real-time data is collected in parallel from multiple heterogeneous data sources; The policy text data is cleaned and segmented to obtain processed policy text data; The satellite remote sensing image data is denoised and key regions are cropped to obtain the processed satellite remote sensing image data. The capacity market data is corrected for outliers and normalized to obtain the processed capacity market data. Text feature vectors are extracted from the processed policy text data, visual feature vectors are extracted from the processed satellite remote sensing image data, and numerical feature vectors are extracted from the processed capacity market data. The text feature vector, the visual feature vector, and the numerical feature vector are spatiotemporally aligned and mapped to the same feature space to obtain the fused data.

3. The dynamic planning method for cross-border multimodal transport routes as described in claim 2, characterized in that, The process of performing policy analysis on the fused data to obtain policy evaluation information includes: By calling a pre-trained multilingual large language model, semantic understanding is performed on the fused data to extract key policy clauses, their binding objects, and effective conditions, thereby obtaining structured policy evaluation information. Image recognition is performed on the satellite remote sensing image data to obtain the status information of the transportation hub; Based on the structured policy assessment information and the transportation hub status information, a quantitative assessment result is generated; The policy assessment information is generated by synthesizing the quantitative assessment results of all policy provisions.

4. The dynamic planning method for cross-border multimodal transport routes as described in claim 1, characterized in that, The process of quantifying the policy evaluation information into dynamic constraints includes: The policy evaluation information is analyzed to identify the affected optimization dimensions corresponding to the policy provisions. The optimization dimensions include at least one of compliance, cost and timeliness. Calculate the cost weight adjustment factor and the timeliness weight adjustment factor corresponding to the affected optimization dimensions; The dynamic constraints are generated based on the cost weight adjustment factor and the timeliness weight adjustment factor.

5. The cross-border multimodal transport route dynamic planning method as described in claim 1, characterized in that, The step of inputting the dynamic constraints and the fused data into a multi-objective optimization model to obtain a Pareto optimal solution set includes: A multi-objective optimization model is constructed, and a non-dominated sorting genetic algorithm with an elitist strategy is used as the solver for the multi-objective optimization model. The dynamic constraints and the capacity market data in the fused data are used as input parameters to input the multi-objective optimization model. The multi-objective optimization model is run to perform search and iteration, generating a set of Pareto optimal solutions that represent the trade-offs between different objectives, where each solution corresponds to a complete cross-border multimodal transport route scheme.

6. The cross-border multimodal transport route dynamic planning method as described in claim 1, characterized in that, Determining the target path scheme from the Pareto optimal solution set includes: Calculate the trade-off score for each path in the Pareto optimal solution set under different optimization objectives, including compliance, total cost, and transportation timeliness; Obtain the preference weights for different optimization objectives, and perform a weighted summation of the trade-off scores for each path scheme based on the preference weights to generate a comprehensive priority score for each path scheme. The path scheme with the highest overall priority score is determined as the target path scheme.

7. The cross-border multimodal transport route dynamic planning method as described in claim 1, characterized in that, After determining the target path scheme from the Pareto optimal solution set, the method further includes: Based on the laws, languages, and cultural customs of the destination countries involved in the target route plan, a customs clearance strategy is generated; The target route plan is executed, and the status of key nodes is monitored in real time during the execution process, and actual transportation data is collected. The actual transportation data is compared with the expected data to generate feedback information, and the parameters of the multi-objective optimization model are dynamically adjusted based on the feedback information.

8. A dynamic planning device for cross-border multimodal transport routes, characterized in that, The device includes: The acquisition module is used to acquire real-time data from multiple sources and fuse the real-time data from multiple sources to obtain fused data. The parsing module is used to perform policy analysis on the fused data to obtain policy evaluation information; The quantification module is used to quantify the policy evaluation information into dynamic constraints. The input module is used to input the dynamic constraints and the fused data into the multi-objective optimization model to obtain the Pareto optimal solution set, which includes at least one cross-border multimodal transport route scheme. The determination module is used to determine the target path scheme from the Pareto optimal solution set.

9. A terminal device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the cross-border multimodal transport route dynamic planning method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the cross-border multimodal transport route dynamic planning method as described in any one of claims 1 to 7.

Images