Cross-border trade risk solution generation method and device, equipment and medium

Abstract

The invention relates to the technical field of cross-border trade, and discloses a cross-border trade risk solution generation method and device, equipment and a medium, and the method comprises the steps: obtaining a plurality of pieces of first real-time data information related to a specified cross-border trade; detecting whether the first real-time data information contains risk data information or not; when the first real-time data information contains risk data information, inputting the risk data information into a target simulation engine to obtain a target solution; and solving the risk brought by the risk data information based on the solution. The method has the advantages that the target simulation engine is introduced, multiple crisis scenes are constructed based on the large language model, influences under different risk situations can be comprehensively understood and predicted, the flexibility of the algorithm enables enterprises to timely generate an optimization solution for a dynamic international market environment, and the risk prediction efficiency is improved. And the loss caused by emergencies is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of cross-border trade, and in particular, to a method, device, equipment and medium for generating solutions to cross-border trade risks. Background Art

[0002] With the acceleration of the global economic integration process, the importance of cross-border trade in the international market has become increasingly prominent. However, this field is also full of uncertainties and risks. Some unpredictable and significant events often lead to supply chain disruptions, causing problems such as trade delays and increased costs.

[0003] In traditional cross-border trade decision-making, enterprises often rely on historical data and static models for risk assessment and decision-making. However, these methods have limited capabilities in dealing with rapidly changing market environments, with a long decision-making time and difficulty in quickly responding to emergencies. Summary of the Invention

[0004] Based on this, it is necessary to propose a method, device, equipment and medium for generating solutions to existing cross-border trade risks.

[0005] A method for generating solutions to cross-border trade risks, the method comprising: Obtaining a plurality of first real-time data information related to a specified cross-border trade; Detecting whether the first real-time data information contains risk data information; When the first real-time data information contains risk data information, inputting the risk data information into a target simulation engine to obtain a target solution; Based on the solution, solving the risks brought by the risk data information; Wherein, the training method of the target simulation engine includes: Obtaining a plurality of second real-time data information; Inputting the plurality of second real-time data information into a preset simulation engine to obtain a target simulation engine; Based on a preset large language model, constructing a plurality of crisis scenarios in the target simulation engine, and generating solutions for each crisis scenario in combination with preset rules.

[0006] Further, the step of when the first real-time data information contains risk data information, inputting the risk data information into a target simulation engine to obtain a solution includes: Extracting risk characteristics of the risk data information according to a preset semantic parsing model; Performing similarity matching between the risk characteristics and the pre-stored characteristics of each crisis scenario; Extract the crisis scenario with the highest similarity as the target crisis scenario, and determine whether the corresponding similarity value is greater than a preset similarity value; If the corresponding similarity value is greater than the preset similarity value, select the solution corresponding to the target crisis scenario as the target solution.

[0007] Further, after the step of extracting the crisis scenario with the highest similarity as the target crisis scenario and determining whether the corresponding similarity value is greater than a preset similarity value, the following steps are also included: If the corresponding similarity value is less than or equal to the preset similarity value, input the risk data information into the target simulation engine to construct a corresponding target crisis scenario based on a preset large language model; Generate a target solution for the target crisis scenario in combination with the preset rules.

[0008] Further, after the step of constructing multiple crisis scenarios based on a preset large language model in the target simulation engine and generating solutions for each crisis scenario in combination with preset rules, the following steps are also included: Based on the solution, obtain the corresponding third real-time data; Based on the third real-time data, use an improved non-dominated sorting genetic algorithm to generate multiple solution strategies for each crisis scenario; Obtain the dimension values corresponding to multiple preset dimensions of each solution strategy, and perform weighted summation of the respective dimension values to obtain the index values corresponding to each solution strategy; Select the solution strategy with the smallest index value as the target solution corresponding to the crisis scenario.

[0009] Further, after the step of selecting the solution strategy with the smallest index value as the solution to the corresponding crisis scenario, the following steps are also included: Input the target solution and the third real-time data into the target simulation engine to adjust the parameters of the target simulation engine, thereby obtaining an adjusted simulation engine.

[0010] Further, the step of using an improved non-dominated sorting genetic algorithm to generate multiple solution strategies for each crisis scenario based on the third real-time data also includes: Obtain manually set parameters; Based on the manually set parameters and the third real-time data, use an improved non-dominated sorting genetic algorithm to generate multiple solution strategies for each crisis scenario.

[0011] Further, after the step of constructing multiple crisis scenarios based on a preset large language model in the target simulation engine and generating solutions for each crisis scenario in combination with preset rules, the method further includes: Obtain a first predicted cost corresponding to the solution of the target simulation engine, send the solution to a specified terminal, and obtain a second predicted cost uploaded by the specified terminal; Calculate whether the first predicted cost exceeds a preset percentage of the second predicted cost; If the first predicted cost exceeds the preset percentage of the second predicted cost, adjust the cost constraint parameter in the target simulation engine and regenerate the solution.

[0012] A device for generating a solution to cross-border trade risks, the device includes: An acquisition module, configured to acquire a plurality of first real-time data information related to a specified cross-border trade; A detection module, configured to detect whether the first real-time data information contains risk data information; An input module, configured to, when the first real-time data information contains risk data information, input the risk data information into a target simulation engine to obtain a target solution; A solution module, configured to solve the risks brought by the risk data information based on the solution; Wherein, the training method of the target simulation engine includes: A second real-time data information acquisition sub-module, configured to acquire a plurality of second real-time data information; A second real-time data information input sub-module, configured to input the plurality of second real-time data information into a preset simulation engine to obtain a target simulation engine; A crisis scenario construction sub-module, configured to construct multiple crisis scenarios based on a preset large language model in the target simulation engine and generate solutions for each crisis scenario in combination with preset rules.

[0013] A computer device includes a memory and a processor. When the computer program stored in the memory is executed by the processor, the processor performs the following steps: Obtain a plurality of first real-time data information related to a specified cross-border trade; Detect whether the first real-time data information contains risk data information; When the first real-time data information contains risk data information, input the risk data information into a target simulation engine to obtain a target solution; Solve the risks brought by the risk data information based on the solution; Among them, the training method of the target simulation engine includes: Obtain a plurality of second real-time data information; Input the plurality of second real-time data information into a preset simulation engine to obtain a target simulation engine; Build a plurality of crisis scenarios in the target simulation engine based on a preset large language model, and generate solutions for each crisis scenario in combination with preset rules.

[0014] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor is caused to execute the following steps: Obtain a plurality of first real-time data information related to specified cross-border trade; Detect whether the first real-time data information contains risk data information; When the first real-time data information contains risk data information, input the risk data information into the target simulation engine to obtain a target solution; Solve the risks brought by the risk data information based on the solution; Among them, the training method of the target simulation engine includes: Obtain a plurality of second real-time data information; Input the plurality of second real-time data information into a preset simulation engine to obtain a target simulation engine; Build a plurality of crisis scenarios in the target simulation engine based on a preset large language model, and generate solutions for each crisis scenario in combination with preset rules.

[0015] The beneficial effects of the present invention: By introducing the target simulation engine and building a plurality of crisis scenarios based on the large language model, it is possible to more comprehensively understand and predict the impacts in different risk situations. The flexibility of this algorithm enables enterprises to timely generate optimized solutions for the dynamic international market environment and reduce losses caused by unexpected events. Description of the Drawings

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0017] Among them: Figure 1 It is an application environment diagram of a method for generating a solution to cross-border trade risks in an embodiment; Figure 2Flowchart of a method for generating a solution to cross-border trade risk in an embodiment; Figure 3 Structural block diagram of a device for generating a solution to cross-border trade risk in an embodiment; Figure 4 Structural block diagram of a computer device in an embodiment. Detailed implementation manners

[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0019] Figure 1 Application environment diagram for generating a solution to cross-border trade risk in an embodiment. Refer to Figure 1 , the method for generating a solution to cross-border trade risk is applied to a system for generating a solution to cross-border trade risk. The system for generating a solution to cross-border trade risk includes a terminal 110 and a server 120. The terminal 110 and the server 120 are connected through a network. The terminal 110 may specifically be a desktop terminal or a mobile terminal, and the mobile terminal may specifically be at least one of a mobile phone, a tablet computer, a notebook computer, etc. The server 120 may be implemented by an independent server or a server cluster composed of multiple servers. The terminal 110 is used to obtain data information in real time, and the server 120 is used to generate a target solution.

[0020] As Figure 2 shown, in an embodiment, a method for generating a solution to cross-border trade risk is provided. This method can be applied to both the terminal and the server. In this embodiment, it is exemplified by being applied to the server. The method for generating a solution to cross-border trade risk specifically includes the following steps: S1: Obtain multiple first real-time data information related to a specified cross-border trade; S2: Detect whether the first real-time data information contains risk data information; S3: When the first real-time data information contains risk data information, input the risk data information into a target simulation engine to obtain a target solution; S4: Solve the risk brought by the risk data information based on the solution; Among them, the training method of the target simulation engine includes: S201: Obtain multiple second real-time data information; S202: Input multiple pieces of the second real-time data information into a preset simulation engine to obtain a target simulation engine; S203: Based on a preset large language model, construct multiple crisis scenarios within the target simulation engine, and generate solutions for each crisis scenario in combination with preset rules.

[0021] As described in step S1 above, obtain multiple pieces of first real-time data information related to specified cross-border trade. Among them, the first real-time data information includes various factors such as market price, exchange rate changes, international transportation status, regulatory changes, weather forecasts, political stability, etc. Specifically, the first real-time data includes geopolitics related to the specified trade (GDELT event database), logistics status (AIS real-time data of Shipxy), and customs policies (ASEAN country gazette machine translation system). The acquisition method can be through multiple channels such as the Internet, social media, news reports, and financial markets. The means of data acquisition can be API interfaces, web scraping technology, data information platforms, etc. By collecting this information in a highly automated manner, enterprises can obtain a comprehensive market view.

[0022] As described in step S2 above, detect whether the first real-time data information contains risk data information. After obtaining the first real-time data information, conduct in-depth analysis of these data to detect whether they contain risk data information. Specifically, machine learning or rule-based methods can be used to search for potential risk signals in the data, such as abnormal price fluctuations, logistics delays, and public notice policy changes. By analyzing data patterns and regularities, abnormal situations that may cause business interruptions or economic losses can be identified. In a preferred embodiment, it can be further divided into data preprocessing, feature extraction, and risk signal identification. For example, using time series analysis, data trends and periodic fluctuations can be identified to determine whether there are abnormalities.

[0023] As described in step S3 above, when the first real-time data information contains risk data information, input the risk data information into the target simulation engine to obtain a target solution. When it is confirmed that the first real-time data information contains risk data information, input these risk data into the target simulation engine. The design of the target simulation engine needs to be based on rich data input and accurate algorithms. By simulating different business scenarios, evaluate the possible impact of risks on enterprise operations. The simulation engine can adopt various physical models, economic models, or behavioral models, etc., to capture potential risk factors from a multi-dimensional perspective. After a series of simulation calculations, the simulation engine will generate solutions for specific risks. In a specific embodiment, the target simulation engine includes the following data hierarchical architecture: Data Layer: Integrate 12 types of global risk signal sources (such as the GDELT geopolitical event database, the AIS real-time logistics data of Shipxy, and the ASEAN customs policy machine translation system) to provide real-time data input for the engine.

[0024] Simulation Layer: The engine dynamically constructs 5-level crisis scenarios (such as port congestion, tariff adjustment, and full-link interruption) based on the large language model (LLM), and uses Monte Carlo simulation combined with a rule engine to generate three-dimensional response plans.

[0025] Optimization Layer: Transmit the plans output by the engine (such as alternative route planning and bonded warehouse activation strategies) to the multi-objective optimization algorithm module for secondary optimization calculation.

[0026] Application Layer: The supplier intelligent selection system directly calls the compliance policies generated by the engine (such as the certificate of origin reconstructed according to the CPTPP rules).

[0027] The underlying operating mechanism of the engine is as follows: Data Access: Access external data through a standardized API gateway, and embed a knowledge graph to structurally process multi-source heterogeneous data (such as policy texts and logistics status).

[0028] Scenario Construction: Utilize the sequence generation ability of the LLM to transform discrete events into a crisis scenario tree with weights, supporting dynamic probability assignment (such as the triggering probability of a port strike event being 35%).

[0029] Policy Deduction: Test the feasibility of the plan in a virtual environment. For example, simulate the time limit and cost changes of enabling the China-Europe Land-Sea Express Line when the Suez Canal is blocked.

[0030] Output Integration: The three-dimensional plan is output in a standardized JSON format for direct parsing and calling by the upper-layer module to ensure the system's collaborative efficiency.

[0031] As described in step S4 above, address the risks brought by the risk data information based on the solution. Implement the solution generated by the target simulation engine to deal with potential problems caused by the risk data information. Specifically, after obtaining the solution, it may involve short-term emergency measures such as resource reconfiguration, supplier adjustment, and transportation route change, as well as long-term strategic adjustments to enhance the enterprise's risk resistance ability. In this link, the enterprise may need to form a cross-departmental team to ensure the effective transmission and execution of information, specifically referring to the specific solution.

[0032] As described in the above steps S201 - S202, during the training of the target simulation engine, multiple second real - time data information are obtained. The sources of this information can be very diverse, covering data from different fields and levels. For example, enterprises need to pay attention to market dynamics, industry trends, policy changes, competitor behavior, and customer feedback, etc. In a specific embodiment, the second real - time data information is also geopolitics (GDELT event database), logistics status (VesselsValue AIS real - time data), and customs policies (ASEAN country gazette machine translation system). Different from the first real - time data, the second real - time data information is not only information related to the specified cross - border trade, but can be information related to the entire enterprise. The acquisition method can be through multiple channels such as the Internet, social media, news reports, and financial markets. The means of data acquisition can be API interfaces, web crawler technology, data information platforms, etc. By collecting this information in real - time through highly automated means, enterprises can obtain a comprehensive market view.

[0033] Input the multiple second real - time data information obtained into the preset simulation engine. The key to this process lies in how to effectively integrate diverse data into the simulation engine. The preset simulation engine is usually built based on historical models and already has certain analysis and decision - making capabilities. However, after inputting real - time data information, the simulation engine can perform dynamic adjustment and optimization. That is, integrate each data in the data layer of the simulation engine to generate the target simulation engine. After the target simulation engine is constructed, use the preset large language model (LLM) to construct multiple crisis scenarios, and based on these scenarios, generate corresponding solutions in combination with preset rules. The core of this process lies in how to effectively utilize the powerful natural language processing ability of the LLM to transform complex risk factors and events into specific crisis scenarios. For example, the LLM can identify potential risk events according to the input real - time data information, such as transportation delays, market demand fluctuations, policy changes, etc. Specifically, 5 - level crisis scenarios can be generated. For example, the first level is local port congestion (simulation of port strike events), the third level is regional tariff adjustment (simulation of sudden changes in tax rates of RCEP member states), and the fifth level is full - link interruption (re - creating canal blockage events). Semantic analysis is performed on these events to generate corresponding crisis scenarios. These crisis scenarios can be synthesized into a weighted scenario tree, reflecting the relationships and dynamic interactions between different events. And combining preset rules means that the processing of these scenarios not only depends on data analysis but also follows industry norms, enterprise policies, and best practices. This combination makes the generated solutions more practical and adaptable, providing targeted coping strategies for enterprises to ensure that corresponding countermeasures can be quickly and effectively formulated in the face of emergencies. Finally, the completion of this process will enable the target simulation engine to have the ability to respond to different crisis situations, thereby improving the risk management level of enterprises in cross - border trade.

[0034] In one embodiment, step S3 of inputting the risk data information into the target simulation engine to obtain a solution when the first real-time data information contains risk data information includes: S301: Extract the risk features of the risk data information according to a preset semantic parsing model; S302: Perform similarity matching between the risk features and the pre-stored features of each crisis scenario; S303: Extract the crisis scenario with the maximum similarity as the target crisis scenario, and determine whether the corresponding similarity value is greater than a preset similarity value; S304: If the corresponding similarity value is greater than the preset similarity value, select the solution corresponding to the target crisis scenario as the target solution.

[0035] As described in steps S301 - S304 above, it is first necessary to extract key risk features from the input risk data information using a preset semantic parsing model. This process involves natural language processing of the original data to identify important information segments related to risks. The semantic parsing model can analyze the structure and meaning of the text, and by identifying keywords, phrases, and context information contained in sentences, convert complex data into structured features that can be used for further analysis. For example, when analyzing a news article about supply chain delays, the model may extract key information such as "transportation delay", "weather factors", "supplier", etc. These extracted risk features can be quantified and standardized to make them comparable, laying the foundation for risk matching in subsequent steps. Effective risk feature extraction helps ensure that in the subsequent crisis scenario matching process, the actual risk types faced can be more accurately corresponded, thereby improving the accuracy and timeliness of overall decision-making.

[0036] For the extracted risk features, it is necessary to perform similarity matching with the pre-stored crisis scenario features. The core of this process is to establish a similarity calculation model. Commonly used algorithms may include cosine similarity, Euclidean distance, or other machine learning models to quantify the similarity between different risk features. This ensures that newly input risk information can effectively identify the most relevant crisis scenario through comparison. During the processing, the feature data of the pre-stored crisis scenarios should cover key elements of various scenarios, such as the impact degree, historical occurrence frequency, and specific consequences caused. Through such matching, the system can determine which crisis scenarios may have a direct impact on the current risk, and thus make a more targeted and effective response in the subsequent process. The effectiveness of similarity matching is crucial for subsequent decision support because it will lay the foundation for the selection of corresponding solutions.

[0037] From the results of matching risk characteristics with the characteristics of each crisis scenario, extract the crisis scenario with the highest similarity as the target crisis scenario. At this time, the system needs to first record the similarity value of each match and find the maximum value from them. If this maximum similarity value exceeds the preset threshold, it indicates that there is a significant association between the current risk characteristics and the target crisis scenario. The preset similarity value is usually adjusted according to historical data and the characteristics of the business scenario to ensure that potential impact crisis scenarios can be identified in a timely and accurate manner. For example, it can be set to 85%. If the similarity value corresponding to the target crisis scenario does not reach the preset standard, it may indicate that the current risk characteristics fail to effectively match any known crisis scenario, thus further analysis and countermeasures need to be taken. If it is detected that the similarity value of the target crisis scenario is greater than the preset similarity value, the system will select the solution corresponding to the target crisis scenario as the final target solution. This process is a key link in providing specific solution strategies for risk management. If the target crisis scenario is properly selected and has a high similarity, it can effectively guide the enterprise to take corresponding countermeasures for this scenario. For example, if the target crisis scenario is "port congestion", the corresponding solutions may include adjusting the transportation route and increasing logistics resources. This selection process requires the system to ensure the accuracy and timeliness of the decision-making to ensure that the formulated solution can effectively solve the current risk problem. At the same time, if the selection of the target crisis scenario and its corresponding solution is reasonable, it helps the enterprise to improve its emergency response ability in actual operation, reduce potential economic losses, and provide guarantee for the continuous operation of the enterprise.

[0038] In one embodiment, after step S303 of extracting the crisis scenario with the highest similarity as the target crisis scenario and determining whether the corresponding similarity value is greater than the preset similarity value, it further includes: S3041: If the corresponding similarity value is less than or equal to the preset similarity value, input the risk data information into the target simulation engine to construct a corresponding target crisis scenario based on the preset large language model; S3042: Generate the target solution for the target crisis scenario in combination with the preset rules.

[0039] In one embodiment, if the similarity value of the derived target crisis scenario is less than or equal to the preset similarity value, it means that the relevance between the current risk characteristics and the stored crisis scenarios is insufficient. In this case, the system needs to take further measures to address potential risks. Therefore, the risk data information will be directly input into the target simulation engine to reconstruct a corresponding target crisis scenario using the preset large language model (LLM) therein. Based on the newly constructed target crisis scenario, the system will generate corresponding target solutions in combination with preset rules. The preset rules may cover various strategies, including response time, resource allocation, legal compliance, etc., which are rules preset for the engine. During this process, through in-depth analysis of the target crisis scenario, the system can generate specific countermeasures. For example, if the new target crisis scenario is "uncertainty of new trade policies", the corresponding solutions may involve re-evaluating suppliers, adjusting prices, or enhancing customer communication, etc. This process is usually achieved through a rule engine to ensure that the generated solutions are both operable and in line with the company's strategic goals and market demands.

[0040] In one embodiment, after step S203 of constructing multiple crisis scenarios based on a preset large language model in the target simulation engine and generating solutions for each crisis scenario in combination with preset rules, the following steps are further included: S2041: Obtain corresponding third real-time data based on the solution. S2042: Use the improved non-dominated sorting genetic algorithm to generate multiple solution strategies for each crisis scenario based on the third real-time data. S2043: Obtain the dimension values corresponding to multiple preset dimensions of each solution strategy, and perform weighted summation of the respective dimension values to obtain the index values corresponding to each solution strategy. S2044: Select the solution strategy with the smallest index value as the target solution corresponding to the crisis scenario.

[0041] As described in the above steps S2041 - S2044, first, obtain the corresponding third real - time data according to the generated solution. The third real - time data may include information directly related to the solution, such as market reaction, industry trends, competitors' behaviors, customer feedback, operating costs, real - time capacity data of each logistics node (such as Zhengzhou International Land Port), etc. These data can be obtained through multiple channels, including third - party data service providers, industry reports, social media, market analysis tools, etc. The acquisition methods can be through multiple channels such as the Internet, social media, news reports, financial markets, etc. The means of data acquisition can be API interfaces, web crawler technology, data information platforms, etc. By collecting this information in real - time through highly automated means, enterprises can obtain a comprehensive market view. Based on the third real - time data, an improved Non - dominated Sorting Genetic Algorithm II (NSGA - II) is used to generate multiple solution strategies for each crisis scenario. NSGA - II is a multi - objective optimization algorithm suitable for solving complex decision - making problems, especially when multiple objectives need to be balanced. By introducing an improvement mechanism, the convergence and diversity of the algorithm can be enhanced, making it more effective in the strategy generation process. Specifically, the system analyzes the current third real - time data and transforms the solution into a series of optimization problems. For example, the algorithm can consider multiple objectives such as cost, efficiency, compliance, etc. simultaneously and balance these objectives to generate a set of feasible solution strategies. These strategies will reflect the best responses to the corresponding crisis scenarios under different environments and constraints, ensuring that enterprises can take flexible and effective actions in different situations. This process not only enhances the scientific nature of decision - making but also improves the ability to handle complex situations, providing enterprises with diverse choices. Evaluate each generated solution strategy. This process first involves obtaining the performance of each strategy in multiple preset dimensions, such as cost, time efficiency, risk management performance, and compliance, etc. The values of each dimension can be quantified through simulation, expert evaluation, or historical data. Next, these dimension values are weighted and summed to obtain the comprehensive index value of each solution strategy. The weighting process here is very important because different dimensions may have different importance in decision - making. The preset weights can be set based on past experience, industry standards, or the enterprise's own strategic goals. Through the obtained index values, the system can quantitatively compare various solution strategies, providing a clear basis for decision - makers to more intuitively understand the advantages and disadvantages of each strategy. This evaluation process not only improves the information transparency of decision - making but also provides an important reference for the subsequent selection of target solutions, ensuring that enterprises can choose the most feasible strategy in the process of coping with risks, and select the solution strategy with the smallest index value as the target solution corresponding to the crisis scenario.

[0042] In one embodiment, after step S2044 of selecting the solution strategy with the smallest index value as the solution corresponding to the crisis scenario, the following steps are further included: S2045: Input the target solution and the third real-time data into the target simulation engine to adjust the parameters of the target simulation engine, so as to obtain a simulation engine with adjusted parameters.

[0043] As described in step S2045 above, input the target solution and the third real-time data into the target simulation engine to adjust the parameters of the target simulation engine, so as to obtain a simulation engine with adjusted parameters. The purpose is to optimize and adjust the parameters of the simulation engine by inputting the selected solution and relevant real-time data. The simulation technology is used to simulate various possible situations and results to help decision-makers understand the potential impact of a specific solution strategy. In this process, the target solution provides a theoretical and strategic framework, while the third real-time data reflects the actual operating environment, including current social and economic conditions, environmental changes, technological progress, etc. By combining these two, the simulation engine can more accurately reflect the real situation and make corresponding parameter adjustments. By continuously adjusting the parameters, the simulation engine can generate more accurate prediction results to ensure that the solution can be effectively implemented in practice.

[0044] In one embodiment, step 2042 of generating multiple solution strategies for each crisis scenario by using the improved non-dominated sorting genetic algorithm based on the third real-time data further includes: S20421: Obtain manually set parameters; S20422: Based on the manually set parameters and the third real-time data, use the improved non-dominated sorting genetic algorithm to generate multiple solution strategies for each crisis scenario.

[0045] As described in the above steps S20421 - S20422, a series of manually set parameters need to be obtained from decision - makers or experts in relevant fields. These parameters are usually the key inputs required for the specific needs, constraints, and strategies of the current crisis scenario. Manually set parameters can include basic configurations such as population size, crossover rate, mutation rate, selection method, etc. in the genetic algorithm. These parameters have a direct impact on the performance of the genetic algorithm. For example, the population size determines the diversity of solutions, and the crossover and mutation rates affect the evolution rate and search ability of the population. The process of obtaining these parameters generally needs to combine past experience and data analysis to ensure the rationality and adaptability of the set values. In addition, the manually set parameters can be combined with real - time data to ensure flexible deployment in a dynamic environment, thus helping to generate effective and practical solution strategies. This process provides the necessary basic support for subsequent strategy generation, enabling the genetic algorithm to operate effectively in complex situations. Using the obtained manually set parameters and the third real - time data, an improved non - dominated sorting genetic algorithm is used to generate multiple solution strategies for each crisis scenario. The improved non - dominated sorting genetic algorithm is an evolutionary algorithm suitable for multi - objective optimization problems and can effectively handle the optimization between multiple conflicting objectives. At the same time, this algorithm pays special attention to maintaining the diversity of solutions to avoid premature convergence. By using the manually set parameters as algorithm configurations, the operating characteristics of the algorithm can be flexibly adjusted to better adapt to the existing problems. At this time, the third real - time data provides an accurate reflection of the current environmental conditions, such as relevant economic variables, resource status, social dynamics, etc., which are important bases for generating solution strategies. The process of realizing the generation usually includes multiple iterative steps, in which the advantages and disadvantages of candidate solutions are evaluated in each iteration, and a new generation of solutions is generated through operations such as selection, crossover, and mutation. In each iteration, the algorithm relies on real - time data and artificially set criteria to adjust the strategy, enabling it to show flexibility and adaptability in complex and dynamic environments.

[0046] In one embodiment, after step S203 of constructing multiple crisis scenarios based on a preset large - language model in the target simulation engine and generating solution strategies for each crisis scenario in combination with preset rules, the following steps are further included: S2141: Obtain the first predicted cost of the solution corresponding to the target simulation engine, send the solution to a specified terminal, and obtain the second predicted cost uploaded by the specified terminal; S2142: Calculate whether the first predicted cost exceeds a preset percentage of the second predicted cost; S2143: If the first predicted cost exceeds the preset percentage of the second predicted cost, adjust the cost constraint parameters in the target simulation engine and regenerate the solution.

[0047] As described in the above steps S2141 - S2143, the system first needs to extract the first predicted cost of the solution corresponding to the preset crisis scenario from the target simulation engine. This cost is usually based on the calculation model within the simulation engine, taking into account cost elements from various sources, including the consumption of human resources, materials, time, and other resources. Subsequently, the system sends this solution to a designated terminal, which can be a decision - maker of the team, relevant experts, or other collaborative personnel, etc. After receiving the solution, the designated terminal can upload a second predicted cost obtained through evaluation based on its specific implementation conditions and practical experience. This second predicted cost may involve factors such as the characteristics of on - site actual operations, fluctuations in material prices, and execution difficulties. Therefore, the comparison between the first predicted cost and the second predicted cost can provide important reference for subsequent decision - making. Through this process, the system can not only evaluate the rationality of the simulation results but also enable participants to incorporate actual data into the decision - making process to improve the practicality and reliability of the solution. Comparing the first predicted cost and the second predicted cost aims to evaluate the effectiveness and rationality of the latter. Specifically, the system will compare whether the first predicted cost exceeds a preset percentage of the second predicted cost. In the decision - making process, cost control is an important factor. If the first predicted cost far exceeds the set standard of the second predicted cost, it may indicate that the solution generated by the simulation engine is theoretically unrealistic or some practical factors have not been considered during the implementation process. The calculation process usually requires mathematical operations to determine the gap between the two and uses the preset percentage as the judgment criterion, such as the common tolerance ranges of "10%" or "20%". If it is found in the calculation that the first predicted cost indeed exceeds the allowed preset percentage, the system will adjust the cost constraint parameters. This adjustment is a necessary operation for the system to enhance the feasibility and rationality of the solution. The target simulation engine will re - evaluate the cost constraints and may adjust the corresponding parameters based on multiple factors such as market price changes, resource availability, and implementation difficulties. For example, the system may increase resource utilization, reduce certain non - critical costs, or adjust the strategy to control the overall budget if possible. After the adjustment is completed, the system will regenerate the solution based on the new cost constraint conditions. Through this process, the system can effectively optimize the original solution to make it more in line with the actual situation. This cycle ensures that the decision - making can quickly respond to changes in the dynamic environment while enhancing the practical availability of the solution. Finally, the adjusted and regenerated solution will be more realistic, providing a more reliable basis for decision - makers to support strategic deployments in crisis management.

[0048] Referring to Figure 3 , the present invention also provides a device for generating a solution to cross - border trade risks, and the device includes: An acquisition module 902, configured to acquire a plurality of first real - time data information related to a specified cross - border trade; A detection module 904, configured to detect whether the first real-time data information contains risk data information; An input module 906, configured to, when the first real-time data information contains risk data information, input the risk data information into a target simulation engine to obtain a target solution; A solution module 908, configured to solve the risk brought by the risk data information based on the solution; Wherein, the training method of the target simulation engine includes: A second real-time data information acquisition sub-module, configured to acquire a plurality of second real-time data information; A second real-time data information input sub-module, configured to input the plurality of second real-time data information into a preset simulation engine to obtain a target simulation engine; A crisis scenario construction sub-module, configured to construct a plurality of crisis scenarios in the target simulation engine based on a preset large language model, and generate solutions for each crisis scenario in combination with preset rules.

[0049] In one embodiment, the input module 906 includes: A risk feature extraction sub-module, configured to extract risk features of the risk data information according to a preset semantic parsing model; A similarity matching sub-module, configured to perform similarity matching between the risk features and pre-stored features of each crisis scenario; A target crisis scenario marking sub-module, configured to extract the crisis scenario with the largest similarity as the target crisis scenario, and determine whether the corresponding similarity value is greater than a preset similarity value; A target solution sub-module, configured to, if the corresponding similarity value is greater than the preset similarity value, select the solution corresponding to the target crisis scenario as the target solution.

[0050] In one embodiment, the input module 906 further includes: A risk data information input sub-module, configured to, if the corresponding similarity value is less than or equal to the preset similarity value, input the risk data information into the target simulation engine to construct a corresponding target crisis scenario based on a preset large language model; A target solution generation sub-module, configured to generate a target solution for the target crisis scenario in combination with the preset rules.

[0051] In one embodiment, the solution generation device for cross-border trade risks further includes: A third real-time data acquisition module, configured to acquire corresponding third real-time data based on the solution; A solution strategy generation module, configured to generate multiple solution strategies for each of the crisis scenarios based on the third real-time data by using an improved non-dominated sorting genetic algorithm; A dimension value acquisition module, configured to acquire dimension values corresponding to multiple preset dimensions of each solution strategy, and perform weighted summation on the respective dimension values to obtain an index value corresponding to each solution strategy; A target solution marking module, configured to select the solution strategy with the minimum index value as the target solution corresponding to the crisis scenario.

[0052] In one embodiment, the solution generation device for cross-border trade risks further includes: A third real-time data input module, configured to input the target solution and the third real-time data into the target simulation engine to adjust parameters of the target simulation engine, so as to obtain an adjusted simulation engine.

[0053] In one embodiment, the solution strategy generation module includes: An artificially set parameter acquisition sub-module, configured to acquire artificially set parameters; A solution strategy generation sub-module, configured to generate multiple solution strategies for each of the crisis scenarios based on the artificially set parameters and the third real-time data by using an improved non-dominated sorting genetic algorithm.

[0054] In one embodiment, the solution generation device for cross-border trade risks further includes: A first predicted cost acquisition module, configured to acquire a first predicted cost corresponding to the solution of the target simulation engine, send the solution to a specified terminal, and acquire a second predicted cost uploaded by the specified terminal; A percentage judgment module, configured to calculate whether the first predicted cost exceeds a preset percentage of the second predicted cost; A cost constraint parameter adjustment module, configured to, if the first predicted cost exceeds the preset percentage of the second predicted cost, adjust cost constraint parameters in the target simulation engine and regenerate a solution.

[0055] Figure 4 Shows the internal structure diagram of a computer device in one embodiment. The computer device may specifically be a terminal or a server. As Figure 4As shown in the figure, the computer device includes a processor, a memory, and a network interface connected via a system bus. Among them, the memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium of the computer device stores an operating system and can also store a computer program. When the computer program is executed by the processor, the processor can implement a method for generating a solution to cross-border trade risks. The internal memory can also store a computer program. When the computer program is executed by the processor, the processor can execute a method for generating a solution to cross-border trade risks. Those skilled in the art can understand that Figure 4 The structure shown in the figure is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0056] In one embodiment, a computer device is proposed, including a memory and a processor. The memory stores a computer program. When the computer program is executed by the processor, the processor performs the following steps: Obtain a plurality of first real-time data information related to a specified cross-border trade; Detect whether the first real-time data information contains risk data information; When the first real-time data information contains risk data information, based on inputting the risk data information into a target simulation engine to obtain a target solution; Based on the solution, solve the risks brought by the risk data information; Among them, the training method of the target simulation engine includes: Obtain a plurality of second real-time data information; Input a plurality of the second real-time data information into a preset simulation engine to obtain a target simulation engine; Based on a preset large language model, construct a plurality of crisis scenarios in the target simulation engine, and generate solutions for each crisis scenario in combination with preset rules.

[0057] By introducing a target simulation engine and constructing a plurality of crisis scenarios based on a large language model, it is possible to more comprehensively understand and predict the impacts in different risk situations. The flexibility of this algorithm enables enterprises to timely generate optimized solutions for the dynamic international market environment and reduce losses caused by emergencies.

[0058] In one embodiment, a computer-readable storage medium is proposed, storing a computer program. When the computer program is executed by the processor, the processor performs the following steps: Obtain a plurality of first real-time data information related to a specified cross-border trade; Check whether the first real-time data information contains risk data information; When the first real-time data information contains risk data information, input the risk data information into the target simulation engine to obtain a target solution; Solve the risks brought by the risk data information based on the solution; Among them, the training method of the target simulation engine includes: Obtain multiple second real-time data information; Input multiple pieces of the second real-time data information into a preset simulation engine to obtain a target simulation engine; Build multiple crisis scenarios in the target simulation engine based on a preset large language model, and generate solutions for each crisis scenario in combination with preset rules.

[0059] By introducing the target simulation engine and building multiple crisis scenarios based on the large language model, it is possible to more comprehensively understand and predict the impacts in different risk situations. The flexibility of this algorithm enables enterprises to generate optimized solutions in a timely manner for the dynamic international market environment and reduce losses caused by emergencies.

[0060] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the various embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0061] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0062] The above-described embodiments merely represent several implementation manners of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the patent of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the appended claims.

Claims

1. A method for generating a solution to cross-border trade risks, characterized in that The method includes: Obtaining a plurality of first real-time data information related to specified cross-border trade; Detecting whether the first real-time data information contains risk data information; When the first real-time data information contains risk data information, inputting the risk data information into a target simulation engine to obtain a target solution; Solving the risks brought by the risk data information based on the solution; Wherein, the training method of the target simulation engine includes: Obtaining a plurality of second real-time data information; Inputting the plurality of second real-time data information into a preset simulation engine to obtain a target simulation engine; Constructing a plurality of crisis scenarios in the target simulation engine based on a preset large language model, and generating solutions for each crisis scenario in combination with preset rules.

2. The method for generating a solution to cross-border trade risks according to claim 1, wherein The step of, when the first real-time data information contains risk data information, inputting the risk data information into a target simulation engine to obtain a solution, includes: Extracting the risk characteristics of the risk data information according to a preset semantic parsing model; Performing similarity matching between the risk characteristics and the pre-stored characteristics of each crisis scenario; Extracting the crisis scenario with the largest similarity as the target crisis scenario, and determining whether the corresponding similarity value is greater than a preset similarity value; If the corresponding similarity value is greater than the preset similarity value, selecting the solution corresponding to the target crisis scenario as the target solution.

3. The method for generating a solution to cross-border trade risks according to claim 2, wherein After the step of extracting the crisis scenario with the largest similarity as the target crisis scenario and determining whether the corresponding similarity value is greater than a preset similarity value, it further includes: If the corresponding similarity value is less than or equal to the preset similarity value, inputting the risk data information into the target simulation engine to construct a corresponding target crisis scenario based on a preset large language model; Generating a target solution for the target crisis scenario in combination with the preset rules.

4. The method for generating a solution to cross-border trade risks according to claim 1, characterized in that, After the step of constructing a plurality of crisis scenarios in the target simulation engine based on a preset large language model and generating solutions for each crisis scenario in combination with preset rules, it further includes: Obtaining corresponding third real-time data based on the solution; Using an improved non-dominated sorting genetic algorithm to generate multiple solution strategies for each crisis scenario based on the third real-time data; Obtaining the dimension values corresponding to a plurality of preset dimensions of each solution strategy, and performing weighted summation of the respective dimension values to obtain the index values corresponding to each solution strategy; Selecting the solution strategy with the smallest index value as the target solution corresponding to the crisis scenario.

5. The method for generating a solution to cross-border trade risks according to claim 4, wherein After the step of selecting the solution strategy with the smallest index value as the solution to the crisis scenario, it further includes: Inputting the target solution and the third real-time data into the target simulation engine to adjust the parameters of the target simulation engine, thereby obtaining a simulation engine with adjusted parameters.

6. The method for generating a solution to cross-border trade risks according to claim 4, wherein The step of using an improved non-dominated sorting genetic algorithm to generate multiple solution strategies for each crisis scenario based on the third real-time data further includes: Obtaining manually set parameters; Based on the artificially set parameters and the third real-time data, an improved non-dominated sorting genetic algorithm is used to generate multiple solution strategies for each of the crisis scenarios.

7. The method for generating a solution to cross-border trade risks according to claim 1, characterized in that, After the step of constructing multiple crisis scenarios based on a preset large language model in the target simulation engine and generating solution plans for each crisis scenario in combination with preset rules, the method further includes: Obtaining a first predicted cost corresponding to the solution plan of the target simulation engine, sending the solution plan to a specified terminal, and obtaining a second predicted cost uploaded by the specified terminal; Calculating whether the first predicted cost exceeds a preset percentage of the second predicted cost; If the first predicted cost exceeds the preset percentage of the second predicted cost, adjusting the cost constraint parameters in the target simulation engine and regenerating the solution plan.

8. A solution generation device for cross-border trade risks, characterized in that, The device includes: An acquisition module, configured to acquire a plurality of first real-time data information related to a specified cross-border trade; A detection module, configured to detect whether the first real-time data information contains risk data information; An input module, configured to, when the first real-time data information contains risk data information, input the risk data information into a target simulation engine to obtain a target solution plan; A solution module, configured to solve the risks brought by the risk data information based on the solution plan; Wherein, the training method of the target simulation engine includes: A second real-time data information acquisition sub-module, configured to acquire a plurality of second real-time data information; A second real-time data information input sub-module, configured to input the plurality of second real-time data information into a preset simulation engine to obtain a target simulation engine; A crisis scenario construction sub-module, configured to construct a plurality of crisis scenarios in the target simulation engine based on a preset large language model and generate solution plans for each crisis scenario in combination with preset rules.

9. A computer-readable storage medium, characterized in that, A computer program is stored, and when the computer program is executed by a processor, the processor is caused to execute the steps of the method for generating a solution plan for cross-border trade risks according to any one of claims 1 to 7.

10. A computer device, characterized in that, The device includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor is caused to execute the steps of the method for generating a solution plan for cross-border trade risks according to any one of claims 1 to 7.

Images