Digital twinborn-based cross-border trade supply chain risk early warning system

By building a cross-border trade supply chain risk early warning system using digital twin technology, integrating multi-source data collection and deep learning models, and dynamically adjusting early warning thresholds, the system addresses the shortcomings of traditional early warning methods, achieving accurate identification and real-time early warning of cross-border trade supply chain risks, and supporting enterprise decision-making.

CN121882695APending Publication Date: 2026-04-17赵茗庄
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202511996865.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-12-27
Publication Date
2026-04-17

AI Technical Summary

Technical Problem

Existing methods for risk warning in cross-border trade supply chains rely on manual statistics and experience-based judgment. Incomplete data collection, delayed risk identification, and static assessment models result in low accuracy of warnings, making it difficult to adapt to the dynamic changes in cross-border trade supply chains and hindering enterprises from responding to risks in a timely manner.

Method used

A cross-border trade supply chain risk early warning system based on digital twins is adopted, which integrates IoT sensors, government data interfaces, logistics information interfaces, financial data interfaces and market data crawler modules to build a multi-source data acquisition layer. Combined with digital twin modeling, data processing and risk assessment modules, the system achieves accurate identification and dynamic early warning of risk factors through an improved CNN-LSTM hybrid model and reinforcement learning algorithm.

Benefits of technology

It enables full-process early warning of risks in cross-border trade supply chains, improves the coverage of risk identification and the accuracy of early warning, adapts to dynamic changes in the supply chain, reduces false alarm and missed alarm rates, provides practical decision support, and reduces corporate losses.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121882695A_ABST
    Figure CN121882695A_ABST
Patent Text Reader

Abstract

The invention provides a cross-border trade supply chain risk early warning system based on digital twinning, and relates to the technical field of cross-border trade. The cross-border trade supply chain risk early warning system based on digital twinning comprises an entity layer, a data acquisition layer, a digital twinning modeling layer, a data processing layer, a risk early warning layer, a safety protection layer and an application output layer, and the entity layer is a monitoring object basis of the system and covers all key physical objects in a cross-border trade supply chain; comprising suppliers, logistics carriers, storage facilities, customs nodes and purchasers in the cross-border trade supply chain. The system integrates a multi-source data acquisition module, covers full-dimensional data of policies, logistics, finance, markets, Internet of Things perception and the like, ensures the comprehensiveness of risk factor extraction, improves the coverage of risk identification, greatly improves the accuracy and real-time performance of early warning, and solves the problem that a traditional early warning mode is inaccurate and lagged in evaluation.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of cross-border trade technology, specifically a cross-border trade supply chain risk early warning system based on digital twins. Background Technology

[0002] Cross-border trade supply chains involve multiple entities, including suppliers, logistics providers, customs, and buyers from various countries and regions. The business processes are complex and influenced by a variety of factors such as policy, logistics, finance, and market conditions, resulting in numerous risk points and high uncertainty. Traditional cross-border trade supply chain risk early warning methods mainly rely on manual statistics and experience-based judgment, which suffers from incomplete data collection, delayed risk identification, static assessment models, and low early warning accuracy. These methods are ill-suited to the dynamic and complex scenarios of cross-border trade supply chains, leading to enterprises' inability to respond to risks in a timely manner and causing significant economic losses. Digital twin technology can construct a digital mirror of the physical world, enabling full lifecycle management and real-time monitoring of physical objects, providing a new technological path for supply chain risk early warning. However, existing digital twin-based supply chain management systems primarily focus on logistics tracking and visualization, lacking in-depth analysis of multi-source risk factors and dynamic early warning capabilities, thus failing to fully leverage the simulation, prediction, and early warning value of digital twins.

[0003] Therefore, this invention proposes a cross-border trade supply chain risk early warning system based on digital twins to achieve accurate risk identification, dynamic assessment and early warning. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a cross-border trade supply chain risk early warning system based on digital twins, including an entity layer, a data acquisition layer, a digital twin modeling layer, a data processing layer, a risk early warning layer, an application output layer, and a security protection layer. Each layer works collaboratively to achieve full-process early warning of risks in the cross-border trade supply chain.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A cross-border trade supply chain risk early warning system based on digital twins includes an entity layer, a data acquisition layer, a digital twin modeling layer, a data processing layer, a risk early warning layer, a security protection layer, and an application output layer. Its features include: the entity layer forming the foundation of the system's monitoring objects, encompassing all key physical objects in the cross-border trade supply chain, including suppliers, logistics carriers, warehousing facilities, customs nodes, and buyers; the data acquisition layer collecting multi-source operational data from the entity layer supply chain; the digital twin modeling layer constructing digital twins of physical objects and an overall digital twin model of the supply chain based on the collected data; the data processing layer preprocessing and feature extraction of the collected data; the risk early warning layer identifying and issuing early warnings based on the digital twin model and the processed data; and the application output layer outputting early warning information and decision-making suggestions. The data acquisition layer includes multiple types of data acquisition modules, including an IoT sensor module, a government data interface module, a logistics information interface module, a financial data interface module, and a market data crawler module. The digital twin modeling layer includes an object modeling module, a relationship modeling module, and a dynamic update module; The data processing layer includes a data preprocessing module and a risk factor extraction module; The risk warning layer includes a risk assessment module, a warning threshold dynamic adjustment module, and a warning generation module; The security protection layer includes a data encryption module, an access control module, and an abnormal behavior monitoring module; Through the above technical solutions, the entity layer is the foundation of the system's monitoring objects, covering all key physical objects in the cross-border trade supply chain; the data acquisition layer is the core of data acquisition, achieving full-dimensional data coverage through multiple types of data acquisition modules; the digital twin modeling layer is the core technical support of the system, constructing a digital twin sub-entity for each physical object through the object modeling module, which not only includes a three-dimensional visualization model of the geometric shape, but also integrates physical attributes and operating parameters; the data processing layer is responsible for data cleaning and feature extraction; the risk warning layer is the core of the system's early warning; the application output layer is responsible for displaying early warning information and providing decision support; and the security protection layer ensures the system's data security and operational security.

[0006] Furthermore, the IoT sensor module of the data acquisition layer is used to collect real-time operational data such as the location, temperature, humidity, vibration, and load of the logistics carrier; the government data interface module is used to obtain policy data of various countries in cross-border trade; the logistics information interface module is used to collect data on transportation node status, transportation timeliness, and logistics costs; the financial data interface module is used to obtain data on exchange rates, interest rates, and payment settlement status; and the market data crawler module is used to collect data on supply and demand, price fluctuations, and competitor dynamics. Through the above technical solutions, the IoT sensor module is deployed on physical objects such as logistics carriers and warehousing facilities to collect real-time sensing data such as location, temperature, and humidity. The government data interface module connects with the official systems of customs and trade management departments of various countries to obtain policy data. The logistics information interface module connects with the information systems of mainstream logistics companies to collect logistics data such as transportation and customs clearance. The financial data interface module connects with the systems of banks and payment institutions to obtain data such as exchange rates and payment status. The market data crawler module uses web crawling technology to collect market data from e-commerce platforms and industry databases to ensure the comprehensiveness and real-time nature of the data sources.

[0007] Furthermore, the object modeling module of the digital twin modeling layer adopts parametric modeling and 3D visualization technology to construct a digital twin sub-body containing geometric features, physical attributes, and operating parameters for each physical object. The relationship modeling module constructs the logistics connection relationship, information flow interaction relationship, and capital flow association relationship between the digital twin sub-bodies based on the business process of the cross-border trade supply chain. The dynamic update module dynamically adjusts the parameters of the digital twin model according to the changes in the operating status of the physical object through real-time data acquisition and edge computing technology to ensure the synchronization between the digital twin model and the physical world. Through the above technical solutions, the relationship modeling module constructs the logistics, information flow, and capital flow relationships between sub-entities based on the supply chain business process, the supply relationship between suppliers and warehousing facilities, and the customs clearance process relationship between warehousing facilities and customs nodes. The dynamic update module reduces data transmission latency through edge computing technology, ensuring that the digital twin model and the physical object's operating status are synchronized in real time, providing an accurate digital foundation for risk warning.

[0008] Furthermore, the data preprocessing module of the data processing layer removes invalid data based on the box plot and isolated forest fusion algorithm, and standardizes and unifies the data format through Z-Score. The risk factor extraction module uses an improved CNN-LSTM hybrid model based on deep learning to extract policy risk factors, logistics risk factors, financial risk factors, supply and demand risk factors, and compliance risk factors from the preprocessed data. Through the above technical solution, the box plot and isolated forest fusion algorithm initially screens out obvious outliers, marking data that exceeds the range as preliminary outliers. Then, the pre-processed data is input into the isolated forest algorithm, which isolates the data by constructing multiple decision trees. Outliers, due to their unique features, are more easily isolated in shallower tree layers. By calculating the outlier score of the data, the outliers are finally determined and removed. The CNN network consists of convolutional layers and pooling layers. The convolutional layers use 3×3 convolutional kernels and extract spatial features from the data through a sliding window. The LSTM network consists of input gate, forget gate, and output gate, which can capture the long and short-term time series features of the data. An attention mechanism layer is added between the CNN and LSTM. By calculating the attention weights of different features, the contribution of important features to risk factor extraction is highlighted, improving the accuracy of feature extraction.

[0009] Furthermore, the risk assessment module of the risk warning layer uses the AHP (Analytic Hierarchy Process) to determine the weight of each risk factor, and combines the fuzzy comprehensive evaluation method and the Markov chain prediction model to quantitatively assess the supply chain risk level for the current and future period. The risk level is divided into low risk, medium risk, and high risk. The warning threshold dynamic adjustment module dynamically adjusts the warning threshold for different risk types based on historical risk data and supply chain operation status using the Q-Learning reinforcement learning algorithm. The warning generation module generates corresponding warning information based on the comparison between the risk assessment results and the warning threshold, including the warning level, risk type, scope of impact, probability of occurrence, duration, and response suggestions. Through the above technical solutions, the AHP method constructs a hierarchical model and calculates the weights of each indicator after consistency testing. The fuzzy comprehensive evaluation method transforms the quantitative values ​​of each risk factor into a fuzzy evaluation matrix. Combined with the weights determined by the AHP method, a fuzzy comprehensive evaluation vector is calculated, thereby obtaining a quantitative score for the risk level. The Markov chain prediction model treats the risk level as a state space and calculates a state transition probability matrix based on historical risk level data. This matrix is ​​used to predict the risk level state over a period of time, achieving risk warning. The Q-Learning reinforcement learning algorithm treats the adjustment of the warning threshold as a decision-making process of reinforcement learning. The state space is the supply chain operation status and historical warning accuracy, and the action space is the adjustment range of the warning threshold. Through interaction with the environment, it continuously updates and learns the optimal threshold adjustment strategy, achieving dynamic optimization of the warning threshold.

[0010] A method for risk early warning of cross-border trade supply chains based on digital twins, characterized in that: the method is implemented through the cross-border trade supply chain risk early warning system based on digital twins as described in claims 1-5, and the method steps include: Step 1: Data Acquisition; Collect multi-source operational data of physical entities in the cross-border trade supply chain through multiple data acquisition modules, including policy data, logistics data, financial data, market data, and IoT sensing data; Step 2: Digital Twin Modeling; Construct digital twin sub-entities of physical objects and an overall digital twin model of the supply chain, establish business relationships between digital twin sub-entities, and dynamically update the digital twin model based on real-time data; Step 3: Data processing; The collected data undergoes preprocessing operations such as outlier removal, normalization, and missing value imputation, and then various risk factors are extracted using a deep learning model; Step 4: Risk assessment; The AHP method is used to determine the weights of risk factors, and the risk level is quantitatively assessed by combining the fuzzy comprehensive evaluation method and the Markov chain prediction model. Step 5: Warning generation; Based on the dynamically adjusted warning threshold, generate the corresponding warning information and output it to the application output layer.

[0011] Furthermore, the dynamic update process of the digital twin model in step 2 includes receiving incremental data from the data acquisition layer in real time, processing the incremental data quickly through edge computing nodes, extracting the state change characteristics of the physical object, and triggering parameter updates of the digital twin model if the state change exceeds a preset threshold. At the same time, the model update log is recorded for traceability and verification.

[0012] Furthermore, the training process of the Markov chain prediction model in step 4 includes using historical risk data as the training set, using the risk level as the state space, calculating the state transition probability matrix, and optimizing the model parameters through the maximum likelihood estimation method to maintain a high prediction accuracy of the model's risk level.

[0013] This invention provides a cross-border trade supply chain risk early warning system based on digital twins. It has the following beneficial effects: 1. This invention provides a cross-border trade supply chain risk early warning system based on digital twins. It integrates multi-source data acquisition modules, covering data from all dimensions such as policy, logistics, finance, market, and IoT sensing, ensuring the comprehensiveness of risk factor extraction and improving the coverage of risk identification. Furthermore, it adopts an improved CNN-LSTM hybrid model to extract risk factors, achieving accurate risk identification and forward-looking early warning, which greatly improves the accuracy and real-time performance of early warning and solves the problems of inaccurate and delayed assessment in traditional early warning methods.

[0014] 2. This invention provides a cross-border trade supply chain risk early warning system based on digital twins. It uses reinforcement learning algorithms to dynamically adjust the early warning threshold, adapting to the dynamic changes in the cross-border trade supply chain, improving the flexibility and accuracy of early warnings, reducing false alarm rates and missed alarm rates, and integrating a decision support module to generate targeted supply chain optimization solutions based on risk early warning results. This provides enterprises with practical decision support, helps them reduce risk losses, ensures the stable operation of the supply chain, supports data interoperability with existing enterprise management systems, has good compatibility and scalability, and reduces deployment costs and usage barriers for enterprises. Attached Figure Description

[0015] Figure 1 This is a system architecture diagram of the cross-border trade supply chain risk early warning system based on digital twins according to the present invention; Figure 2 This is a diagram illustrating the system operation steps of the cross-border trade supply chain risk early warning system based on digital twins according to the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Example 1

[0017] like Figure 1 As shown in the figure, this embodiment of the invention provides a cross-border trade supply chain risk early warning system based on digital twins, including an entity layer, a data acquisition layer, a digital twin modeling layer, a data processing layer, a risk early warning layer, a security protection layer, and an application output layer. The entity layer is the foundation of the system's monitoring objects, covering all key physical objects in the cross-border trade supply chain, including suppliers, logistics carriers, warehousing facilities, customs nodes, and purchasers. The data acquisition layer is used to collect multi-source operational data from the supply chain in the entity layer. The digital twin modeling layer constructs digital twins of physical objects and an overall digital twin model of the supply chain based on the collected data. The data processing layer preprocesses and extracts features from the collected data. The risk early warning layer realizes risk identification and early warning based on the digital twin model and the processed data. The application output layer is used to output early warning information and decision suggestions. The data acquisition layer includes multiple types of data acquisition modules, including an IoT sensor module, a government data interface module, a logistics information interface module, a financial data interface module, and a market data crawler module. The digital twin modeling layer includes an object modeling module, a relationship modeling module, and a dynamic update module; The data processing layer includes a data preprocessing module and a risk factor extraction module; The risk warning layer includes a risk assessment module, a warning threshold dynamic adjustment module, and a warning generation module; The security protection layer includes a data encryption module, an access control module, and an abnormal behavior monitoring module; The entity layer is the foundation of the system's monitoring objects, covering all key physical objects in the cross-border trade supply chain. The data acquisition layer is the core of data acquisition, achieving full-dimensional data coverage through multiple types of data acquisition modules. The digital twin modeling layer is the core technical support of the system, constructing a digital twin for each physical object through the object modeling module. This includes not only a three-dimensional visualization model of the geometric shape but also integrates physical attributes and operating parameters. The data processing layer is responsible for data cleaning and feature extraction. The risk warning layer is the core of the system's early warning system. The application output layer is responsible for displaying early warning information and providing decision support. The security protection layer ensures the system's data security and operational security. The IoT sensor module in the data acquisition layer is used to collect real-time operational data such as the location, temperature, humidity, vibration, and load of the logistics carrier. The government data interface module is used to obtain policy data from various countries in cross-border trade. The logistics information interface module is used to collect data on transportation node status, transportation timeliness, and logistics costs. The financial data interface module is used to obtain data on exchange rates, interest rates, and payment settlement status. The market data crawler module is used to collect data on supply and demand, price fluctuations, and competitor dynamics. The IoT sensor module is deployed on physical objects such as logistics carriers and warehousing facilities to collect real-time sensing data such as location and temperature. The government data interface module connects with the official systems of customs and trade management departments of various countries to obtain policy data. The logistics information interface module connects with the information systems of mainstream logistics companies to collect logistics data such as transportation and customs clearance. The financial data interface module connects with the systems of banks and payment institutions to obtain data such as exchange rates and payment status. The market data crawler module uses web crawling technology to collect market data from e-commerce platforms and industry databases to ensure the comprehensiveness and real-time nature of the data sources. The object modeling module of the digital twin modeling layer uses parametric modeling and 3D visualization technology to construct a digital twin sub-entity for each physical object, including geometric features, physical attributes, and operating parameters. The relationship modeling module, based on the business process of the cross-border trade supply chain, constructs the logistics connection relationship, information flow interaction relationship, and capital flow association relationship between the digital twin sub-entities. The dynamic update module, through real-time data acquisition and edge computing technology, dynamically adjusts the parameters of the digital twin model according to the changes in the operating status of the physical object, ensuring the synchronization of the digital twin model with the physical world. The relationship modeling module, based on the supply chain business process, constructs the logistics, information flow, and capital flow association between the sub-entities, the supply relationship between suppliers and warehousing facilities, and the customs clearance process relationship between warehousing facilities and customs nodes. The dynamic update module, through edge computing technology, reduces data transmission latency, ensuring that the operating status of the digital twin model and the physical object is synchronized in real time, providing an accurate digital foundation for risk warning. The data preprocessing module of the data processing layer removes invalid data based on a box plot and isolated forest fusion algorithm and standardizes the data format using Z-Score. The risk factor extraction module uses an improved CNN-LSTM hybrid model based on deep learning to extract policy risk factors, logistics risk factors, financial risk factors, supply and demand risk factors, and compliance risk factors from the preprocessed data. The box plot and isolated forest fusion algorithm initially screens out obvious outliers, marking data outside the range as preliminary outliers. Then, the preprocessed data is input into the isolated forest algorithm to construct multiple decision trees. Data isolation is performed, as outliers, due to their unique features, are more easily isolated in shallower tree layers. By calculating the outlier score, outliers are finally identified and removed. The CNN network consists of convolutional layers and pooling layers, with 3×3 convolutional kernels used in the convolutional layers. Spatial features in the data are extracted through a sliding window. The LSTM network consists of an input gate, a forget gate, and an output gate, which can capture the long and short-term time series features of the data. An attention mechanism layer is added between the CNN and LSTM. By calculating the attention weights of different features, the contribution of important features to risk factor extraction is highlighted, thereby improving the accuracy of feature extraction. The risk assessment module of the risk warning layer uses the Analytic Hierarchy Process (AHP) to determine the weights of each risk factor. Combined with fuzzy comprehensive evaluation and a Markov chain prediction model, it quantitatively assesses the supply chain risk level for the current and future periods, categorizing risks as low, medium, and high. The dynamic adjustment module for warning thresholds uses a Q-Learning reinforcement learning algorithm to dynamically adjust warning thresholds for different risk types based on historical risk data and supply chain operational status. The warning generation module generates corresponding warning information based on a comparison between the risk assessment results and the warning thresholds, including the warning level, risk type, scope of impact, probability of occurrence, duration, and response recommendations. The AHP method constructs a hierarchical model and uses consistency checks... After verification, the weights of each indicator are calculated. The fuzzy comprehensive evaluation method transforms the quantitative values ​​of each risk factor into a fuzzy evaluation matrix. Combined with the weights determined by the AHP method, the fuzzy comprehensive evaluation vector is calculated, thereby obtaining the quantitative score of the risk level. The Markov chain prediction model regards the risk level as a state space. Based on historical risk level data, the state transition probability matrix is ​​calculated. The risk level state over a period of time is predicted through this matrix to achieve risk warning. The Q-Learning reinforcement learning algorithm regards the adjustment of the warning threshold as a decision-making process of reinforcement learning. The state space is the supply chain operation status and historical warning accuracy, and the action space is the adjustment range of the warning threshold. Through interaction with the environment, the algorithm continuously updates and learns the optimal threshold adjustment strategy to achieve dynamic optimization of the warning threshold. Example 2

[0018] like Figure 2 As shown, this embodiment of the invention provides a method for risk early warning of cross-border trade supply chains based on digital twins. The method is implemented through the cross-border trade supply chain risk early warning system based on digital twins as described in claims 1-5, and the method steps include: Step 1: Data Acquisition; Collect multi-source operational data of physical entities in the cross-border trade supply chain through multiple data acquisition modules, including policy data, logistics data, financial data, market data, and IoT sensing data; Step 2: Digital Twin Modeling; Construct digital twin sub-entities of physical objects and an overall digital twin model of the supply chain, establish business relationships between digital twin sub-entities, and dynamically update the digital twin model based on real-time data; Step 3: Data processing; The collected data undergoes preprocessing operations such as outlier removal, normalization, and missing value imputation, and then various risk factors are extracted using a deep learning model; Step 4: Risk assessment; The AHP method is used to determine the weights of risk factors, and the risk level is quantitatively assessed by combining the fuzzy comprehensive evaluation method and the Markov chain prediction model. Step 5: Warning Generation; Based on dynamically adjusted warning thresholds, generate corresponding warning information and output it to the application output layer; The dynamic update process of the digital twin model in step 2 includes receiving incremental data from the data acquisition layer in real time, processing the incremental data quickly through edge computing nodes, extracting the state change characteristics of the physical object, and triggering parameter updates of the digital twin model if the state change exceeds a preset threshold. At the same time, the model update log is recorded for traceability and verification. Step 4 involves training the Markov chain prediction model using historical risk data as the training set, using risk levels as the state space, calculating the state transition probability matrix, and optimizing the model parameters through maximum likelihood estimation to maintain high prediction accuracy for the model's risk levels.

[0019] The following points should be noted in this article: 1. The accompanying drawings of the embodiments disclosed herein only relate to the structures involved in the embodiments disclosed herein; other structures can be referred to in general design.

[0020] 2. Where there is no conflict, the embodiments of this disclosure and the features in the embodiments can be combined with each other to obtain new embodiments.

[0021] Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

Claims

1. A cross-border trade supply chain risk early warning system based on digital twinning, comprising an entity layer, a data collection layer, a digital twinning modeling layer, a data processing layer, a risk early warning layer, a security protection layer, and an application output layer, characterized in that: The entity layer is the foundation of the system's monitoring objects, covering all key physical objects in the cross-border trade supply chain, including suppliers, logistics carriers, warehousing facilities, customs nodes, and purchasers. The data acquisition layer is used to collect multi-source operational data from the entity layer supply chain. The digital twin modeling layer constructs digital twins of physical objects and an overall digital twin model of the supply chain based on the collected data. The data processing layer preprocesses and extracts features from the collected data. The risk warning layer identifies and warns risks based on the digital twin model and the processed data. The application output layer outputs warning information and decision-making suggestions. The data acquisition layer includes multiple types of data acquisition modules, including an IoT sensor module, a government data interface module, a logistics information interface module, a financial data interface module, and a market data crawler module. The digital twin modeling layer includes an object modeling module, a relationship modeling module, and a dynamic update module; The data processing layer includes a data preprocessing module and a risk factor extraction module; The risk warning layer includes a risk assessment module, a warning threshold dynamic adjustment module, and a warning generation module; The security protection layer includes a data encryption module, an access control module, and an abnormal behavior monitoring module.

2. The digital-twin-based cross-border trade supply chain risk early warning system according to claim 1, characterized in that: The IoT sensor module of the data acquisition layer is used to collect real-time operational data such as the location, temperature, humidity, vibration, and load of the logistics carrier. The government data interface module is used to obtain policy data of various countries in cross-border trade. The logistics information interface module is used to collect data on transportation node status, transportation timeliness, and logistics costs. The financial data interface module is used to obtain data on exchange rates, interest rates, and payment settlement status. The market data crawler module is used to collect data on supply and demand, price fluctuations, and competitor dynamics.

3. The cross-border trade supply chain risk early warning system based on digital twinning according to claim 1, characterized in that: The object modeling module of the digital twin modeling layer uses parametric modeling and 3D visualization technology to construct a digital twin sub-entity for each physical object, which includes geometric features, physical attributes, and operating parameters. The relationship modeling module constructs the logistics connection relationship, information flow interaction relationship, and capital flow association relationship between the digital twin sub-entities based on the business process of the cross-border trade supply chain. The dynamic update module dynamically adjusts the parameters of the digital twin model according to the changes in the operating status of the physical object through real-time data acquisition and edge computing technology, ensuring the synchronization between the digital twin model and the physical world.

4. The digital-twin-based cross-border trade supply chain risk early warning system according to claim 1, characterized in that: The data preprocessing module of the data processing layer removes invalid data based on a box plot and isolated forest fusion algorithm, and standardizes the data format using Z-Score. The risk factor extraction module uses an improved CNN-LSTM hybrid model based on deep learning to extract policy risk factors, logistics risk factors, financial risk factors, supply and demand risk factors, and compliance risk factors from the preprocessed data.

5. The cross-border trade supply chain risk early warning system based on digital twins according to claim 1, characterized in that: The risk assessment module of the risk warning layer uses the Analytic Hierarchy Process (AHP) to determine the weights of each risk factor. It combines the fuzzy comprehensive evaluation method and the Markov chain prediction model to quantitatively assess the supply chain risk level for the current and future period. The risk levels are divided into low risk, medium risk, and high risk. The dynamic adjustment module of the warning threshold uses the Q-Learning reinforcement learning algorithm to dynamically adjust the warning thresholds for different risk types based on historical risk data and supply chain operation status. The warning generation module generates corresponding warning information based on the comparison between the risk assessment results and the warning thresholds, including the warning level, risk type, scope of impact, probability of occurrence, duration, and response suggestions.

6. A method for early warning of risks in cross-border trade supply chains based on digital twins, characterized in that: The method is implemented through the cross-border trade supply chain risk early warning system based on digital twins as described in claims 1-5, and the method steps include: Step 1: Data Acquisition; Collect multi-source operational data of physical entities in the cross-border trade supply chain through multiple data acquisition modules, including policy data, logistics data, financial data, market data, and IoT sensing data; Step 2: Digital Twin Modeling; Construct digital twin sub-entities of physical objects and an overall digital twin model of the supply chain, establish business relationships between digital twin sub-entities, and dynamically update the digital twin model based on real-time data; Step 3: Data processing; The collected data undergoes preprocessing operations such as outlier removal, normalization, and missing value imputation, and then various risk factors are extracted using a deep learning model; Step 4: Risk assessment; The AHP method is used to determine the weights of risk factors, and the risk level is quantitatively assessed by combining the fuzzy comprehensive evaluation method and the Markov chain prediction model. Step 5: Warning generation; Based on the dynamically adjusted warning threshold, generate the corresponding warning information and output it to the application output layer.

7. The cross-border trade supply chain risk early warning method based on digital twins according to claim 6, characterized in that: The dynamic update process of the digital twin model in step 2 includes receiving incremental data from the data acquisition layer in real time, processing the incremental data quickly through edge computing nodes, extracting the state change characteristics of the physical object, and triggering parameter updates of the digital twin model if the state change exceeds a preset threshold. At the same time, the model update log is recorded for traceability and verification.

8. The cross-border trade supply chain risk early warning method based on digital twins according to claim 6, characterized in that: The training process of the Markov chain prediction model in step 4 includes using historical risk data as the training set, using the risk level as the state space, calculating the state transition probability matrix, and optimizing the model parameters through the maximum likelihood estimation method to maintain a high prediction accuracy of the model's risk level.

Citation Information

Cited By

  • Cross-border e-commerce multi-currency settlement fund flow and logistics collaborative control method and system

    CN122198962A