Freight service verification method and system fused with multi-dimensional data

Abstract

The invention provides a freight service verification method and system fused with multi-dimensional data in the cross technical field of artificial intelligence and freight information verification. The method comprises the following steps: S1, creating a freight service risk identification model; s2, acquiring a large amount of historical freight service data to construct a data set to train a freight service risk identification model; s3, the server sets a service verification rule set comprising a plurality of verification rules; s4, the server obtains real-time freight service data through ETL; s5, calling a service verification rule set through a rule engine, and carrying out pre-verification on the real-time freight service data; s6, performing in-event verification on the real-time freight service data through the freight service risk identification model; and S7, constructing an incremental data set based on the real-time freight service data to carry out post optimization on the freight service risk identification model. The freight service verification method has the advantages that the accuracy and timeliness of freight service verification are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intersection between artificial intelligence and freight information verification, and in particular to a freight business verification method and system that integrates multidimensional data. Background Art

[0002] With the rapid development of information technology, particularly the deep integration and widespread adoption of the internet, big data, the Internet of Things (IoT), and cloud computing, the traditional road freight transportation industry is undergoing a profound digital and intelligent transformation. Against this backdrop, network freight platforms have emerged and are rapidly becoming a core component of the modern logistics system.

[0003] Online freight platforms leverage the internet's powerful information integration and resource allocation capabilities to efficiently connect and optimize the allocation of dispersed transportation resources (such as carrier vehicles and drivers). Acting as a "carrier," online freight platforms directly sign transportation service contracts with shippers, assuming overall transportation responsibility. At the same time, they do not directly participate in actual transportation operations, instead delegating specific cargo transportation tasks to qualified carriers connected to the platform. This innovative model significantly enhances the transparency of the logistics chain, improves the efficiency of transport resource utilization, reduces transaction costs caused by information asymmetry, and provides important support for promoting cost reduction, efficiency improvement, and high-quality development in the logistics industry.

[0004] However, while online freight platforms achieve efficient operations and value creation, their unique "asset-light, platform-based" operating model also introduces significant challenges and risks. Among these, the risk of business fraud is particularly prominent, directly threatening the platform's robust operations, integrity system development, data value mining, and compliance requirements. This risk manifests itself in the following forms: 1. Falsified waybill risk: Platform users (such as some shippers or actual carriers) may fabricate non-existent freight transportation requests or transaction information to generate false waybill data. Motives for this may include, but are not limited to, obtaining platform subsidies or freight rate differentials, inflating transaction volumes to boost platform rankings or secure financial support (such as waybill financing), and circumventing tax regulations. 2. Falsified track risk: To conceal irregularities (such as detours, unusual stops, or transporting prohibited items) or to facilitate fictitious waybill creation, actual carriers may fabricate or modify vehicle tracks through techniques such as GPS signal simulators, artificial interference with vehicle positioning devices, and back-end tampering with positioning data. Falsified tracks severely undermine the visibility and traceability of the transportation process, rendering regulatory and risk control measures ineffective. 3. Fund flow mismatch risk: There are unexplained discrepancies between the transaction amounts and payment settlement information recorded by the platform and the actual fund flows (such as bank statements and third-party payment statements). This may be due to falsified waybill information, offline transactions outside the platform's supervision, and off-book fund circulation.

[0005] The above-mentioned behaviors not only erode the trust foundation and market fairness of online freight platforms, leading to losses on the platforms and reducing user stickiness, but are also likely to trigger serious compliance risks and affect the construction of a healthy industry ecosystem. Therefore, it is necessary to verify the freight business. The verification of the freight business of online freight platforms has traditionally been done manually, but there are the following pain points: 1. Data silos and collaborative verification barriers: The freight business process of online freight platforms involves multiple links and a large amount of heterogeneous data, such as: transport contracts and cargo information on the shipper side; order information and user identity information on the platform side; vehicle / driver qualifications and waybill status on the actual carrier side; real-time vehicle positioning and tracking on the IoT side; and funds receipt and payment flow and invoice information on the payment settlement side. Under traditional architectures, this critical data is often stored in various subsystems within the platform (such as contract management systems, order systems, tracking systems, and financial settlement systems) or external third-party systems (such as payment gateways and tax systems). These systems often lack unified data interface standards, real-time and efficient data sharing mechanisms, and enhanced access control. Manual cross-system data extraction, comparison, and analysis are extremely inefficient and even technically challenging. This physical dispersion and logical isolation of data creates a severe "data silo" effect, making it difficult for verification personnel to conduct cross-verification and cross-corroboration. This makes it easy for fraudsters to exploit the incomplete information recorded in a single system, making it impossible to identify complex fraud schemes that involve coordinated fraud across multiple links and systems.

[0006] 2. Severe lag leads to failure of risk control: The manual verification process is often cumbersome and lengthy, relying on spot checks, report analysis, or post-event reporting, and exhibits significant lags. Verification often occurs after the problematic waybill is completed and the transaction is settled, even days or weeks later. This "post-event audit" model implies that the risk has already occurred: the problematic waybill has already been completed and included in the business volume, and the mismatched funds may have been transferred or concealed. Online freight platforms passively respond to problems after they occur, akin to "closing the fold after the horse has bolted." While they can pursue accountability and penalties for identified issues, they are unable to conduct real-time monitoring and proactively intercept key risk-prone links (such as waybill generation, track anomalies, and the eve of payment). This allows fraudulent behavior ample time to evade inspection, significantly reducing the preventative effectiveness of risk control measures.

[0007] 3. The single verification dimension leads to insufficient verification depth: Limited by the time constraints of manual verification and the limitations of data silos, traditional methods tend to focus on verifying the authenticity of single-point data. For example, a verifier might only check the format of an uploaded electronic waybill (for obvious signs of forgery) or perform a simple position comparison of individual vehicle track points. This verification model, based on a single data stream, has obvious flaws: it can verify the superficial authenticity of a single link (such as the waybill), but it cannot effectively assess the correlation and consistency of this data with other core data throughout the entire business process (such as the signing status of the corresponding contract, the consistency and reasonable speed of the waybill's track points, and the matching payment transaction amount and time). Counterfeiters can exploit the weaknesses of this "point-by-point verification" and carefully fabricate data at specific links to "pass" single-dimensional inspections. This results in superficial verification results that fail to reveal the deeper, multi-linked fraud chain.

[0008] In summary, while online freight platforms are improving logistics efficiency, they are also facing severe and increasingly complex risks of business fraud. Traditional verification methods that rely on manual operations, are limited to post-audits, and are trapped in data silos and a single verification dimension are inherently inefficient, lagging, and superficial. They can no longer meet the high requirements of online freight platform governance and the need to handle huge business volumes. Therefore, how to provide a freight business verification method and system that integrates multi-dimensional data to improve the accuracy and timeliness of freight business verification has become a technical problem that needs to be solved urgently. Summary of the Invention

[0009] The technical problem to be solved by the present invention is to provide a freight business verification method and system that integrates multi-dimensional data to improve the accuracy and timeliness of freight business verification.

[0010] In a first aspect, the present invention provides a method for verifying freight business by integrating multi-dimensional data, comprising the following steps: Step S1: Create a freight business risk identification model and set a loss function of the freight business risk identification model; Step S2: obtaining a large amount of historical freight business data including historical contract data, historical business data, historical trajectory data, historical capital data, and historical bill data, and constructing a data set after preprocessing and annotating each of the historical freight business data; Step S3: training a freight business risk identification model using the data set and the loss function, and deploying the trained freight business risk identification model to a server; Step S4: The server sets a business verification rule set including several verification rules; Step S5: The server uses ETL tools to obtain real-time freight business data, including real-time contract data, real-time business data, real-time track data, real-time fund data, and real-time bill data, from the contract system, business system, track system, fund system, and bill system. Step S6: The server calls the business verification rule set through a rule engine to perform a pre-verification on the real-time freight business data; Step S7: The server verifies the real-time freight business data through the deployed freight business risk identification model; Step S8: The server constructs an incremental data set based on the real-time freight business data, and performs post-optimization on the freight business risk identification model through the incremental data set.

[0011] In a second aspect, the present invention provides a freight business verification system integrating multi-dimensional data, including the following modules: A freight business risk identification model creation module is used to create a freight business risk identification model and set a loss function of the freight business risk identification model; A data set construction module is used to obtain a large amount of historical freight business data including historical contract data, historical business data, historical trajectory data, historical capital data, and historical bill data, and to construct a data set after preprocessing and annotating each of the historical freight business data; A freight business risk identification model training module, configured to train the freight business risk identification model using the data set and the loss function, and deploy the trained freight business risk identification model to a server; The business verification rule set setting module is used for the server to set a business verification rule set containing several verification rules; The multi-dimensional data acquisition module is used by the server to obtain real-time freight business data including real-time contract data, real-time business data, real-time trajectory data, real-time capital data and real-time bill data from the contract system, business system, tracking system, capital system and billing system through ETL tools; A pre-verification module is used for the server to call the business verification rule set through a rule engine to perform pre-verification on the real-time freight business data; An in-process verification module, configured to verify the real-time freight business data in-process by using the freight business risk identification model deployed on the server; A post-optimization module is used for the server to construct an incremental data set based on the real-time freight business data, and to perform post-optimization on the freight business risk identification model through the incremental data set.

[0012] The advantages of the present invention are: 1. Create a freight business risk identification model and set the loss function of the freight business risk identification model; then obtain a large amount of historical freight business data including historical contract data, historical business data, historical trajectory data, historical capital data and historical bill data to build a data set, train the freight business risk identification model through the data set and loss function, deploy the trained freight business risk identification model to the server, and set a business verification rule set containing several verification rules on the server; then the server uses the ETL tool to obtain real-time freight business data including real-time contract data, real-time business data, real-time trajectory data, real-time capital data and real-time bill data from the contract system, business system, trajectory system, capital system and bill system, and calls the business verification rule set to verify the real-time freight business data through the rule engine. The freight business data is verified in advance, the real-time freight business data is verified in-process through the deployed freight business risk identification model, an incremental data set is built based on the real-time freight business data, and the freight business risk identification model is optimized afterward through the incremental data set; that is, multi-dimensional data (real-time contract data, real-time business data, real-time trajectory data, real-time capital data and real-time bill data) is obtained from different systems through ETL tools to verify the freight business, so as to overcome the traditional data silos and collaborative verification barriers, and overcome the traditional problem of single verification dimension; the business verification rule set is called through the rule engine to conduct pre-verification, and the pre-trained freight business risk identification model is used for in-process verification to overcome the traditional serious lag that leads to risk control failure, and ultimately greatly improve the accuracy and timeliness of freight business verification.

[0013] 2. By integrating multi-dimensional data (including contract data, business data, trajectory data, financial data, and bill data), the limitations of a single data source are avoided. This enables the freight business risk identification model to capture more comprehensive risk characteristics (such as contract fraud, financial anomalies, or trajectory deviation) during risk identification, thereby improving the accuracy of verification. For example, the fusion of trajectory data and financial data can effectively identify fraudulent transportation activities.

[0014] 3. Automated verification is achieved through the dual mechanisms of a rules engine and a freight business risk identification model, greatly reducing manual intervention. Specifically, pre-verification (based on the rules engine) quickly screens common violations, while in-process verification (based on a trained model) handles complex risk scenarios (such as abnormal cash flows) in real time, shortening verification response time. At the same time, the use of ETL tools to automatically collect real-time freight business data greatly improves data processing efficiency.

[0015] 4. By constructing incremental data sets and optimizing the freight business risk identification model, post-closed-loop optimization is achieved. The freight business risk identification model can be automatically updated based on new data, enhancing its adaptability to changes in the business environment (such as new fraud methods). Combined with the defined loss function, it ensures continuous optimization of the freight business risk identification model and improves the long-term robustness of verification. This self-learning mechanism can resist "concept drift" (changes in data distribution), making the system more reliable and durable, unlike static solutions.

[0016] 5. An end-to-end risk management closed loop is established through three stages: pre-event, in-event, and post-event. Pre-event rule verification provides basic screening, in-event model verification handles complex logic, and post-event optimization ensures continuous improvement. This not only covers the entire life cycle of freight business (such as order execution to financial settlement), but also supports real-time decision-making (such as in-event verification can automatically trigger risk alerts), effectively reducing overall risk exposure and improving business security.

[0017] 6. Large-scale data is processed through ETL tools and dataset construction methods, ensuring efficient data utilization; historical freight business data is used to train freight business risk identification models, and real-time freight business data supports verification. Combined with a configurable business verification rule set (verification rules can be added at any time), it is easy to expand to new scenarios (such as multimodal freight), optimize data resources, reduce redundant storage and computing overhead, effectively improve hardware resource utilization, and support adaptation of multiple business types.

[0018] 7. By integrating multi-dimensional real-time and historical data (contracts, business, trajectory, funds, bills), combining the pre-verification of the rule engine with the in-process verification of the AI ​​model (freight business risk identification model), a dynamic risk management closed-loop system covering the entire freight business process has been built. Its core advantage is that it significantly improves the accuracy of risk identification (using multi-dimensional data to avoid blind spots from a single perspective) and timeliness (real-time verification and rapid response). At the same time, through layered processing (initial screening of rules + precise judgment of models), computing resources are optimized and labor costs are reduced. Relying on the incremental learning mechanism for continuous self-optimization, the model can be dynamically evolved to adapt to business changes, ultimately achieving the goal of automated, highly robust and scalable risk prevention and control, effectively reducing the risk of fraud and violations.

[0019] 8. The freight business risk identification model integrates multi-source heterogeneous data such as contract data, business data, trajectory data, financial data, and bill data through a feature extraction layer, enabling comprehensive monitoring of freight business risks. This avoids the limitations of a single data source and can capture a wider range of risk factors (such as contract breaches, financial anomalies, and trajectory deviations). By processing different types of data through dedicated modules (such as Transformer for text and CNN / GRU for sequences), the solution's practicality and market adaptability are enhanced, meeting the freight industry's needs for comprehensive risk assessment.

[0020] 9. Contract data is processed through the Transformer bidirectional encoder to capture key semantic dependencies and improve the accuracy of text understanding. Business and bill data are processed through a multi-layer fully connected network to model linear and nonlinear relationships and enhance the ability to analyze complex business indicators. Trajectory and capital data are processed by combining GRU and CNN. GRU captures temporal dependencies and CNN extracts local features (such as transaction patterns) to ensure efficient identification of dynamic risks. This hybrid architecture (combining Transformer, CNN, GRU, etc.) can efficiently process structured, unstructured and temporal data, reduce feature loss, and improve model robustness.

[0021] 10. The feature fusion layer dynamically calculates the feature weights of each feature through the multi-head attention module and performs weighted fusion. LSTM modeling of temporal dependencies and global average pooling for dimensionality reduction ensures that key risk features are given priority while processing the time series characteristics of the data (such as capital flow trends) to improve the fusion effect. The combination of multi-head attention and LSTM enables adaptive feature fusion, reduces redundant information, and improves computational efficiency.

[0022] 11. The prediction output layer adopts a dual-branch structure (risk item prediction and risk level prediction). It outputs the probability of independent risk items and risk levels respectively through Sigmoid and Softmax activation functions, and jointly outputs a comprehensive report. This allows the freight business risk identification model to handle multiple related tasks simultaneously (such as identifying specific risk items and assessing the overall risk level), improving the granularity and accuracy of the prediction. The multi-task learning framework reduces model complexity by sharing features (whole-graph feature vectors). The loss function design (binary cross entropy and categorical cross entropy weighting) balances the weights of different tasks, optimizes the training process, and effectively improves the generalization ability of the freight business risk identification model, making the output business risk identification report more comprehensive.

[0023] 12. The loss function is based on the weighted summation of risk term loss (binary cross entropy) and risk level loss (categorical cross entropy). This allows for dynamic adjustment of the priorities of different tasks during training (e.g., prioritizing high-risk levels or specific risk items), thereby improving the convergence speed and stability of the freight business risk identification model. This innovative weighted loss mechanism can effectively address imbalance issues in multi-task learning (e.g., the low frequency of certain risk items), reduce the risk of overfitting, and enhance the reliability of the freight business risk identification model in real-world scenarios.

[0024] 13. Modular design (e.g., data processing in modules at the feature extraction layer) and feature integration (global average pooling) reduce computational complexity and improve inference speed. The model structure is easy to expand, for example, by adding new data sources (e.g., weather data) or adjusting modules (e.g., replacing GRU with Transformer) to adapt to different freight scenarios. While maintaining high performance, it reduces hardware resource requirements, making it suitable for edge computing or cloud deployment. Its scalability enhances long-term competitiveness.

[0025] 14. By integrating five types of heterogeneous data, namely contracts, business, trajectory, funds, and bills, the system deeply explores multi-dimensional risk features through targeted neural network modules (such as Transformer to extract contract semantics, CNN+GRU to capture fund sequences, and fully connected layer modeling of business indicators). It also uses a multi-head attention mechanism to adaptively weight and fuse features, combining LSTM modeling of temporal dependencies with global average pooling to compress key information. On this basis, a comprehensive risk report is generated through a dual-task collaborative prediction mechanism (Sigmoid output of independent risk item probability + Softmax classification risk level). Its multi-task weighted loss function (binary cross entropy + categorical cross entropy) effectively balances training objectives, ensuring high-precision risk identification while significantly improving the model's generalization ability and computational efficiency for complex freight scenarios, ultimately achieving end-to-end automated risk control decision support.

[0026] 15. By adopting multi-step pre-processing (including deduplication, missing value processing, erroneous data correction, data standardization, data normalization and data encoding), the original data is systematically cleaned and transformed, reducing the impact of noise, inconsistency and bias, and ensuring the consistency and operability of the data set; for example, standardization and normalization processing makes data from different sources (such as speed information and amount information) comparable, thereby improving the accuracy of subsequent risk labeling and model training.

[0027] 16. By integrating multi-dimensional historical freight data (including contract, business, trajectory, financial and bill data), and using systematic pre-processing (such as deduplication and standardization) to improve data quality, combining risk annotation to build a high-quality data set, and then innovatively applying adversarial networks to expand the sample size, the comprehensiveness, accuracy and robustness of freight risk identification have been significantly improved, effectively solving the prediction bias problem caused by single data or insufficient samples in traditional models, and at the same time achieving automated and efficient processing, providing scalable technical support for logistics risk management.

[0028] 17. Stratified sampling was used to divide the dataset into training, validation, and test sets in a ratio of 8:1:1. This ensured balanced and representative data distribution, reduced sampling bias, and improved the model's generalization ability (i.e., the model's performance on unseen data), thereby reducing the misjudgment rate in freight business risk identification. The 8:1:1 division optimized resource utilization, avoided overfitting or underfitting, and enhanced the model's reliability.

[0029] 18. During the training phase, by continuously optimizing hyperparameters (such as learning rate and batch size) until the loss value of the loss function is less than the preset threshold, the model achieves rapid convergence and efficient training, reducing training time and computing resource consumption while ensuring that the model achieves high accuracy in freight risk identification tasks; the iterative optimization mechanism (such as dynamic adjustment based on loss value) improves the adaptability of training, enabling the model to better handle the noise and complexity in freight data.

[0030] 19. The validation set is used to independently evaluate the model accuracy and set an accuracy threshold as the passing standard. If the threshold is not met, the training set is expanded to continue training. This introduces a feedback loop to improve the model by increasing data diversity (such as adding more freight cases). This avoids the model from "stagnating" during the validation phase, improves the robustness of the model, ensures its high recognition accuracy in a changing freight environment (such as risk patterns in different regions or seasons), and reduces the risk of missed or false positives in business operations.

[0031] 20. The test set is used for the final evaluation of the model, and a confidence threshold check (such as the confidence level of the model prediction) is introduced, which increases the credibility of risk identification. If the confidence level does not meet the standard, the training set is expanded and retrained to form an iterative improvement. This ensures strict verification of the model before actual deployment, reduces the probability of misjudgment of high-risk events in the freight business (such as cargo loss or fraud), and provides quantitative evaluation through confidence indicators to facilitate monitoring and optimization.

[0032] 21. By optimizing data set division through stratified sampling, combined with iterative training and dual threshold verification mechanisms, the accuracy, robustness and reliability of the freight business risk identification model have been significantly improved; containerization technology has been used to achieve efficient deployment and elastic expansion, significantly reducing operation and maintenance costs; at the same time, through terminal authentication and structured storage of rule sets, the customizability of business rules and real-time decision-making capabilities are enhanced while ensuring system security, forming a closed-loop optimization system from data preprocessing to model deployment, effectively solving the core pain points of low risk identification efficiency, high misjudgment rate and poor system scalability in freight scenarios.

[0033] 22. Through ETL tools, the real-time freight business data of the five major systems of contract, business, track, capital and bill are automatically integrated. After standardized processing such as cleaning, formatting and aggregation, a unified high-quality data warehouse is built, which completely breaks down information silos and provides real-time, accurate and multi-dimensional data support for business decision-making, significantly improving operational efficiency, risk management capabilities and core competitiveness of the enterprise.

[0034] 23. Automatically call the rules engine through the API interface to trigger the processing of freight business events in real time, avoiding the delay of traditional manual verification; the rules engine reads the verification rule set from the specified path and performs multi-dimensional checks on the data (such as data format, integrity, logical consistency, vehicle qualifications, personnel qualifications, vehicle trajectory, waybill time and invoice information), ensuring that the verification process is efficient and automated, significantly reducing business interruption time and improving the overall throughput of the freight process (for example, reports can be quickly generated in the event of failure), thereby improving operational efficiency and reducing labor costs.

[0035] 24. A multi-layer encryption mechanism (including SM9, RC6, character shifting, and split-and-swap) provides enhanced data protection. Specifically, hash values ​​are used to ensure data integrity and prevent tampering with verification reports and times. The SM9 algorithm provides post-quantum asymmetric encryption, and primary encrypted data ensures the confidentiality of sensitive information. Split-and-swap (splitting in a 7:3 ratio and swapping the order) increases the difficulty of cracking. The RC6 algorithm further encrypts and introduces perturbations by shifting characters right by 5 bits, making the encryption process more complex. This multi-step process significantly improves the data's resistance to attacks and is suitable for high-security freight data environments. Finally, it is uploaded to the blockchain, leveraging its distributed ledger characteristics to ensure that the data cannot be tampered with and is traceable, enhancing the non-repudiation of the entire system.

[0036] 25. Real-time automated verification of freight business data (covering key dimensions such as data format, integrity, logical consistency, qualifications and timeliness) is achieved through a rule engine, significantly improving verification efficiency and accuracy and avoiding manual delays. At the same time, a multi-layer dynamic encryption mechanism (SM9 algorithm, RC6 algorithm combined with data segmentation, transposition and displacement operations) and blockchain evidence storage are adopted to ensure that data is tamper-proof and traceable throughout the process, and verification failure reports are pushed in real time through the TLS protocol to achieve immediate early warning and closed-loop management of business risks. Ultimately, on the basis of ensuring freight safety and compliance, operating costs are significantly reduced and system reliability is enhanced.

[0037] 26. Using hardware acceleration technology (such as GPU or FPGA) to reason about freight business risk identification models significantly reduces the computational delay of real-time data processing, enables in-process verification, avoids the time lag of traditional batch processing, and has a high response speed (millisecond-level reasoning). It can promptly identify and handle high-risk freight incidents (such as cargo damage or fraud), thereby reducing business losses. This real-time performance enhances the practicality of the system, especially for high-frequency and dynamic freight scenarios.

[0038] 27. By adopting multi-level encryption (including XTEA, IDEA algorithms and custom character swapping) and combining it with hash value verification (such as the second hash value based on the report and verification time), strong data security protection is provided: (1) Algorithm fusion improves security: The XTEA algorithm is used for preliminary encryption to ensure basic data security; then, an obfuscation layer is introduced by character swapping (such as swapping 7 with B, 8 with 9, and 10 with A in hexadecimal data) to increase the difficulty of reverse engineering and resist pattern recognition attacks; the IDEA algorithm performs deep encryption to form a "three-layer" protection; (2) Data integrity protection: Hash values ​​are embedded in the hash value calculation and encryption process to ensure that data is not tampered with during transmission and storage, effectively responding to the risk of man-in-the-middle attacks or data tampering; (3) The TLS protocol ensures real-time communication security: When the risk level is higher than the preset value, a report is pushed through the TLS protocol to prevent data from being eavesdropped or intercepted during transmission; (4) Blockchain technology ensures permanent immutability: Finally, the encrypted report is uploaded to the blockchain, using the characteristics of the distributed ledger to provide traceability and audit capabilities.

[0039] 28. Through hardware acceleration, millisecond-level in-process verification of freight business risks is achieved. Combined with an original multi-layer encryption mechanism (integrating XTEA / IDEA algorithms and character swapping and obfuscation technology) and blockchain evidence storage, while ensuring the security of real-time business risk report transmission and data immutability, the freight business risk identification model driven by incremental data sets is optimized in the closed loop, continuously improving risk identification accuracy, forming a closed-loop intelligent risk control system from real-time warning to dynamic evolution, significantly reducing freight business operational risks and manual intervention costs.

[0040] 29. By integrating multi-dimensional heterogeneous data (contracts, business, trajectory, funds, bills), a freight business risk identification model is constructed. Combined with the rule engine, it achieves dual guarantees of pre-verification and in-process model prediction. Incremental data is used to optimize the freight business risk identification model in real time, forming a full-process closed-loop risk control system. The innovation is reflected in the use of modules such as Transformer, GRU, and multi-head attention to achieve deep feature extraction and fusion, improving risk identification accuracy. At the same time, hardware acceleration, containerized deployment and blockchain encryption ensure system efficiency, scalability and data security, significantly reducing freight business risks and improving verification efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0042] Figure 1 It is a flow chart of a freight business verification method integrating multi-dimensional data according to the present invention.

[0043] Figure 2 It is a structural diagram of a freight business verification system integrating multi-dimensional data according to the present invention. DETAILED DESCRIPTION

[0044] The overall idea of ​​the technical solution in the embodiments of the present application is as follows: multi-dimensional data (real-time contract data, real-time business data, real-time trajectory data, real-time fund data and real-time bill data) is obtained from different systems through ETL tools to verify freight business, so as to overcome the traditional data silos and collaborative verification barriers, and overcome the traditional problem of single verification dimension; the business verification rule set is called through the rule engine to perform pre-verification, and the pre-trained freight business risk identification model is used to perform in-process verification, so as to overcome the traditional serious lag that leads to risk control failure, thereby improving the accuracy and timeliness of freight business verification.

[0045] Please refer to Figures 1 to 2 As shown, a preferred embodiment of a freight business verification method integrating multi-dimensional data of the present invention includes the following steps: Step S1: Create a freight business risk identification model and set a loss function of the freight business risk identification model; Step S2: obtaining a large amount of historical freight business data including historical contract data, historical business data, historical trajectory data, historical capital data, and historical bill data, and constructing a data set after preprocessing and annotating each of the historical freight business data; By integrating multi-dimensional data (including contract data, business data, trajectory data, financial data, and bill data), the limitations of a single data source are avoided. This enables the freight business risk identification model to capture more comprehensive risk characteristics (such as contract fraud, financial anomalies, or trajectory deviations) during risk identification, thereby improving the accuracy of verification; for example, the fusion of trajectory data and financial data can effectively identify false transportation activities.

[0046] Step S3: training a freight business risk identification model using the data set and the loss function, and deploying the trained freight business risk identification model to a server; Step S4: The server sets a business verification rule set including several verification rules; Step S5: The server uses ETL tools to obtain real-time freight business data, including real-time contract data, real-time business data, real-time track data, real-time fund data, and real-time bill data, from the contract system, business system, track system, fund system, and bill system. Step S6: The server calls the business verification rule set through a rule engine to perform a pre-verification on the real-time freight business data; Step S7: The server verifies the real-time freight business data through the deployed freight business risk identification model; Automated verification is achieved through the dual mechanisms of a rules engine and a freight business risk identification model, greatly reducing manual intervention. Specifically, pre-verification (based on a rules engine) quickly screens common violations, while in-process verification (based on a trained model) handles complex risk scenarios (such as abnormal cash flows) in real time, shortening verification response time. At the same time, the use of ETL tools to automatically collect real-time freight business data greatly improves data processing efficiency.

[0047] Step S8: The server constructs an incremental data set based on the real-time freight business data, and performs post-optimization on the freight business risk identification model using the incremental data set; By constructing incremental data sets and optimizing the freight business risk identification model, post-closed-loop optimization is achieved. The freight business risk identification model can be automatically updated based on new data, enhancing its adaptability to changes in the business environment (such as new fraud methods). Combined with the defined loss function, it ensures continuous optimization of the freight business risk identification model and improves the long-term robustness of verification. This self-learning mechanism can resist "concept drift" (changes in data distribution), making the system more reliable and durable, different from static solutions.

[0048] Through the three stages of pre-event, in-event and post-event, an end-to-end risk management closed loop is established; pre-event rule verification provides basic screening, in-event model verification handles complex logic, and post-event optimization ensures continuous improvement. It not only covers the entire life cycle of freight business (such as order execution to financial settlement), but also supports real-time decision-making (such as in-event verification can automatically trigger risk alerts), effectively reducing overall risk exposure and improving business security.

[0049] In step S1, the freight business risk identification model is constructed based on the feature extraction layer, the feature fusion layer and the prediction output layer; The feature extraction layer is constructed based on the contract data processing module, the business data processing module, the trajectory data processing module, the capital data processing module and the bill data processing module; the contract data processing module is used to extract key semantic dependencies from the contract data through a bidirectional encoder constructed by Transformer to obtain contract risk features; the business data processing module is used to model the linear and nonlinear relationships of business indicators of the business data through a multi-layer first fully connected network to obtain business operation features; the trajectory data processing module is used to extract trajectory risk features from the trajectory data through a first gated loop unit; the capital data processing module is used to capture local transaction features from the capital data through a one-dimensional convolutional neural network, and capture the serial dependency features of funds from the capital data through a second gated loop unit, and output capital flow features based on the local transaction features and the serial dependency features; the bill data processing module is used to extract bill integrity features from the bill data through a multi-layer second fully connected network; The freight business risk identification model integrates multi-source heterogeneous data such as contract data, business data, trajectory data, financial data, and bill data through a feature extraction layer, enabling comprehensive monitoring of freight business risks. This avoids the limitations of a single data source and can capture a wider range of risk factors (such as contract breaches, financial anomalies, and trajectory deviations). By processing different types of data through dedicated modules (such as Transformer for text and CNN / GRU for sequences), the solution's practicality and market adaptability are enhanced, meeting the freight industry's needs for comprehensive risk assessment.

[0050] Contract data is processed through the Transformer bidirectional encoder to capture key semantic dependencies and improve the accuracy of text understanding; business and bill data are processed through a multi-layer fully connected network to model linear and nonlinear relationships and enhance the ability to parse complex business indicators; trajectory and capital data are processed through a combination of GRU and CNN, with GRU capturing temporal dependencies and CNN extracting local features (such as transaction patterns) to ensure efficient identification of dynamic risks; this hybrid architecture (combining Transformer, CNN, GRU, etc.) can efficiently process structured, unstructured and temporal data, reduce feature loss, and improve model robustness.

[0051] The feature fusion layer is constructed based on a multi-head attention module, a long short-term memory module, and a feature integration module; the multi-head attention module is used to calculate the feature weights of contract risk features, business operation features, trajectory risk features, capital flow features, and bill integrity features through multi-head self-attention units, and fuse the contract risk features, business operation features, trajectory risk features, capital flow features, and bill integrity features based on the feature weights to obtain fused features; the long short-term memory module is used to model the temporal dependency of the fused features through LSTM units to obtain serialized features; the feature integration module is used to perform global average pooling on the serialized features to obtain the whole image feature vector; The feature fusion layer dynamically calculates the feature weights of each feature through the multi-head attention module and performs weighted fusion. It then uses LSTM modeling to model temporal dependencies and global average pooling for dimensionality reduction, ensuring that key risk features are given priority while processing the time series characteristics of the data (such as capital flow trends) to improve the fusion effect. The combination of multi-head attention and LSTM realizes adaptive fusion of features, reduces redundant information, and improves computational efficiency.

[0052] The prediction output layer is constructed based on the risk item prediction branch, the risk level prediction branch and the joint output module; the risk item prediction branch is used to infer the feature vector of the entire graph through the first fully connected layer and the Sigmoid activation function to obtain the independent probabilities of different risk items (such as fraud, delay, default, etc.), supporting multi-label prediction; the risk level prediction branch is used to infer the feature vector of the entire graph through the second fully connected layer and the Softmax activation function to obtain the category probability of the risk level (low, medium, high); the joint output module is used to output a business risk identification report including risk items and risk levels based on the independent probabilities and category probabilities.

[0053] The prediction output layer adopts a dual-branch structure (risk item prediction and risk level prediction), and outputs the probability of independent risk items and risk levels respectively through Sigmoid and Softmax activation functions, and jointly outputs a comprehensive report. This allows the freight business risk identification model to handle multiple related tasks simultaneously (such as identifying specific risk items and assessing the overall risk level), improving the granularity and accuracy of the prediction; the multi-task learning framework reduces model complexity by sharing features (whole-graph feature vectors), and the loss function design (binary cross entropy and categorical cross entropy weighting) balances the weights of different tasks, optimizes the training process, and effectively improves the generalization ability of the freight business risk identification model, making the output business risk identification report more comprehensive.

[0054] In step S1, the loss function is constructed based on the weighted sum of risk item loss and risk level loss; the risk item loss adopts binary cross entropy loss; the risk level loss adopts categorical cross entropy loss; The loss function is based on the weighted summation of risk term loss (binary cross entropy) and risk level loss (categorical cross entropy). This allows for dynamic adjustment of the priorities of different tasks during training (e.g., focusing more on high-risk levels or specific risk items), thereby improving the convergence speed and stability of the freight business risk identification model. This weighted loss mechanism is innovative and can effectively address imbalance issues in multi-task learning (e.g., the low frequency of occurrence of certain risk items), reduce the risk of overfitting, and enhance the reliability of the freight business risk identification model in real-world scenarios.

[0055] The step S2 is specifically as follows: Obtain a large amount of historical freight business data including historical contract data, historical business data, historical trajectory data, historical capital data, and historical bill data; The historical contract data includes at least basic contract information, cooperation entity information, cargo information, transportation service requirements, fee clauses, liability for breach of contract, and dispute resolution methods; The basic contract information at least includes the contract number, signing date, and contract validity period, which are used to classify, search, and manage the contract to ensure the timeliness and traceability of the contract; the cooperative entity information at least includes the shipper name, carrier name, address, contact information, and business license number, which are used to clarify the identity and responsible parties of the contract parties and provide a basis for subsequent business development and dispute resolution; the cargo information at least includes the name, specifications, model, weight, volume, quantity, and value of the cargo, which are used to reasonably arrange transportation vehicles, calculate freight, and ensure cargo safety; the transportation service requirements at least include the departure place, destination, transportation time requirements (such as the latest pickup time, the latest delivery time), transportation mode (road, rail, The terms of the contract shall include at least the calculation method of the freight (such as by weight, volume or mileage), the payment method (such as prepaid, collect, monthly settlement, etc.), the payment time, and the invoice issuance requirements, so as to clarify the financial responsibilities and settlement process of both parties. The liability for breach of contract shall specify the corresponding liability for breach of contract and compensation for possible breach of contract that may occur during the performance of the contract, such as delayed delivery, damage or loss of goods, etc., to protect the legitimate rights and interests of both parties. The dispute resolution method shall stipulate the means of resolving contract disputes, such as negotiation, mediation, arbitration or litigation, as well as the applicable laws and arbitration institutions or competent courts, to provide a resolution mechanism for possible disputes.

[0056] The business data includes at least waybill information, order logs, online transaction logs, customer information, vehicle and driver information, and logistics information; The waybill information at least includes the waybill number, shipper information, consignee information, cargo description (consistent with the cargo information in the contract), transportation route, transportation mileage, estimated transportation time, actual transportation time, waybill status (such as ordered, picked up, in transit, delivered, etc.), which is used to track and manage the entire process of each transportation business; the order log is used to record the time, operator and reason for operations such as order generation, modification, and cancellation, so as to facilitate the tracing of order changes and ensure the transparency and auditability of business operations; the online transaction log at least includes records of users' online transaction behaviors such as logging in, browsing, querying, placing orders, quoting, and closing transactions on the online freight platform, which is used to reflect users' trading habits and the platform's trading activity, and provide a basis for data analysis and business optimization; the customer information, in addition to the consignment information involved in the contract, In addition to the information of people and carriers, it also includes the customer's registration information, credit rating, historical transaction records, preference settings, etc., which helps the online freight platform to better understand customer needs and provide personalized services, and is also conducive to customer relationship management and risk assessment; the vehicle and driver information at least includes license plate number, vehicle model, approved load capacity, vehicle status (such as idle, in transit, maintenance, etc.), driver's name, driver's contact information, driver's license information, professional qualification certificate information, driving experience, and credit record, which are used to reasonably dispatch vehicle and driver resources and ensure transportation safety; the logistics information at least includes the real-time location of the goods, transportation status (such as whether it is transported normally, whether there are any abnormal situations, etc.), loading and unloading time, and transit information. By combining with trajectory data, full monitoring and visual management of the logistics process can be achieved, thereby improving logistics efficiency and customer satisfaction.

[0057] The trajectory data includes at least time information, geographic location information, driving path information, speed information, mileage information and stop point information; The time information is the timestamp of the vehicle at each location point, including the departure time, the time of passing through each node, the time of arrival at the destination, etc. The vehicle's driving speed, stay time, etc. can be calculated through the time information, and then the transportation efficiency and whether there are abnormal stops can be analyzed; the geographical location information is the real-time latitude and longitude coordinates of the vehicle during transportation, which is obtained through positioning equipment (such as GPS, Beidou, etc.) to determine the precise location of the vehicle, draw the vehicle's driving trajectory, and realize dynamic tracking and monitoring of the transportation process; the driving path information is the actual route of the vehicle, including the road sections, intersections, highway entrances and exits passed by, etc., which can be compared with the planned transportation route to determine whether the vehicle is driving according to the planned route and whether there are abnormal situations such as detours and deviations; the speed information The information refers to the vehicle's driving speed in different time periods. By monitoring the speed, it is possible to analyze whether the vehicle is speeding and whether there are abnormal speed changes due to factors such as road conditions and traffic control, providing a reference for transportation safety and efficiency evaluation; the mileage information refers to the mileage of the vehicle during transportation, which can be used to calculate costs such as fuel consumption and vehicle wear and tear. At the same time, combined with information such as cargo weight, it can also evaluate transportation cost-effectiveness and provide data support for the optimization of transportation plans; the stop point information refers to the vehicle's stop location and stop duration during transportation, including loading and unloading locations, gas stations, rest areas, etc. By analyzing the stop points, the vehicle's operating mode and the driver's driving habits can be understood, and it is also helpful to discover potential risk points, such as the risk of cargo theft due to long-term stops.

[0058] The fund data includes at least fund flow information, account balance information and settlement information; The fund flow information includes the income and expenditure records of each fund, such as the collection of freight, freight paid to the actual carrier, collection of platform service fees, etc., and records in detail the flow of funds, amount, transaction time, transaction object and other information to ensure the transparency and traceability of funds; the account balance information is the balance of the shipper and carrier's funds account on the platform, reflecting the financial status and available funds of both parties, and providing a basis for fund settlement and financial management; the settlement information includes at least the settlement period, settlement method, settlement time, and settlement amount, which is used to clarify the settlement rules and procedures of funds, ensure timely and accurate settlement of funds, and safeguard the legitimate rights and interests of all parties.

[0059] The bill data at least includes invoice number, invoice date, invoice amount, and invoice type; The historical freight business data are preprocessed by at least removing duplicate data, processing missing values, correcting erroneous data, standardizing data, normalizing data, and encoding data; the preprocessed historical freight business data are labeled with risk items and risk levels; a data set is constructed using the labeled historical freight business data; and the sample size of the data set is expanded using an adversarial network.

[0060] By adopting multi-step preprocessing (including duplicate data removal, missing value processing, erroneous data correction, data standardization, data normalization and data encoding), the original data is systematically cleaned and transformed, reducing the impact of noise, inconsistency and bias, and ensuring the consistency and operability of the data set; for example, standardization and normalization processing makes data from different sources (such as speed information and amount information) comparable, thereby improving the accuracy of subsequent risk labeling and model training.

[0061] By integrating multi-dimensional historical freight data (including contract, business, trajectory, financial and bill data), and using systematic pre-processing (such as deduplication and standardization) to improve data quality, combining risk annotation to build a high-quality data set, and then innovatively applying adversarial networks to expand the sample size, the comprehensiveness, accuracy and robustness of freight risk identification have been significantly improved, effectively solving the prediction bias problem caused by single data or insufficient samples in traditional models, and at the same time achieving automated and efficient processing, providing scalable technical support for logistics risk management.

[0062] The step S3 is specifically as follows: The data set is divided into a training set, a validation set, and a test set in a ratio of 8:1:1 by a stratified sampling method. The freight business risk identification model is trained using the training set. During the training process, the hyperparameters of the freight business risk identification model are continuously optimized until the loss value of the loss function is less than a preset loss threshold; The trained freight business risk identification model is verified using the validation set to determine whether the recognition accuracy is greater than a preset accuracy threshold. If not, the verification fails and the training set is expanded to continue training. If so, the verification passes, and: The verified freight business risk identification model is tested using the test set to determine whether the confidence level is greater than a preset confidence threshold. If not, the test fails, and the training set is expanded to continue training. If yes, the test passes, and the freight business risk identification model that has passed the test is deployed to the server using containerization technology. Using stratified sampling, the dataset was divided into training, validation, and test sets in a ratio of 8:1:1. This ensured balanced and representative data distribution, reduced sampling bias, and improved the model's generalization ability (i.e., its performance on unseen data), thereby reducing the misjudgment rate in freight business risk identification. The 8:1:1 division optimized resource utilization, avoided overfitting or underfitting problems, and enhanced the model's reliability.

[0063] During the training phase, by continuously optimizing hyperparameters (such as learning rate and batch size) until the loss value of the loss function is less than the preset threshold, the model achieves rapid convergence and efficient training, reducing training time and computing resource consumption, while ensuring that the model achieves high accuracy in freight risk identification tasks; the iterative optimization mechanism (such as dynamic adjustment based on loss value) improves the adaptability of training, enabling the model to better handle the noise and complexity in freight data.

[0064] By optimizing data set division through stratified sampling and combining iterative training with a dual-threshold verification mechanism, the accuracy, robustness, and reliability of the freight business risk identification model have been significantly improved. Containerization technology is used to achieve efficient deployment and elastic expansion, significantly reducing operation and maintenance costs. At the same time, through terminal authentication and structured storage of rule sets, the customizability of business rules and real-time decision-making capabilities are enhanced while ensuring system security. This has formed a closed-loop optimization system from data preprocessing to model deployment, effectively solving the core pain points of low risk identification efficiency, high misjudgment rate, and poor system scalability in freight scenarios.

[0065] The step S4 is specifically as follows: After authenticating the accessing mobile terminal, the server obtains several verification rules set by the mobile terminal, constructs a business verification rule set based on each of the verification rules, and stores the business verification rule set in a structured manner to a designated path.

[0066] The step S5 is specifically as follows: The server uses an ETL tool to extract real-time freight business data, including real-time contract data, real-time business data, real-time track data, real-time fund data, and real-time bill data, from the contract system, business system, track system, fund system, and bill system based on a preset extraction cycle. The server performs data conversion operations on each of the real-time freight business data, including at least data cleansing, data formatting, and data aggregation, and loads the real-time freight business data after the data conversion operations into a preset data warehouse. The step S6 is specifically as follows: Based on a freight business trigger event, the server calls a rule engine through an API interface. The rule engine reads the business verification rule set from a specified path, and uses the business verification rule set to perform pre-verification of real-time freight business data, including at least data format, data integrity, logical consistency, vehicle qualifications, personnel qualifications, vehicle trajectory, waybill timeliness, and invoice information, and generates a pre-verification report containing pre-verification success or event verification failure information; When the pre-verification report indicates that the pre-verification fails, the pre-verification report is pushed to the pre-associated management terminal in real time via the TLS protocol; The server obtains the pre-verification time, calculates the pre-verification report and the first hash value of the pre-verification time, encrypts the pre-verification report, pre-verification time and the first hash value into first-level encrypted data through the SM9 algorithm, divides the first-level encrypted data into a ratio of 7:3 and swaps the order to obtain second-level encrypted data, encrypts the second-level encrypted data into third-level encrypted data through the RC6 algorithm, shifts each character of the third-level encrypted data 5 bits to the right to obtain a pre-encrypted report, and uploads the pre-encrypted report to the blockchain.

[0067] The rule engine is automatically called through the API interface to trigger the processing of freight business events in real time, avoiding the delay of traditional manual verification; the rule engine reads the verification rule set from the specified path and performs multi-dimensional checks on the data (such as data format, integrity, logical consistency, vehicle qualifications, personnel qualifications, vehicle trajectory, waybill timeliness and invoice information), ensuring that the verification process is efficient and automated, significantly reducing business interruption time and improving the overall throughput of the freight process (for example, reports can be quickly generated in the event of failure), thereby improving operational efficiency and reducing labor costs.

[0068] A multi-layer encryption mechanism (including the SM9 algorithm, RC6 algorithm, character shifting, and splitting and transposing) provides enhanced protection for data. Specifically, hash values ​​are used to ensure data integrity and prevent verification reports and time from being tampered with. The SM9 algorithm provides post-quantum-level asymmetric encryption, and the first-level encrypted data ensures the confidentiality of sensitive information. Splitting and transposing (splitting in a 7:3 ratio and transposing the order) is used to increase the difficulty of cracking. The RC6 algorithm further encrypts and introduces disturbances by shifting the characters right by 5 bits, making the encryption process more complex. This multi-step process significantly improves the data's resistance to attacks and is suitable for freight data environments with high security requirements. Finally, it is uploaded to the blockchain, leveraging its distributed ledger characteristics to ensure that the data cannot be tampered with and is traceable, enhancing the non-repudiation of the entire system.

[0069] The step S7 is specifically as follows: The server inputs the real-time freight business data into a deployed freight business risk identification model. The freight business risk identification model performs inference using hardware acceleration technology to obtain a real-time business risk identification report including real-time risk items and real-time risk levels, thereby performing in-process verification of the real-time freight business data. When the risk level is higher than a preset level, the real-time business risk identification report is pushed to a pre-associated management terminal in real time via the TLS protocol; The server obtains the in-process verification time, calculates the real-time business risk identification report and the second hash value of the in-process verification time, encrypts the real-time business risk identification report, the in-process verification time, and the second hash value into a first-layer encrypted data using the XTEA algorithm, converts the first-layer encrypted data into hexadecimal data, swaps the number 7 with the letter B, the number 8 with the number 9, and the number 10 with the letter A in the hexadecimal data to obtain a second-layer encrypted data, encrypts the second-layer encrypted data into a third-layer encrypted data using the IDEA algorithm, adds a random string of a specified length to a specified position in the third-layer encrypted data to obtain an in-process encrypted report, and uploads the in-process encrypted report to the blockchain; By using hardware acceleration technology (such as GPU or FPGA) to infer freight business risk identification models, the computational delay of real-time data processing is significantly reduced, enabling in-process verification and avoiding the time lag of traditional batch processing. With high response speed (millisecond-level inference), high-risk freight incidents (such as cargo damage or fraud) can be identified and handled promptly, thereby reducing business losses. This real-time performance enhances the practicality of the system, making it particularly suitable for high-frequency, dynamic freight scenarios.

[0070] The step S8 is specifically as follows: The server constructs an incremental dataset based on the real-time freight business data, labels the incremental dataset with real-time risk items, real-time risk levels, and actual risk items and levels. When the amount of data in the incremental dataset exceeds a preset threshold, the server performs post-optimization on the freight business risk identification model using the incremental dataset. The real-time risk items and real-time risk levels are predicted data, while the actual risk items and levels are actual data.

[0071] A preferred embodiment of a freight business verification system integrating multi-dimensional data of the present invention includes the following modules: A freight business risk identification model creation module is used to create a freight business risk identification model and set a loss function of the freight business risk identification model; A data set construction module is used to obtain a large amount of historical freight business data including historical contract data, historical business data, historical trajectory data, historical capital data, and historical bill data, and to construct a data set after preprocessing and annotating each of the historical freight business data; By integrating multi-dimensional data (including contract data, business data, trajectory data, financial data, and bill data), the limitations of a single data source are avoided. This enables the freight business risk identification model to capture more comprehensive risk characteristics (such as contract fraud, financial anomalies, or trajectory deviations) during risk identification, thereby improving the accuracy of verification; for example, the fusion of trajectory data and financial data can effectively identify false transportation activities.

[0072] A freight business risk identification model training module, configured to train the freight business risk identification model using the data set and the loss function, and deploy the trained freight business risk identification model to a server; The business verification rule set setting module is used for the server to set a business verification rule set containing several verification rules; The multi-dimensional data acquisition module is used by the server to obtain real-time freight business data including real-time contract data, real-time business data, real-time trajectory data, real-time capital data and real-time bill data from the contract system, business system, tracking system, capital system and billing system through ETL tools; A pre-verification module is used for the server to call the business verification rule set through a rule engine to perform pre-verification on the real-time freight business data; An in-process verification module, configured to verify the real-time freight business data in-process by using the freight business risk identification model deployed on the server; Automated verification is achieved through the dual mechanisms of a rules engine and a freight business risk identification model, greatly reducing manual intervention. Specifically, pre-verification (based on a rules engine) quickly screens common violations, while in-process verification (based on a trained model) handles complex risk scenarios (such as abnormal cash flows) in real time, shortening verification response time. At the same time, the use of ETL tools to automatically collect real-time freight business data greatly improves data processing efficiency.

[0073] A post-optimization module, configured for the server to construct an incremental data set based on the real-time freight business data, and to perform post-optimization on the freight business risk identification model using the incremental data set; By constructing incremental data sets and optimizing the freight business risk identification model, post-closed-loop optimization is achieved. The freight business risk identification model can be automatically updated based on new data, enhancing its adaptability to changes in the business environment (such as new fraud methods). Combined with the defined loss function, it ensures continuous optimization of the freight business risk identification model and improves the long-term robustness of verification. This self-learning mechanism can resist "concept drift" (changes in data distribution), making the system more reliable and durable, different from static solutions.

[0074] Through the three stages of pre-event, in-event and post-event, an end-to-end risk management closed loop is established; pre-event rule verification provides basic screening, in-event model verification handles complex logic, and post-event optimization ensures continuous improvement. It not only covers the entire life cycle of freight business (such as order execution to financial settlement), but also supports real-time decision-making (such as in-event verification can automatically trigger risk alerts), effectively reducing overall risk exposure and improving business security.

[0075] In the freight business risk identification model creation module, the freight business risk identification model is constructed based on a feature extraction layer, a feature fusion layer, and a prediction output layer; The feature extraction layer is constructed based on the contract data processing module, the business data processing module, the trajectory data processing module, the capital data processing module and the bill data processing module; the contract data processing module is used to extract key semantic dependencies from the contract data through a bidirectional encoder constructed by Transformer to obtain contract risk features; the business data processing module is used to model the linear and nonlinear relationships of business indicators of the business data through a multi-layer first fully connected network to obtain business operation features; the trajectory data processing module is used to extract trajectory risk features from the trajectory data through a first gated loop unit; the capital data processing module is used to capture local transaction features from the capital data through a one-dimensional convolutional neural network, and capture the serial dependency features of funds from the capital data through a second gated loop unit, and output capital flow features based on the local transaction features and the serial dependency features; the bill data processing module is used to extract bill integrity features from the bill data through a multi-layer second fully connected network; The freight business risk identification model integrates multi-source heterogeneous data such as contract data, business data, trajectory data, financial data, and bill data through a feature extraction layer, enabling comprehensive monitoring of freight business risks. This avoids the limitations of a single data source and can capture a wider range of risk factors (such as contract breaches, financial anomalies, and trajectory deviations). By processing different types of data through dedicated modules (such as Transformer for text and CNN / GRU for sequences), the solution's practicality and market adaptability are enhanced, meeting the freight industry's needs for comprehensive risk assessment.

[0076] Contract data is processed through the Transformer bidirectional encoder to capture key semantic dependencies and improve the accuracy of text understanding; business and bill data are processed through a multi-layer fully connected network to model linear and nonlinear relationships and enhance the ability to parse complex business indicators; trajectory and capital data are processed through a combination of GRU and CNN, with GRU capturing temporal dependencies and CNN extracting local features (such as transaction patterns) to ensure efficient identification of dynamic risks; this hybrid architecture (combining Transformer, CNN, GRU, etc.) can efficiently process structured, unstructured and temporal data, reduce feature loss, and improve model robustness.

[0077] The feature fusion layer is constructed based on a multi-head attention module, a long short-term memory module, and a feature integration module; the multi-head attention module is used to calculate the feature weights of contract risk features, business operation features, trajectory risk features, capital flow features, and bill integrity features through multi-head self-attention units, and fuse the contract risk features, business operation features, trajectory risk features, capital flow features, and bill integrity features based on the feature weights to obtain fused features; the long short-term memory module is used to model the temporal dependency of the fused features through LSTM units to obtain serialized features; the feature integration module is used to perform global average pooling on the serialized features to obtain the whole image feature vector; The feature fusion layer dynamically calculates the feature weights of each feature through the multi-head attention module and performs weighted fusion. It then uses LSTM modeling to model temporal dependencies and global average pooling for dimensionality reduction, ensuring that key risk features are given priority while processing the time series characteristics of the data (such as capital flow trends) to improve the fusion effect. The combination of multi-head attention and LSTM realizes adaptive fusion of features, reduces redundant information, and improves computational efficiency.

[0078] The prediction output layer is constructed based on the risk item prediction branch, the risk level prediction branch and the joint output module; the risk item prediction branch is used to infer the feature vector of the entire graph through the first fully connected layer and the Sigmoid activation function to obtain the independent probabilities of different risk items (such as fraud, delay, default, etc.), supporting multi-label prediction; the risk level prediction branch is used to infer the feature vector of the entire graph through the second fully connected layer and the Softmax activation function to obtain the category probability of the risk level (low, medium, high); the joint output module is used to output a business risk identification report including risk items and risk levels based on the independent probabilities and category probabilities.

[0079] The prediction output layer adopts a dual-branch structure (risk item prediction and risk level prediction), and outputs the probability of independent risk items and risk levels respectively through Sigmoid and Softmax activation functions, and jointly outputs a comprehensive report. This allows the freight business risk identification model to handle multiple related tasks simultaneously (such as identifying specific risk items and assessing the overall risk level), improving the granularity and accuracy of the prediction; the multi-task learning framework reduces model complexity by sharing features (whole-graph feature vectors), and the loss function design (binary cross entropy and categorical cross entropy weighting) balances the weights of different tasks, optimizes the training process, and effectively improves the generalization ability of the freight business risk identification model, making the output business risk identification report more comprehensive.

[0080] In the freight business risk identification model creation module, the loss function is constructed based on the weighted sum of risk item loss and risk level loss; the risk item loss adopts binary cross entropy loss; the risk level loss adopts categorical cross entropy loss; The loss function is based on the weighted summation of risk term loss (binary cross entropy) and risk level loss (categorical cross entropy). This allows for dynamic adjustment of the priorities of different tasks during training (e.g., focusing more on high-risk levels or specific risk items), thereby improving the convergence speed and stability of the freight business risk identification model. This weighted loss mechanism is innovative and can effectively address imbalance issues in multi-task learning (e.g., the low frequency of occurrence of certain risk items), reduce the risk of overfitting, and enhance the reliability of the freight business risk identification model in real-world scenarios.

[0081] The dataset construction module is specifically used for: Obtain a large amount of historical freight business data including historical contract data, historical business data, historical trajectory data, historical capital data, and historical bill data; The historical contract data includes at least basic contract information, cooperation entity information, cargo information, transportation service requirements, fee clauses, liability for breach of contract, and dispute resolution methods; The basic contract information at least includes the contract number, signing date, and contract validity period, which are used to classify, search, and manage the contract to ensure the timeliness and traceability of the contract; the cooperative entity information at least includes the shipper name, carrier name, address, contact information, and business license number, which are used to clarify the identity and responsible parties of the contract parties and provide a basis for subsequent business development and dispute resolution; the cargo information at least includes the name, specifications, model, weight, volume, quantity, and value of the cargo, which are used to reasonably arrange transportation vehicles, calculate freight, and ensure cargo safety; the transportation service requirements at least include the departure place, destination, transportation time requirements (such as the latest pickup time, the latest delivery time), transportation mode (road, rail, The terms of the contract shall include at least the calculation method of the freight (such as by weight, volume or mileage), the payment method (such as prepaid, collect, monthly settlement, etc.), the payment time, and the invoice issuance requirements, so as to clarify the financial responsibilities and settlement process of both parties. The liability for breach of contract shall specify the corresponding liability for breach of contract and compensation for possible breach of contract that may occur during the performance of the contract, such as delayed delivery, damage or loss of goods, etc., to protect the legitimate rights and interests of both parties. The dispute resolution method shall stipulate the means of resolving contract disputes, such as negotiation, mediation, arbitration or litigation, as well as the applicable laws and arbitration institutions or competent courts, to provide a resolution mechanism for possible disputes.

[0082] The business data includes at least waybill information, order logs, online transaction logs, customer information, vehicle and driver information, and logistics information; The waybill information at least includes the waybill number, shipper information, consignee information, cargo description (consistent with the cargo information in the contract), transportation route, transportation mileage, estimated transportation time, actual transportation time, waybill status (such as ordered, picked up, in transit, delivered, etc.), which is used to track and manage the entire process of each transportation business; the order log is used to record the time, operator and reason for operations such as order generation, modification, and cancellation, so as to facilitate the tracing of order changes and ensure the transparency and auditability of business operations; the online transaction log at least includes records of users' online transaction behaviors such as logging in, browsing, querying, placing orders, quoting, and closing transactions on the online freight platform, which is used to reflect users' trading habits and the platform's trading activity, and provide a basis for data analysis and business optimization; the customer information, in addition to the consignment information involved in the contract, In addition to the information of people and carriers, it also includes the customer's registration information, credit rating, historical transaction records, preference settings, etc., which helps the online freight platform to better understand customer needs and provide personalized services, and is also conducive to customer relationship management and risk assessment; the vehicle and driver information at least includes license plate number, vehicle model, approved load capacity, vehicle status (such as idle, in transit, maintenance, etc.), driver's name, driver's contact information, driver's license information, professional qualification certificate information, driving experience, and credit record, which are used to reasonably dispatch vehicle and driver resources and ensure transportation safety; the logistics information at least includes the real-time location of the goods, transportation status (such as whether it is transported normally, whether there are any abnormal situations, etc.), loading and unloading time, and transit information. By combining with trajectory data, full monitoring and visual management of the logistics process can be achieved, thereby improving logistics efficiency and customer satisfaction.

[0083] The trajectory data includes at least time information, geographic location information, driving path information, speed information, mileage information and stop point information; The time information is the timestamp of the vehicle at each location point, including the departure time, the time of passing through each node, the time of arrival at the destination, etc. The vehicle's driving speed, stay time, etc. can be calculated through the time information, and then the transportation efficiency and whether there are abnormal stops can be analyzed; the geographical location information is the real-time latitude and longitude coordinates of the vehicle during transportation, which is obtained through positioning equipment (such as GPS, Beidou, etc.) to determine the precise location of the vehicle, draw the vehicle's driving trajectory, and realize dynamic tracking and monitoring of the transportation process; the driving path information is the actual route of the vehicle, including the road sections, intersections, highway entrances and exits passed by, etc., which can be compared with the planned transportation route to determine whether the vehicle is driving according to the planned route and whether there are abnormal situations such as detours and deviations; the speed information The information refers to the vehicle's driving speed in different time periods. By monitoring the speed, it is possible to analyze whether the vehicle is speeding and whether there are abnormal speed changes due to factors such as road conditions and traffic control, providing a reference for transportation safety and efficiency evaluation; the mileage information refers to the mileage of the vehicle during transportation, which can be used to calculate costs such as fuel consumption and vehicle wear and tear. At the same time, combined with information such as cargo weight, it can also evaluate transportation cost-effectiveness and provide data support for the optimization of transportation plans; the stop point information refers to the vehicle's stop location and stop duration during transportation, including loading and unloading locations, gas stations, rest areas, etc. By analyzing the stop points, the vehicle's operating mode and the driver's driving habits can be understood, and it is also helpful to discover potential risk points, such as the risk of cargo theft due to long-term stops.

[0084] The fund data includes at least fund flow information, account balance information and settlement information; The fund flow information includes the income and expenditure records of each fund, such as the collection of freight, freight paid to the actual carrier, collection of platform service fees, etc., and records in detail the flow of funds, amount, transaction time, transaction object and other information to ensure the transparency and traceability of funds; the account balance information is the balance of the shipper and carrier's funds account on the platform, reflecting the financial status and available funds of both parties, and providing a basis for fund settlement and financial management; the settlement information includes at least the settlement period, settlement method, settlement time, and settlement amount, which is used to clarify the settlement rules and procedures of funds, ensure timely and accurate settlement of funds, and safeguard the legitimate rights and interests of all parties.

[0085] The bill data at least includes invoice number, invoice date, invoice amount, and invoice type; The historical freight business data are preprocessed by at least removing duplicate data, processing missing values, correcting erroneous data, standardizing data, normalizing data, and encoding data; the preprocessed historical freight business data are labeled with risk items and risk levels; a data set is constructed using the labeled historical freight business data; and the sample size of the data set is expanded using an adversarial network.

[0086] By adopting multi-step preprocessing (including duplicate data removal, missing value processing, erroneous data correction, data standardization, data normalization and data encoding), the original data is systematically cleaned and transformed, reducing the impact of noise, inconsistency and bias, and ensuring the consistency and operability of the data set; for example, standardization and normalization processing makes data from different sources (such as speed information and amount information) comparable, thereby improving the accuracy of subsequent risk labeling and model training.

[0087] By integrating multi-dimensional historical freight data (including contract, business, trajectory, financial and bill data), and using systematic pre-processing (such as deduplication and standardization) to improve data quality, combining risk annotation to build a high-quality data set, and then innovatively applying adversarial networks to expand the sample size, the comprehensiveness, accuracy and robustness of freight risk identification have been significantly improved, effectively solving the prediction bias problem caused by single data or insufficient samples in traditional models, and at the same time achieving automated and efficient processing, providing scalable technical support for logistics risk management.

[0088] The freight business risk identification model training module is specifically used to: The data set is divided into a training set, a validation set, and a test set in a ratio of 8:1:1 by a stratified sampling method. The freight business risk identification model is trained using the training set. During the training process, the hyperparameters of the freight business risk identification model are continuously optimized until the loss value of the loss function is less than a preset loss threshold; The trained freight business risk identification model is verified using the validation set to determine whether the recognition accuracy is greater than a preset accuracy threshold. If not, the verification fails and the training set is expanded to continue training. If so, the verification passes, and: The verified freight business risk identification model is tested using the test set to determine whether the confidence level is greater than a preset confidence threshold. If not, the test fails, and the training set is expanded to continue training. If yes, the test passes, and the freight business risk identification model that has passed the test is deployed to the server using containerization technology. Using stratified sampling, the dataset was divided into training, validation, and test sets in a ratio of 8:1:1. This ensured balanced and representative data distribution, reduced sampling bias, and improved the model's generalization ability (i.e., its performance on unseen data), thereby reducing the misjudgment rate in freight business risk identification. The 8:1:1 division optimized resource utilization, avoided overfitting or underfitting problems, and enhanced the model's reliability.

[0089] During the training phase, by continuously optimizing hyperparameters (such as learning rate and batch size) until the loss value of the loss function is less than the preset threshold, the model achieves rapid convergence and efficient training, reducing training time and computing resource consumption, while ensuring that the model achieves high accuracy in freight risk identification tasks; the iterative optimization mechanism (such as dynamic adjustment based on loss value) improves the adaptability of training, enabling the model to better handle the noise and complexity in freight data.

[0090] By optimizing data set division through stratified sampling and combining iterative training with a dual-threshold verification mechanism, the accuracy, robustness, and reliability of the freight business risk identification model have been significantly improved. Containerization technology is used to achieve efficient deployment and elastic expansion, significantly reducing operation and maintenance costs. At the same time, through terminal authentication and structured storage of rule sets, the customizability of business rules and real-time decision-making capabilities are enhanced while ensuring system security. This has formed a closed-loop optimization system from data preprocessing to model deployment, effectively solving the core pain points of low risk identification efficiency, high misjudgment rate, and poor system scalability in freight scenarios.

[0091] The business verification rule set setting module is specifically used to: After authenticating the accessing mobile terminal, the server obtains several verification rules set by the mobile terminal, constructs a business verification rule set based on each of the verification rules, and stores the business verification rule set in a structured manner to a designated path.

[0092] The multidimensional data acquisition module is specifically used for: The server uses an ETL tool to extract real-time freight business data, including real-time contract data, real-time business data, real-time track data, real-time fund data, and real-time bill data, from the contract system, business system, track system, fund system, and bill system based on a preset extraction cycle. The server performs data conversion operations on each of the real-time freight business data, including at least data cleansing, data formatting, and data aggregation, and loads the real-time freight business data after the data conversion operations into a preset data warehouse. The prior verification module is specifically used to: Based on a freight business trigger event, the server calls a rule engine through an API interface. The rule engine reads the business verification rule set from a specified path, and uses the business verification rule set to perform pre-verification of real-time freight business data, including at least data format, data integrity, logical consistency, vehicle qualifications, personnel qualifications, vehicle trajectory, waybill timeliness, and invoice information, and generates a pre-verification report containing pre-verification success or event verification failure information; When the pre-verification report indicates that the pre-verification fails, the pre-verification report is pushed to the pre-associated management terminal in real time via the TLS protocol; The server obtains the pre-verification time, calculates the pre-verification report and the first hash value of the pre-verification time, encrypts the pre-verification report, pre-verification time and the first hash value into first-level encrypted data through the SM9 algorithm, divides the first-level encrypted data into a ratio of 7:3 and swaps the order to obtain second-level encrypted data, encrypts the second-level encrypted data into third-level encrypted data through the RC6 algorithm, shifts each character of the third-level encrypted data 5 bits to the right to obtain a pre-encrypted report, and uploads the pre-encrypted report to the blockchain.

[0093] The rule engine is automatically called through the API interface to trigger the processing of freight business events in real time, avoiding the delay of traditional manual verification; the rule engine reads the verification rule set from the specified path and performs multi-dimensional checks on the data (such as data format, integrity, logical consistency, vehicle qualifications, personnel qualifications, vehicle trajectory, waybill timeliness and invoice information), ensuring that the verification process is efficient and automated, significantly reducing business interruption time and improving the overall throughput of the freight process (for example, reports can be quickly generated in the event of failure), thereby improving operational efficiency and reducing labor costs.

[0094] A multi-layer encryption mechanism (including the SM9 algorithm, RC6 algorithm, character shifting, and splitting and transposing) provides enhanced protection for data. Specifically, hash values ​​are used to ensure data integrity and prevent verification reports and time from being tampered with. The SM9 algorithm provides post-quantum-level asymmetric encryption, and the first-level encrypted data ensures the confidentiality of sensitive information. Splitting and transposing (splitting in a 7:3 ratio and transposing the order) is used to increase the difficulty of cracking. The RC6 algorithm further encrypts and introduces disturbances by shifting the characters right by 5 bits, making the encryption process more complex. This multi-step process significantly improves the data's resistance to attacks and is suitable for freight data environments with high security requirements. Finally, it is uploaded to the blockchain, leveraging its distributed ledger characteristics to ensure that the data cannot be tampered with and is traceable, enhancing the non-repudiation of the entire system.

[0095] The in-process verification module is specifically used to: The server inputs the real-time freight business data into a deployed freight business risk identification model. The freight business risk identification model performs inference using hardware acceleration technology to obtain a real-time business risk identification report including real-time risk items and real-time risk levels, thereby performing in-process verification of the real-time freight business data. When the risk level is higher than a preset level, the real-time business risk identification report is pushed to a pre-associated management terminal in real time via the TLS protocol; The server obtains the in-process verification time, calculates the real-time business risk identification report and the second hash value of the in-process verification time, encrypts the real-time business risk identification report, the in-process verification time, and the second hash value into a first-layer encrypted data using the XTEA algorithm, converts the first-layer encrypted data into hexadecimal data, swaps the number 7 with the letter B, the number 8 with the number 9, and the number 10 with the letter A in the hexadecimal data to obtain a second-layer encrypted data, encrypts the second-layer encrypted data into a third-layer encrypted data using the IDEA algorithm, adds a random string of a specified length to a specified position in the third-layer encrypted data to obtain an in-process encrypted report, and uploads the in-process encrypted report to the blockchain; By using hardware acceleration technology (such as GPU or FPGA) to infer freight business risk identification models, the computational delay of real-time data processing is significantly reduced, enabling in-process verification and avoiding the time lag of traditional batch processing. With high response speed (millisecond-level inference), high-risk freight incidents (such as cargo damage or fraud) can be identified and handled promptly, thereby reducing business losses. This real-time performance enhances the practicality of the system, making it particularly suitable for high-frequency, dynamic freight scenarios.

[0096] The post-optimization module is specifically used for: The server constructs an incremental dataset based on the real-time freight business data, labels the incremental dataset with real-time risk items, real-time risk levels, and actual risk items and levels. When the amount of data in the incremental dataset exceeds a preset threshold, the server performs post-optimization on the freight business risk identification model using the incremental dataset. The real-time risk items and real-time risk levels are predicted data, while the actual risk items and levels are actual data.

[0097] Although the specific embodiments of the present invention are described above, those skilled in the art should understand that the specific embodiments described are merely illustrative and are not intended to limit the scope of the present invention. Equivalent modifications and changes made by those skilled in the art in accordance with the spirit of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A freight business verification method integrating multi-dimensional data, characterized by: The steps include: Step S1: Create a freight business risk identification model and set a loss function of the freight business risk identification model; Step S2: obtaining a large amount of historical freight business data including historical contract data, historical business data, historical trajectory data, historical capital data, and historical bill data, and constructing a data set after preprocessing and annotating each of the historical freight business data; Step S3: training a freight business risk identification model using the data set and the loss function, and deploying the trained freight business risk identification model to a server; Step S4: The server sets a business verification rule set including several verification rules; Step S5: The server uses ETL tools to obtain real-time freight business data, including real-time contract data, real-time business data, real-time track data, real-time fund data, and real-time bill data, from the contract system, business system, track system, fund system, and bill system. Step S6: The server calls the business verification rule set through a rule engine to perform a pre-verification on the real-time freight business data; Step S7: The server verifies the real-time freight business data through the deployed freight business risk identification model; Step S8: The server constructs an incremental data set based on the real-time freight business data, and performs post-optimization on the freight business risk identification model through the incremental data set.

2. The freight business verification method integrating multi-dimensional data according to claim 1, characterized in that: In step S1, the freight business risk identification model is constructed based on the feature extraction layer, the feature fusion layer and the prediction output layer; The feature extraction layer is constructed based on the contract data processing module, the business data processing module, the trajectory data processing module, the capital data processing module and the bill data processing module; the contract data processing module is used to extract key semantic dependencies from the contract data through a bidirectional encoder constructed by Transformer to obtain contract risk features; the business data processing module is used to model the linear and nonlinear relationships of business indicators of the business data through a multi-layer first fully connected network to obtain business operation features; the trajectory data processing module is used to extract trajectory risk features from the trajectory data through a first gated loop unit; the capital data processing module is used to capture local transaction features from the capital data through a one-dimensional convolutional neural network, and capture the serial dependency features of funds from the capital data through a second gated loop unit, and output capital flow features based on the local transaction features and the serial dependency features; the bill data processing module is used to extract bill integrity features from the bill data through a multi-layer second fully connected network; The feature fusion layer is constructed based on a multi-head attention module, a long short-term memory module, and a feature integration module; the multi-head attention module is used to calculate the feature weights of contract risk features, business operation features, trajectory risk features, capital flow features, and bill integrity features through multi-head self-attention units, and fuse the contract risk features, business operation features, trajectory risk features, capital flow features, and bill integrity features based on the feature weights to obtain a fused feature; The long short-term memory module is used to model the temporal dependency of the fusion features through the LSTM unit to obtain the serialized features; the feature integration module is used to perform global average pooling on the serialized features to obtain the whole image feature vector; The prediction output layer is constructed based on the risk item prediction branch, the risk level prediction branch, and the joint output module; the risk item prediction branch is used to infer the entire image feature vector through the first fully connected layer and the Sigmoid activation function to obtain the independent probabilities of different risk items; the risk level prediction branch is used to infer the entire image feature vector through the second fully connected layer and the Softmax activation function to obtain the category probability of the risk level; the joint output module is used to output a business risk identification report including risk items and risk levels based on the independent probabilities and category probabilities; The loss function is constructed based on the weighted sum of risk item loss and risk level loss; the risk item loss adopts binary cross entropy loss; the risk level loss adopts categorical cross entropy loss; The step S2 is specifically as follows: Obtain a large amount of historical freight business data including historical contract data, historical business data, historical trajectory data, historical capital data, and historical bill data; The historical contract data includes at least basic contract information, cooperation entity information, cargo information, transportation service requirements, fee clauses, liability for breach of contract, and dispute resolution methods; The business data includes at least waybill information, order logs, online transaction logs, customer information, vehicle and driver information, and logistics information; The trajectory data includes at least time information, geographic location information, driving path information, speed information, mileage information and stop point information; The fund data includes at least fund flow information, account balance information and settlement information; The bill data at least includes invoice number, invoice date, invoice amount, and invoice type; The historical freight business data are preprocessed by at least removing duplicate data, processing missing values, correcting erroneous data, standardizing data, normalizing data, and encoding data; the preprocessed historical freight business data are labeled with risk items and risk levels; a data set is constructed using the labeled historical freight business data; and the sample size of the data set is expanded using an adversarial network.

3. The freight business verification method integrating multi-dimensional data according to claim 1, characterized in that: The step S3 is specifically as follows: The data set is divided into a training set, a validation set, and a test set in a ratio of 8:1:1 by a stratified sampling method. The freight business risk identification model is trained using the training set. During the training process, the hyperparameters of the freight business risk identification model are continuously optimized until the loss value of the loss function is less than a preset loss threshold; The trained freight business risk identification model is verified using the validation set to determine whether the identification accuracy is greater than a preset accuracy threshold. If not, the verification fails, and the training set is expanded to continue training. If yes, the verification passes, and: The verified freight business risk identification model is tested using the test set to determine whether the confidence level is greater than a preset confidence threshold. If not, the test fails, and the training set is expanded to continue training. If yes, the test passes, and the freight business risk identification model that has passed the test is deployed to the server using containerization technology. The step S4 is specifically as follows: After authenticating the accessing mobile terminal, the server obtains several verification rules set by the mobile terminal, constructs a business verification rule set based on each of the verification rules, and stores the business verification rule set in a structured manner to a designated path.

4. The method for verifying freight business by integrating multi-dimensional data according to claim 1, characterized in that: The step S5 is specifically as follows: The server uses an ETL tool to extract real-time freight business data, including real-time contract data, real-time business data, real-time track data, real-time fund data, and real-time bill data, from the contract system, business system, track system, fund system, and bill system based on a preset extraction cycle. The server performs data conversion operations on each of the real-time freight business data, including at least data cleansing, data formatting, and data aggregation, and loads the real-time freight business data after the data conversion operations into a preset data warehouse. The step S6 is specifically as follows: Based on a freight business trigger event, the server calls a rule engine through an API interface. The rule engine reads the business verification rule set from a specified path, and uses the business verification rule set to perform pre-verification of real-time freight business data, including at least data format, data integrity, logical consistency, vehicle qualifications, personnel qualifications, vehicle trajectory, waybill timeliness, and invoice information, and generates a pre-verification report containing pre-verification success or event verification failure information; When the pre-verification report indicates that the pre-verification fails, the pre-verification report is pushed to the pre-associated management terminal in real time via the TLS protocol; The server obtains the pre-verification time, calculates the pre-verification report and the first hash value of the pre-verification time, encrypts the pre-verification report, pre-verification time and the first hash value into first-level encrypted data through the SM9 algorithm, divides the first-level encrypted data into a ratio of 7:3 and swaps the order to obtain second-level encrypted data, encrypts the second-level encrypted data into third-level encrypted data through the RC6 algorithm, shifts each character of the third-level encrypted data 5 bits to the right to obtain a pre-encrypted report, and uploads the pre-encrypted report to the blockchain.

5. The freight business verification method integrating multi-dimensional data according to claim 1, characterized in that: The step S7 is specifically as follows: The server inputs the real-time freight business data into a deployed freight business risk identification model. The freight business risk identification model performs inference using hardware acceleration technology to obtain a real-time business risk identification report including real-time risk items and real-time risk levels, thereby performing in-process verification of the real-time freight business data. When the risk level is higher than a preset level, the real-time business risk identification report is pushed to a pre-associated management terminal in real time via the TLS protocol; The server obtains the in-process verification time, calculates the real-time business risk identification report and the second hash value of the in-process verification time, encrypts the real-time business risk identification report, the in-process verification time, and the second hash value into a first-layer encrypted data using the XTEA algorithm, converts the first-layer encrypted data into hexadecimal data, swaps the number 7 with the letter B, the number 8 with the number 9, and the number 10 with the letter A in the hexadecimal data to obtain a second-layer encrypted data, encrypts the second-layer encrypted data into a third-layer encrypted data using the IDEA algorithm, adds a random string of a specified length to a specified position in the third-layer encrypted data to obtain an in-process encrypted report, and uploads the in-process encrypted report to the blockchain; The step S8 is specifically as follows: The server constructs an incremental data set based on the real-time freight business data, and labels the incremental data set with real-time risk items, real-time risk levels, real risk items, and real risk levels. When the data volume of the incremental data set exceeds a preset quantity threshold, the freight business risk identification model is post-optimized through the incremental data set.

6. A freight business verification system integrating multi-dimensional data, characterized by: Includes the following modules: A freight business risk identification model creation module is used to create a freight business risk identification model and set a loss function of the freight business risk identification model; A data set construction module is used to obtain a large amount of historical freight business data including historical contract data, historical business data, historical trajectory data, historical capital data, and historical bill data, and to construct a data set after preprocessing and annotating each of the historical freight business data; A freight business risk identification model training module, configured to train the freight business risk identification model using the data set and the loss function, and deploy the trained freight business risk identification model to a server; The business verification rule set setting module is used for the server to set a business verification rule set containing several verification rules; The multi-dimensional data acquisition module is used by the server to obtain real-time freight business data including real-time contract data, real-time business data, real-time trajectory data, real-time capital data and real-time bill data from the contract system, business system, tracking system, capital system and billing system through ETL tools; A pre-verification module is used for the server to call the business verification rule set through a rule engine to perform pre-verification on the real-time freight business data; An in-process verification module, configured to verify the real-time freight business data in-process by using the freight business risk identification model deployed on the server; A post-optimization module is used for the server to construct an incremental data set based on the real-time freight business data, and to perform post-optimization on the freight business risk identification model through the incremental data set.

7. The freight business verification system integrating multi-dimensional data according to claim 6, characterized in that: In the freight business risk identification model creation module, the freight business risk identification model is constructed based on a feature extraction layer, a feature fusion layer, and a prediction output layer; The feature extraction layer is constructed based on the contract data processing module, the business data processing module, the trajectory data processing module, the capital data processing module and the bill data processing module; the contract data processing module is used to extract key semantic dependencies from the contract data through a bidirectional encoder constructed by Transformer to obtain contract risk features; the business data processing module is used to model the linear and nonlinear relationships of business indicators of the business data through a multi-layer first fully connected network to obtain business operation features; the trajectory data processing module is used to extract trajectory risk features from the trajectory data through a first gated loop unit; the capital data processing module is used to capture local transaction features from the capital data through a one-dimensional convolutional neural network, and capture the serial dependency features of funds from the capital data through a second gated loop unit, and output capital flow features based on the local transaction features and the serial dependency features; the bill data processing module is used to extract bill integrity features from the bill data through a multi-layer second fully connected network; The feature fusion layer is constructed based on a multi-head attention module, a long short-term memory module, and a feature integration module; the multi-head attention module is used to calculate the feature weights of contract risk features, business operation features, trajectory risk features, capital flow features, and bill integrity features through multi-head self-attention units, and fuse the contract risk features, business operation features, trajectory risk features, capital flow features, and bill integrity features based on the feature weights to obtain a fused feature; The long short-term memory module is used to model the temporal dependency of the fusion features through the LSTM unit to obtain the serialized features; the feature integration module is used to perform global average pooling on the serialized features to obtain the whole image feature vector; The prediction output layer is constructed based on the risk item prediction branch, the risk level prediction branch, and the joint output module; the risk item prediction branch is used to infer the entire image feature vector through the first fully connected layer and the Sigmoid activation function to obtain the independent probabilities of different risk items; the risk level prediction branch is used to infer the entire image feature vector through the second fully connected layer and the Softmax activation function to obtain the category probability of the risk level; the joint output module is used to output a business risk identification report including risk items and risk levels based on the independent probabilities and category probabilities; The loss function is constructed based on the weighted sum of risk item loss and risk level loss; the risk item loss adopts binary cross entropy loss; the risk level loss adopts categorical cross entropy loss; The dataset construction module is specifically used for: Obtain a large amount of historical freight business data including historical contract data, historical business data, historical trajectory data, historical capital data, and historical bill data; The historical contract data includes at least basic contract information, cooperation entity information, cargo information, transportation service requirements, fee clauses, liability for breach of contract, and dispute resolution methods; The business data includes at least waybill information, order logs, online transaction logs, customer information, vehicle and driver information, and logistics information; The trajectory data includes at least time information, geographic location information, driving path information, speed information, mileage information and stop point information; The fund data includes at least fund flow information, account balance information and settlement information; The bill data at least includes invoice number, invoice date, invoice amount, and invoice type; The historical freight business data are preprocessed by at least removing duplicate data, processing missing values, correcting erroneous data, standardizing data, normalizing data, and encoding data; the preprocessed historical freight business data are labeled with risk items and risk levels; a data set is constructed using the labeled historical freight business data; and the sample size of the data set is expanded using an adversarial network.

8. The freight business verification system integrating multi-dimensional data according to claim 6, characterized in that: The freight business risk identification model training module is specifically used to: The data set is divided into a training set, a validation set, and a test set in a ratio of 8:1:1 by a stratified sampling method. The freight business risk identification model is trained using the training set. During the training process, the hyperparameters of the freight business risk identification model are continuously optimized until the loss value of the loss function is less than a preset loss threshold; The trained freight business risk identification model is verified using the validation set to determine whether the identification accuracy is greater than a preset accuracy threshold. If not, the verification fails, and the training set is expanded to continue training. If yes, the verification passes, and: The verified freight business risk identification model is tested using the test set to determine whether the confidence level is greater than a preset confidence threshold. If not, the test fails, and the training set is expanded to continue training. If yes, the test passes, and the freight business risk identification model that passes the test is deployed to the server using containerization technology; The business verification rule set setting module is specifically used to: After authenticating the accessing mobile terminal, the server obtains several verification rules set by the mobile terminal, constructs a business verification rule set based on each of the verification rules, and stores the business verification rule set in a structured manner to a designated path.

9. The freight business verification system integrating multi-dimensional data according to claim 6, characterized in that: The multidimensional data acquisition module is specifically used for: The server uses an ETL tool to extract real-time freight business data, including real-time contract data, real-time business data, real-time track data, real-time fund data, and real-time bill data, from the contract system, business system, track system, fund system, and bill system based on a preset extraction cycle. The server performs data conversion operations on each of the real-time freight business data, including at least data cleansing, data formatting, and data aggregation, and loads the real-time freight business data after the data conversion operations into a preset data warehouse. The prior verification module is specifically used to: Based on a freight business trigger event, the server calls a rule engine through an API interface. The rule engine reads the business verification rule set from a specified path, and uses the business verification rule set to perform pre-verification of real-time freight business data, including at least data format, data integrity, logical consistency, vehicle qualifications, personnel qualifications, vehicle trajectory, waybill timeliness, and invoice information, and generates a pre-verification report containing pre-verification success or event verification failure information; When the pre-verification report indicates that the pre-verification fails, the pre-verification report is pushed to the pre-associated management terminal in real time via the TLS protocol; The server obtains the pre-verification time, calculates the pre-verification report and the first hash value of the pre-verification time, encrypts the pre-verification report, pre-verification time and the first hash value into first-level encrypted data through the SM9 algorithm, divides the first-level encrypted data into a ratio of 7:3 and swaps the order to obtain second-level encrypted data, encrypts the second-level encrypted data into third-level encrypted data through the RC6 algorithm, shifts each character of the third-level encrypted data 5 bits to the right to obtain a pre-encrypted report, and uploads the pre-encrypted report to the blockchain.

10. The freight business verification system integrating multi-dimensional data according to claim 6, characterized in that: The in-process verification module is specifically used to: The server inputs the real-time freight business data into a deployed freight business risk identification model. The freight business risk identification model performs inference using hardware acceleration technology to obtain a real-time business risk identification report including real-time risk items and real-time risk levels, thereby performing in-process verification of the real-time freight business data. When the risk level is higher than a preset level, the real-time business risk identification report is pushed to a pre-associated management terminal in real time via the TLS protocol; The server obtains the in-process verification time, calculates the real-time business risk identification report and the second hash value of the in-process verification time, encrypts the real-time business risk identification report, the in-process verification time, and the second hash value into a first-layer encrypted data using the XTEA algorithm, converts the first-layer encrypted data into hexadecimal data, swaps the number 7 with the letter B, the number 8 with the number 9, and the number 10 with the letter A in the hexadecimal data to obtain a second-layer encrypted data, encrypts the second-layer encrypted data into a third-layer encrypted data using the IDEA algorithm, adds a random string of a specified length to a specified position in the third-layer encrypted data to obtain an in-process encrypted report, and uploads the in-process encrypted report to the blockchain; The post-optimization module is specifically used for: The server constructs an incremental data set based on the real-time freight business data, and labels the incremental data set with real-time risk items, real-time risk levels, real risk items, and real risk levels. When the data volume of the incremental data set exceeds a preset quantity threshold, the freight business risk identification model is post-optimized through the incremental data set.

Images