Cross-border logistics information visualization management method and system based on big data

Abstract

The application discloses a cross-border logistics information visual management method and system based on big data, and relates to the technical field of cross-border logistics information management. The cross-border logistics information visual management method based on big data acquires customs declaration information and transshipment operation data of a plurality of cross-border logistics tasks, analyzes task execution risk characteristic values of each cross-border logistics task based on a pre-trained natural language recognition model and in combination with the customs declaration information of each cross-border logistics task, analyzes task safety and completeness characteristic values of each cross-border logistics task based on the transshipment operation data of each cross-border logistics task, and performs visual mapping processing in combination with the task execution risk characteristic values of each cross-border logistics task. The application performs joint management on corresponding cross-border logistics tasks through visual mapping processing results, thereby realizing fine assessment at the task level and greatly improving the fine and intelligent level of cross-border logistics task management and control.

Description

Technical Field

[0001] This invention relates to the field of cross-border logistics information management technology, specifically to a method and system for visual management of cross-border logistics information based on big data. Background Technology

[0002] With the rapid development of cross-border e-commerce and global trade, the number of cross-border logistics tasks continues to grow. Its business process covers multiple links such as customs declaration, transit operations, transportation scheduling, transshipment and customs clearance. In this process, a large amount of customs declaration information and operational data are generated in real time. Against the backdrop of the digital transformation of cross-border logistics, big data technology has been gradually applied to the monitoring and management of the entire logistics chain. By integrating and analyzing customs declaration information and transit operation data, refined management of cross-border logistics tasks can be achieved. At the same time, the development of visualization technology allows complex multi-dimensional task characteristics to be presented in an intuitive way, providing an intuitive expression for cross-border logistics tasks.

[0003] Existing technology, such as the cross-border logistics information visualization management system and method disclosed in patent application CN118569759B, belongs to the field of cross-border logistics information management. This big data-based cross-border logistics information visualization management system includes: a logistics information collection module, a visualization design module, a visualization effect analysis module, and a management quality optimization module. This invention analyzes the logistics information management quality index, and then optimizes the visualization design module based on this index. It notifies managers to regularly maintain the visualization design module of the cross-border logistics information visualization management system, thereby improving the quality of cross-border logistics information visualization management and solving the problem of low quality in existing cross-border logistics information management technologies.

[0004] Based on the above findings, the limitations of existing technologies include at least the following issues: Existing technologies lack the ability to systematically and quantitatively assess and jointly manage the risks and safety integrity of task execution. Specifically, they fail to perform intelligent semantic recognition and risk feature extraction on customs declaration information, resulting in risk assessment still relying on human experience, with insufficient accuracy and real-time performance, and difficulty in reflecting the safety integrity level of the operational process. Furthermore, they fail to introduce cargo damage detection methods based on machine learning and image recognition, making it difficult to promptly detect problems such as packaging damage, surface contamination, or deformation in the logistics process. Moreover, the visualization display of existing technologies remains static, lacking a joint mapping and zoning management mechanism for task execution risk feature values ​​and task safety integrity feature values, thus making it difficult to provide refined risk classification and decision support for cross-border logistics tasks. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and system for visualized management of cross-border logistics information based on big data, which solves the problem that existing technologies lack quantitative assessment of risks and completeness, making it difficult to achieve refined management.

[0006] To achieve the above objectives, this invention provides the following technical solution: a big data-based method for visual management of cross-border logistics information, comprising the following steps: acquiring customs declaration information and transit operation data for several cross-border logistics tasks; analyzing the task execution risk characteristic value of each cross-border logistics task based on a pre-trained natural language processing model and in conjunction with the customs declaration information of each cross-border logistics task; analyzing the task security and completeness characteristic value of each cross-border logistics task based on the transit operation data of each cross-border logistics task; performing visual mapping processing on the task execution risk characteristic value and task security and completeness characteristic value of each cross-border logistics task; and jointly managing the corresponding cross-border logistics tasks based on the results of the visual mapping processing.

[0007] Furthermore, the specific steps for analyzing the task execution risk characteristic value of each cross-border logistics task are as follows: input the customs declaration information of each cross-border logistics task into the pre-trained natural language processing model, analyze the cross-border customs assessment characteristic set of each cross-border logistics task, including declaration compliance characteristic value and customs operation complexity characteristic value; based on the cross-border customs assessment characteristic set of each cross-border logistics task, analyze the task execution risk characteristic value of each cross-border logistics task.

[0008] Furthermore, the natural language processing model includes an input layer, a semantic recognition layer, a feature extraction layer, and an output layer.

[0009] Furthermore, the specific steps for analyzing the cross-border customs assessment feature set of each cross-border logistics task are as follows: In the input layer of the natural language processing model, the customs declaration information of each cross-border logistics task is received and preprocessed; in the semantic recognition layer of the natural language processing model, the preprocessed customs declaration information of each cross-border logistics task is semantically processed; in the feature extraction layer of the natural language processing model, the cross-border customs feature vector of each cross-border logistics task is extracted based on the semantically processed customs declaration information; in the output layer of the natural language processing model, the cross-border customs assessment feature set of each cross-border logistics task is output based on the cross-border customs feature vector of each cross-border logistics task.

[0010] Furthermore, the transit operation data includes transit difference rate, transshipment load ratio, transshipment completion rate, and transit operation image data. The specific steps for analyzing the task safety and completeness feature value of each cross-border logistics task are as follows: Based on the transit difference rate, transshipment load ratio, and transshipment completion rate of each cross-border logistics task, analyze the cross-border logistics operation completion feature value of each cross-border logistics task; Based on the pre-trained cargo damage recognition model and combined with the transit operation image data of each cross-border logistics task, analyze the cargo damage feature value of each cross-border logistics task; Based on the cross-border logistics operation completion feature value and cargo damage feature value of each cross-border logistics task, analyze the task safety and completeness feature value of each cross-border logistics task.

[0011] Further, the specific steps for analyzing the cargo damage feature values ​​of each cross-border logistics task are as follows: input the transit operation image data of each cross-border logistics task into the pre-trained cargo damage recognition model, analyze the transit damage feature set of each cross-border logistics task, including the feature values ​​of protective film integrity, surface contamination coverage, and packaging deformation; based on the transit damage feature set of each cross-border logistics task, analyze the cargo damage feature values ​​of each cross-border logistics task.

[0012] Furthermore, the customs clearance operation image data specifically includes the pixel value and two-dimensional coordinates of each pixel in the customs clearance operation image, and the cargo damage recognition model includes a transfer input layer, a damage extraction layer, and a damage output layer.

[0013] Furthermore, the specific steps for analyzing the transit damage feature set of each cross-border logistics task are as follows: In the transshipment input layer of the cargo damage identification model, the image data of the customs clearance operation for each cross-border logistics task is received and preprocessed. In the damage extraction layer of the cargo damage identification model, the damage feature vector of each cross-border logistics task is extracted based on the preprocessed image data of the customs clearance operation for each cross-border logistics task. In the damage output layer of the cargo damage identification model, the customs clearance damage feature set of each cross-border logistics task is output based on the damage feature vector of each cross-border logistics task.

[0014] Furthermore, the specific steps of the visualization mapping process are as follows: normalize the task execution risk feature value and task safety and completeness feature value of each cross-border logistics task, and construct a two-dimensional mapping space for task features; based on the normalized task execution risk feature value and task safety and completeness feature value, map the corresponding cross-border logistics task to the two-dimensional mapping space for task features.

[0015] The big data-based cross-border logistics information visualization management system includes: a data acquisition module for acquiring customs declaration information and transit operation data for several cross-border logistics tasks; an execution risk analysis module for analyzing the task execution risk characteristics of each cross-border logistics task based on a pre-trained natural language processing model and the customs declaration information of each task; a transit operation analysis module for analyzing the task safety and completeness characteristics of each cross-border logistics task based on the transit operation data; a visualization mapping module for visually mapping the task execution risk characteristics and task safety and completeness characteristics of each cross-border logistics task; and a joint management feedback module for jointly managing the corresponding cross-border logistics tasks based on the visualization mapping results.

[0016] The present invention has the following beneficial effects: (1) This big data-based cross-border logistics information visualization management method introduces a pre-trained natural language processing model to perform semantic analysis and feature extraction on customs declaration information of cross-border logistics tasks. It can quantify the risk feature value of task execution from two dimensions: declaration compliance and customs operation complexity. At the same time, it combines the data analysis of transit operations to analyze the safety and completeness feature value of the task, thereby realizing the accurate characterization of cross-border logistics tasks at the execution level. Combined with visualization mapping processing, the task is partitioned and displayed in a two-dimensional feature space, making the risk level and completeness status clear at a glance. This achieves refined assessment at the task level and greatly improves the refinement and intelligence level of cross-border logistics task management.

[0017] (2) The big data-based cross-border logistics information visualization management method uses a pre-trained cargo damage identification model to conduct in-depth analysis of transit operation image data, thereby forming cargo damage feature values. These feature values, together with the cross-border logistics operation completion feature values, constitute task safety and completeness feature values. These feature values ​​can not only comprehensively reflect the completeness level of cross-border logistics tasks during execution, but also dynamically reveal potential safety hazards. In particular, when cargo damage just occurs, it can be identified and quantified by the model, thereby improving the real-time and accuracy of problem detection. This can effectively reduce the risks of claims and returns caused by damage in the cross-border transit process, thereby ensuring that cross-border logistics tasks have higher safety and quality assurance capabilities.

[0018] (3) This big data-based cross-border logistics information visualization management method normalizes the task execution risk characteristic value and the task safety and integrity characteristic value, and maps each cross-border logistics task to a unified two-dimensional feature space, thereby realizing the partitioned presentation and visualization of the task status. This allows for a clear distinction between the differences in risk level and safety and integrity of different tasks, enabling a quick grasp of the overall distribution of tasks in a unified interface. It also helps relevant personnel to achieve overall situational awareness of task groups at the macro level and focus on individual tasks at the micro level, thereby making the hierarchical management of cross-border logistics tasks more efficient and intuitive, thus comprehensively improving the transparency and management accuracy of cross-border logistics operations.

[0019] (4) This big data-based cross-border logistics information visualization management system, through modular integrated analysis, not only makes the system architecture clearer and more scalable, but also ensures smooth data flow and unified processing logic between various functional links. The data acquisition module realizes the automated acquisition of customs declaration information and transit operation data, avoiding redundant manual input. The execution risk analysis module and the transit operation analysis module respectively complete feature extraction and quantitative analysis in different dimensions. The visualization mapping module further transforms the analysis results into intuitive displays. The joint management feedback module provides task-level management strategies based on the mapping results. Through this modular collaborative operation, the system can not only ensure the real-time and efficient processing of cross-border logistics tasks, but also has good flexibility, thereby improving the stability and reliability of the entire cross-border logistics process management.

[0020] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0021] Figure 1 This is a flowchart of the cross-border logistics information visualization management method based on big data, as described in this invention.

[0022] Figure 2 This is a flowchart illustrating the specific steps involved in analyzing the task execution risk characteristic values ​​of each cross-border logistics task in the big data-based cross-border logistics information visualization management method of this invention.

[0023] Figure 3 This is a schematic diagram of the cross-border logistics task sequence data in the transit damage feature set of the big data-based cross-border logistics information visualization management method of the present invention.

[0024] Figure 4 This is a block diagram of the cross-border logistics information visualization management system based on big data, as described in this invention. Detailed Implementation

[0025] Please see Figure 1This invention provides a technical solution: a big data-based method for visual management of cross-border logistics information, comprising the following steps: acquiring customs declaration information (including but not limited to: commodity name, declared amount, declared commodity quantity, HS code, tariff rate, customs declaration number, consignee information, invoice information, etc.) and transit operation data (which can be obtained based on a transportation management mini-program) for several cross-border logistics tasks (based on BTDS transit and direct connection with customs declaration enterprises QP); analyzing the task execution risk characteristic value of each cross-border logistics task based on a pre-trained natural language processing model and combined with the customs declaration information of each cross-border logistics task; analyzing the task security and completeness characteristic value of each cross-border logistics task based on the transit operation data of each cross-border logistics task; performing visual mapping processing on the task execution risk characteristic value and task security and completeness characteristic value of each cross-border logistics task; and jointly managing the corresponding cross-border logistics tasks based on the results of the visual mapping processing, specifically as follows: For cross-border logistics tasks mapped to Region A (low risk, high completeness), the task is classified as a routine control level, marked as normal execution, without triggering additional intervention, and only retains routine tracking logs. For cross-border logistics tasks mapped to Region B (high risk, high completeness), the task is classified as an early warning control level, marked as a risk warning. In the visualization interface, the task mapping point is displayed as a yellow-to-red gradient, and a prompt icon is automatically generated on the risk monitoring panel. The risk control department is notified, prompting them to strengthen spot checks or manual review during the execution process. For cross-border logistics tasks mapped to Region C (low risk, low completeness), the task is classified as a completeness and supplementation level. The system marks the task as complete; automatically generates a completion work order; distributes the work order to the declarant or carrier, prompting them to upload missing customs declaration materials or supplement transit operation data. After completion, the system recalculates the task's safety completeness characteristic value and updates its position in the two-dimensional mapping space; for cross-border logistics tasks mapped to region D (high risk, low completeness), the system determines them to be at the key control level and marks the task as key control; immediately triggers the cross-departmental joint management mechanism, generates a linkage work order, and automatically pushes it to the customs, risk control, and transportation management parties; records the entire process log of the task and enters closed-loop tracking until the task is completed or intercepted.

[0026] Specifically, such as Figure 2As shown below, the specific steps for analyzing the task execution risk eigenvalue of each cross-border logistics task are as follows: Input the customs declaration information of each cross-border logistics task into a pre-trained natural language processing model to analyze the cross-border customs evaluation feature set of each cross-border logistics task, including the declaration compliance eigenvalue and the customs operation complexity eigenvalue; Based on the cross-border customs evaluation feature set of each cross-border logistics task, analyze the task execution risk eigenvalue of each cross-border logistics task, specifically: perform weighted processing on the declaration compliance eigenvalue and the customs operation complexity eigenvalue of each cross-border logistics task to obtain the task execution risk eigenvalue of each cross-border logistics task.

[0027] The natural language processing model includes an input layer, a semantic recognition layer, a feature extraction layer, and an output layer.

[0028] The specific steps for analyzing the cross-border customs evaluation feature set of each cross-border logistics task are as follows: In the input layer of the natural language processing model, receive the customs declaration information of each cross-border logistics task and perform preprocessing, specifically: clean the customs declaration information by removing irrelevant information such as special characters, HTML tags, and emojis, and perform standardization processing to ensure the consistency of date, time, and number formats. Split the text into meaningful words or phrases through word segmentation and annotation techniques (such as jieba, SpaCy, Stanford NLP, etc.) and label the词性 (such as nouns, verbs, adjectives, etc.) for each word. The word segmentation technique divides the continuous text sequence into words or phrases to help the model understand each independent language unit; The词性 annotation technique classifies each word to identify its function in the sentence. In addition, remove stop words such as "of", "is", "in", etc. These common invalid words do not have an actual effect on semantic analysis. Finally, use word vectorization techniques (such as BERT, Word2Vec, etc.) to convert the cleaned text into a vector representation to ensure that the semantic information of each word or phrase is effectively captured in its context, and during all operations, the original text of the customs declaration information is not modified to ensure that the authenticity and compliance of the information are not affected; In the semantic recognition layer of the natural language processing model, semantic processing is performed on the customs declaration information of each preprocessed cross-border logistics task. Specifically, key information entities in the customs declaration information are identified based on Named Entity Recognition (NER) technology and marked as relevant entities, such as: commodity name (e.g., electronic products), declared amount (e.g., US$10,000), HS code (e.g., 85423990), tariff rate (e.g., 5%), customs declaration number (e.g., 1234567890), consignee information (e.g., Company A, address: 123), invoice information (…). For example, invoice number: INV1234), and based on relation extraction techniques (such as support vector machines or conditional random fields, analyzing the context in the text to identify semantic relationships between different entities), the logical relationships between entities are identified, such as the relationship between the declared amount and the tariff rate: the relationship between these two entities is established based on the correspondence between the declared amount and the tariff rate (for example, the tariff rate corresponding to $10,000 is 5%), and the relationship between the consignee and invoice information: the dependency relationship between the customs declaration number and the invoice number is identified (for example, the invoice number corresponding to "1234567890" is "INV1234"). In the feature extraction layer of the natural language processing model, based on the customs declaration information of each cross-border logistics task after semantic processing, the cross-border customs feature vector of each cross-border logistics task is extracted. In the output layer of the natural language processing model, based on the cross-border customs feature vector of each cross-border logistics task, the cross-border customs evaluation feature set of each cross-border logistics task is output. Specifically, the binarized results of the compliance vector encoding compliance features, transportation compliance features, and tariff compliance features in the cross-border customs feature vector are combined to obtain a final compliance judgment: if all three features are 1, the declaration compliance feature value is 1; if any feature is 0, the declaration compliance feature value is 0, so as to obtain the declaration compliance feature value. The customs operation complexity features in the cross-border customs feature vector are processed by the Sigmoid function, and the result is mapped between 0 and 1 to obtain the customs operation complexity feature value.

[0029] The specific steps for extracting the cross-border customs feature vector for each cross-border logistics task are as follows: First, search for the corresponding HS code for the commodity name of each cross-border logistics task in a pre-set internal coding database and compare it with the HS code corresponding to the declaration. If they are completely identical, the coding compliance feature is determined to be 1; otherwise, it is determined to be 0. Second, search for the corresponding commodity name for each cross-border logistics task in a pre-set prohibited goods database. Specifically, compare the extracted commodity name (e.g., "electronic products") with the commodity names in the prohibited goods database. If the commodity name does not exist, the transportation compliance feature is determined to be 1; otherwise, it is determined to be 0. Third, obtain the corresponding tariff rate from the pre-set tariff database for the HS code of the commodity name of each cross-border logistics task and match it with the tariff rate corresponding to the commodity name. If they are completely identical, the tariff compliance feature is determined to be 1; otherwise, it is determined to be 0. Finally, concatenate the coding compliance feature, transportation compliance feature, and tariff compliance feature to obtain the compliance vector. The total number of entity types for each cross-border logistics task (the number of entities of different types, each type of entity is counted only once, and the duplication of entities is not considered) and the total number of relationships between entities (such as the relationship between the declared amount and the tariff rate, which can be regarded as a relationship, that is, the association between each pair of entities can be regarded as a relationship) are counted. The reference frequency of each entity in each cross-border logistics task is also counted, that is, the number of times the same document information (such as customs declaration number, invoice number) is referenced in the task. If the task involves 3 commodities, and each commodity references the same customs declaration number, then the reference frequency of the customs declaration number is 3. The average value is then processed to obtain the process dependency strength of each cross-border logistics task. The total number of entity types and the total number of relationships between entities are then standardized. Based on the standardization results, a weighted average is applied to extract the complex features of customs operations. The compliance vector and the complex features of customs operations are then concatenated into a cross-border customs feature vector.

[0030] The pre-training steps for natural language processing models are as follows: The labeled dataset consists of customs declaration information from cross-border logistics tasks. The data is labeled by customs experts based on real-time monitoring records and feedback from the declaration environment. Each sample in the labeled dataset includes customs declaration information for each commodity within a designated region, such as commodity name, declared amount, declared quantity, HS code, tariff rate, customs declaration number, consignee information, and invoice information. Each sample contains complete labels and is preprocessed. The preprocessed dataset is then divided into training, validation, and test sets, typically in a ratio of 80% for training, 10% for validation, and 10% for testing.

[0031] The natural language processing model is trained by taking the input layer as an example. The preprocessed text of each customs declaration information (such as commodity name, tariff rate, etc.) is input into the model, and text cleaning, word segmentation, annotation, vectorization and other steps are performed to extract the feature vector of each text unit. Through named entity recognition (NER) and relation extraction technology, the model can identify key information entities and analyze the relationship between entities.

[0032] During training, optimization algorithms (such as the Adam optimizer) are used to minimize prediction errors, and hyperparameters (such as learning rate, number of hidden layer units, etc.) are adjusted to improve model performance. The model is evaluated using a validation set to ensure that it can accurately capture compliance features, complexity features, and other relevant features in cross-border logistics tasks.

[0033] At the output layer, based on the extracted cross-border customs feature vectors (such as compliance vectors, complex customs operation features, etc.), the model outputs a comprehensive evaluation feature set for cross-border logistics tasks. The compliance judgment is generated by concatenating the coded compliance, transportation compliance, and tariff compliance feature values ​​of the compliance vector. The complex customs operation features are processed by mapping through the Sigmoid function, and the result is normalized to a value between 0 and 1.

[0034] After training, the model's generalization ability is evaluated using a test set to ensure that the model can accurately process unseen cross-border logistics data and extract effective features from it. The trained model is saved, and its parameters will be used in subsequent practical application stages to ensure accurate cross-border logistics customs assessment in real-time tasks.

[0035] In this implementation plan, by inputting customs declaration information of cross-border logistics tasks into a pre-trained natural language processing model, automated semantic parsing and feature extraction of text data can be achieved, thereby forming task execution risk feature values. Secondly, preprocessing and standardization operations at the input layer effectively clean up irrelevant information and unify the expression method, ensuring higher data integrity and consistency in subsequent modeling processes. Simultaneously, by combining semantic recognition and vectorized expression, text information is transformed into quantifiable feature vectors, further enhancing the model's ability to capture potential risks. This makes the extraction of task execution risk feature values ​​more accurate and improves analysis efficiency. Finally, by concatenating compliance vectors and complexity indicators, and using methods such as standardization and sigmoid function mapping, complex semantic information is transformed into quantifiable risk feature values, making different tasks comparable. Through pre-training and optimization, the model possesses good generalization ability, thereby achieving automated and efficient risk assessment and significantly improving the intelligence level of risk identification in cross-border logistics tasks.

[0036] Specifically, the transit operation data includes transit difference rate, transshipment load ratio, transshipment completion rate, and transit operation image data. The specific steps for analyzing the task safety and integrity characteristics of each cross-border logistics task are as follows: Based on the transit discrepancy rate, transshipment load ratio, and transshipment completion rate of each cross-border logistics task, the characteristic value of cross-border logistics operation completion for each task is analyzed. Specifically, the transit discrepancy rate, transshipment load ratio, and transshipment completion rate of each cross-border logistics task are standardized. Then, these standardized values ​​are weighted to obtain the characteristic value of cross-border logistics operation completion. During the weighting process, the standardized transit discrepancy rate is inverted, i.e., (1 - standardized transit discrepancy rate). Based on a pre-trained cargo damage recognition model and combined with the transit operation image data of each cross-border logistics task, the characteristic value of cargo damage for each cross-border logistics task is analyzed. Finally, based on the characteristic value of cross-border logistics operation completion and the characteristic value of cargo damage for each cross-border logistics task, the characteristic value of task safety and completeness for each cross-border logistics task is analyzed.

[0037] The specific formula for calculating the task safety completeness characteristic value of a certain cross-border logistics task is as follows: ;in, This refers to the task safety and completeness feature value for a specific cross-border logistics mission. This represents the characteristic value of the completion rate of a cross-border logistics task. This represents the cargo damage characteristic value for a specific cross-border logistics task. The coordination coefficients are stored in the database. The damage coefficient is stored in the database, and the coordination coefficient is used in this embodiment. Damage coefficient The values ​​were 0.867 and 1.500, respectively.

[0038] The transit discrepancy rate is the ratio of the actual quantity of goods to the expected quantity during transit operations. It can be calculated by reading the tags on the goods using RFID sensors or barcode scanners, uploading the results to the database, and then comparing them with the total quantity of goods for that task stored in the database. The result is calculated as (total quantity of goods - actual quantity of goods) / total quantity of goods, and the result is used as the transit discrepancy rate.

[0039] The transshipment load ratio is the ratio between the actual weight of the loaded cargo and the maximum load capacity of the transport vehicle during the transshipment operation. It can be obtained by weighing the actual weight of the transport vehicle through a weighing sensor and then compared with the maximum load weight of the transport vehicle stored in the database. That is, the actual weight / the maximum load weight, and the result is used as the transshipment load ratio.

[0040] The transshipment completion rate is the ratio between the number of transshipment operations completed according to plan and the total number of transshipment operations during the transit process. It can be obtained by weighing sensors to obtain the actual loaded weight of the transshipment transport vehicle and the total weight of the goods for the task stored in the database, and the ratio is processed to obtain the transshipment completion rate.

[0041] The specific steps for analyzing the cargo damage feature values ​​of each cross-border logistics task are as follows: Input the transit operation image data of each cross-border logistics task into the pre-trained cargo damage recognition model, analyze the transit damage feature set of each cross-border logistics task, including the feature values ​​of protective film integrity, surface contamination coverage, and packaging deformation; based on the transit damage feature set of each cross-border logistics task, analyze the cargo damage feature values ​​of each cross-border logistics task.

[0042] The specific formula for calculating the cargo damage characteristic value of a certain cross-border logistics task is as follows: ;in, This represents the cargo damage characteristic value for a specific cross-border logistics task. This represents the integrity characteristic value of the protective film for a specific cross-border logistics task. The protection coefficient is stored in the database. For a specific cross-border logistics task, the surface contamination coverage characteristic value is... The pollution coverage coefficient is stored in the database. This represents the characteristic value of packaging deformation for a specific cross-border logistics task. These are the deformation coefficients stored in the database. These are adjustment coefficients stored in the database. The smoothing coefficients are stored in the database, and In this implementation example, the protection coefficient is stored in the database. Pollution coverage coefficient Deformation coefficient Adjustment coefficient Smoothing coefficient The values ​​were 0.412, 0.347, 0.241, 0.458, and 0.333, respectively.

[0043] The following is a specific implementation example for calculating the cargo damage characteristic value of a cross-border logistics task. The available data includes: the protective film integrity characteristic value, the outer surface contamination coverage characteristic value, and the packaging deformation characteristic value of 5 randomly selected cross-border logistics tasks, as detailed in Table 1 and... Figure 3 As shown: Table 1. Example of Cross-border Logistics Task Sequence Data for Transit Damage Feature Set Cross-border logistics task 1 0.872 0.145 0.213 Cross-border logistics task 2 0.843 0.167 0.261 Cross-border logistics task 3 0.934 0.224 0.328 Cross-border logistics task 4 0.819 0.112 0.275 Cross-border logistics task 5 0.908 0.193 0.394 Protection coefficients stored in the database The value is: 0.412; Pollution coverage coefficients stored in the database The value is: 0.347; Deformation coefficients stored in the database The value is: 0.241; Adjustment coefficients stored in the database The value is: 0.458; Smoothing coefficients stored in the database The value is: 0.333; Substituting the data from Table 1 and the coefficients mentioned above into the specific formula for calculating the cargo damage characteristic value of a cross-border logistics task, we obtain: The cargo damage characteristic value of the first cross-border logistics task = 0.412×(1−0.872)+0.347×0.145+0.241×0.213+0.458×[(1−0.872)×0.145×0.213]^0.333≈0.227; The cargo damage characteristic value of the second cross-border logistics task = 0.412×(1−0.843)+0.347×0.167+0.241×0.261+0.458×[(1−0.843)×0.167×0.261]^0.333≈0.272; The cargo damage characteristic value of the third cross-border logistics task = 0.412×(1−0.934)+0.347×0.224+0.241×0.328+0.458×[(1−0.934)×0.224×0.328]^0.333≈0.252; The cargo damage characteristic value for the fourth cross-border logistics task is 0.412×(1−0.819)+0.347×0.112+0.241×0.275+0.458×[(1−0.819)×0.112×0.275]^0.333≈0.253; The cargo damage characteristic value of the fifth cross-border logistics task is 0.412×(1−0.908)+0.347×0.193+0.241×0.394+0.458×[(1−0.908)×0.193×0.394]^0.333≈0.289.

[0044] In this implementation plan, the standardization and weighted calculation of the transit difference rate, transshipment load ratio, and transshipment completion rate, combined with reverse processing to correct abnormal fluctuations, enable a more realistic portrayal of the operational completion level of cross-border logistics tasks during execution. Secondly, a pre-trained cargo damage identification model is introduced to automatically identify and extract features from transit operation images, generating cargo damage feature values. This allows potential safety issues during transit to be captured in real time. Finally, by fusing the operational completion feature values ​​with the cargo damage feature values, a task safety completeness feature value is formed. Coordination coefficients and damage coefficients are introduced to achieve differentiated measurement of the safety completeness of different tasks in complex scenarios, thereby improving the accuracy of feature extraction and evaluation, and significantly enhancing the safety assurance level of cross-border logistics.

[0045] Specifically, the customs clearance operation image data consists of the pixel value and two-dimensional coordinates of each pixel in the customs clearance operation image, and the cargo damage recognition model includes a transfer input layer, a damage extraction layer, and a damage output layer.

[0046] The specific steps for analyzing the transit damage feature set of each cross-border logistics task are as follows: In the transit input layer of the cargo damage identification model, the transit operation image data of each cross-border logistics task is received and image preprocessing is performed. Specifically, Gaussian filtering and median filtering are used to denoise the original image, eliminating noise or blur caused by shooting. Contrast stretching and histogram equalization are used to enhance the brightness and clarity of the image, ensuring that damage features (such as tearing, deformation, and contamination marks) are easier to identify. In the damage extraction layer of the cargo damage identification model, the damage feature vector of each cross-border logistics task is extracted based on the preprocessed transit operation image data of each cross-border logistics task. In the damage output layer of the cargo damage identification model, the transit damage feature set of each cross-border logistics task is output based on the damage feature vector of each cross-border logistics task. Specifically, the protective film integrity feature, outer contamination coverage feature, and packaging deformation feature in the damage feature vector are processed by the Sigmoid function, and the result is mapped between 0 and 1 to obtain the protective film integrity feature value, outer contamination coverage feature value, and packaging deformation feature value.

[0047] The specific steps for extracting the damage feature vector for each cross-border logistics task are as follows: Based on edge detection algorithms (such as Hough line detection), the pixel value and two-dimensional coordinates of each pixel in the transit operation image of each cross-border logistics task are used to identify the rectangular region of the cargo outline of each cross-border logistics task. This involves performing gradient calculations on the pixel value of each pixel in the transit operation image, calculating its gradient strength and direction in the horizontal and vertical directions, obtaining an edge point set based on threshold segmentation, and mapping the edge point set to the Hough parameter space. This involves converting the two-dimensional coordinates of the edge points in the transit operation image into a straight line representation in the parameter space. By iterating through different angle values ​​and performing voting statistics in an accumulator, possible straight lines are obtained. Based on the voting results, several lines with the highest cumulative values ​​are identified, and each line is filtered according to geometric constraints, including parallelism constraints (two lines are considered parallel when the difference in their directional angles is less than a preset threshold) and perpendicularity constraints (two lines are considered perpendicular when the difference in their directional angles is close to 90°). By combining lines that satisfy the parallel and perpendicular relationships, candidate rectangles are obtained. The coordinates of the intersection points of the candidate rectangles are solved, and a closed region is formed by the intersection points of four lines. The minimum bounding rectangle of this closed region is fitted to obtain the rectangular bounding box of the cargo's outer contour, thereby identifying the rectangular region of the cargo's outer contour for each cross-border logistics task. HSV conversion is performed on the pixel values ​​of each pixel within the rectangular area of ​​the cargo's outer contour. The luminance and saturation components of each pixel are extracted and compared with preset luminance and saturation thresholds. Pixels with luminance values ​​higher than the luminance threshold and saturation values ​​lower than the saturation threshold are selected and marked as candidate pixels for the protective film. This yields a set of candidate pixels for the protective film. Connectivity analysis (e.g., Flood analysis) is then performed on this set. Using algorithms such as Fill, four-neighborhood, and eight-neighborhood labeling, several candidate connected regions for protective film are obtained. The area ratio (the ratio of the number of candidate pixels in the protective film to the number of pixels within the rectangular area of ​​the cargo's outer contour) and color consistency value (the mean of the variances of the luminance and saturation components of each candidate pixel) are extracted for each candidate connected region. Based on a preset protective film integrity judgment rule: if the area ratio of a candidate connected region is greater than a first threshold (the proportion of the area of ​​the candidate connected region to the area of ​​the rectangular area of ​​the cargo's outer contour, for example, 80%), and its color consistency value is less than a second threshold (indicating uniform color in the region), then the candidate connected region is determined to be a valid protective film region. If multiple regions meet the conditions, the connected region with the largest area is selected as the protective film region (including several protective film pixels). The Canny edge detection algorithm is used to extract the boundary of the protective film region, obtaining a set of boundary pixels (including several boundary protective film pixels). Boundary processing is then performed (morphological dilation and erosion operations are used to preprocess the boundary pixel set, removing isolated noise points and smoothing boundary lines. Dilation fills small gaps between boundaries, and erosion removes false edge points, ensuring boundary continuity). For each boundary protective film pixel after boundary processing, the Euclidean distance between adjacent boundary protective film pixels is iterated. When the distance between adjacent boundary protective film pixels exceeds a preset breakage threshold (e.g., greater than 3 pixels), a breakage is detected. If the gap is marked as a break point, the number of boundary break points in the protective film area is counted and the ratio of the break point to the total number of pixels in the protective film area is calculated to obtain the boundary break rate of the protective film area. The mean value of the brightness component of the protective film pixels in the protective film area (the mean value of the brightness component of each protective film pixel) is extracted. At the same time, the area ratio in the protective film area is read and standardized. Based on the standardized boundary break rate, the mean value of the brightness component of the protective film pixels, and the area ratio, a weighted process is performed to extract the complete features of the protective film. In the weighted process, the standardized boundary break rate is taken as its complement, i.e., 1 - the standardized boundary break rate. The luminance and saturation components of each pixel within the rectangular region of the cargo's outer contour are read and compared with preset pollution luminance and saturation thresholds. Pixels with luminance components below the pollution luminance threshold and saturation components above the pollution saturation threshold are selected as pollution pixels. Based on a four-neighbor or eight-neighbor labeling algorithm, several pollution connected regions are extracted. The coverage ratio (the ratio of the number of pollution pixels in the pollution connected region to the total number of pixels in the rectangular region of the cargo's outer contour) and color dispersion (the ratio of the saturation of all pollution pixels in the pollution connected region) of each pollution connected region are extracted. The variances of the degree component and the luminance component are used, and the mean is taken as the color dispersion of the region. The contamination deposition degree is calculated by extracting the mean luminance and standard deviation of luminance based on the luminance components of all contaminated pixels in the contaminated connected region, and extracting the third and fourth central moments. Skewness and kurtosis are then extracted, i.e., the ratio of the third central moment to the cube of the luminance standard deviation and the ratio of the fourth central moment to the fourth power of the luminance standard deviation. These are then weighted to obtain the contamination deposition degree, and the absolute value of skewness is taken during the weighting process. The results are then standardized, and a weighted average is performed based on the standardized results to extract the surface contamination coverage features. The process involves reading the 2D coordinates of all pixels within the rectangular region of the cargo's outer contour, extracting the cargo's outer contour curve using an edge detection algorithm, and fitting the curve with the minimum bounding rectangle method. This yields the coordinates of the four vertices and the area of ​​the fitted rectangle. The ratio of the cargo contour area to the fitted rectangle area is analyzed and used as the rectangle filling rate. The average perpendicular distance from the cargo contour boundary points to the four sides of the rectangle is analyzed and normalized to obtain the boundary straightness residual. Simultaneously, the included angles between adjacent sides of the four vertices of the fitted rectangle are extracted, and the difference between each angle and 90° is calculated to obtain the angular orthogonality deviation. A convex hull region is constructed based on the cargo contour, and the ratio of the difference between the convex hull area and the cargo contour area to the convex hull area is analyzed as the convex defect rate. Finally, the geometric center of the fitted rectangle is extracted. Using this as a symmetry benchmark, the cargo outline is horizontally and vertically flipped to obtain horizontally symmetrical and vertically symmetrical outlines. The overlap ratio (using intersection-union ratio) between the original outline (i.e., edge detection of pixels within the rectangular area of ​​the cargo outline, such as the Canny algorithm, to obtain a continuous closed curve representing the true outer boundary of the cargo) and the horizontally and vertically symmetrical outlines is analyzed to obtain the horizontal overlap ratio and the vertical overlap ratio. The complement of the maximum value is taken as the symmetry loss. The rectangular filling rate, boundary straightness residual, angular orthogonality deviation, convex defect rate, and symmetry loss are normalized and weighted averaged based on preset weights to obtain the packaging deformation features. The protective film integrity features, external contamination coverage features, and packaging deformation features are concatenated into a damage feature vector.

[0048] Furthermore, the pre-training steps for the cargo damage recognition model are as follows: A labeled image dataset was obtained, consisting of cargo images taken at the customs clearance operation site of cross-border logistics tasks. The data was labeled by logistics supervision experts based on actual photographed samples. The labeling content includes the intact area of ​​the protective film, the area covered by external contamination, the area of ​​packaging deformation, and other potentially damaged areas. Each image sample has complete labeling information, including: the pixel segmentation mask of the protective film area, the pixel segmentation mask of the contamination area, the bounding box and geometric defect parameters of the packaging deformation area. The labeled image dataset was divided into training set, validation set and test set, usually in a ratio of 80% for training, 10% for validation and 10% for testing.

[0049] In the input layer, each cargo image in the training set is preprocessed, including Gaussian filtering, median filtering for noise reduction, histogram equalization, and contrast stretching to enhance image clarity, and the processed image is then input into the model.

[0050] In the damage extraction layer, the model uses a convolutional neural network to perform multi-scale convolution operations on the image, extracting texture, edge, and color features layer by layer. To ensure accurate identification of damage features, a multi-task learning mechanism is introduced during model training, which simultaneously optimizes the extraction of three types of features within the same network structure: Protective film integrity features: Through supervised segmentation tasks, the model is trained to accurately segment the protective film region and extract features such as region area ratio, boundary continuity, and average brightness; Surface contamination coverage features: Through region classification and segmentation tasks, the model is trained to identify contaminated areas and extract features such as coverage area ratio, color dispersion, and contamination deposition; Packaging deformation features: Through boundary detection and geometric fitting tasks, the model is trained to identify the packaging deformation contour and extract features such as rectangular filling rate, boundary straightness residual, angular orthogonality deviation, and symmetry loss.

[0051] During training, a supervised learning method is adopted to optimize the model by minimizing the loss function. The loss function includes: cross-entropy loss for image segmentation task, IoU loss for boundary detection task, and mean squared error (MSE) for regression feature prediction task. The Adam optimizer or SGD optimizer is used to iteratively update the network parameters and adjust hyperparameters such as learning rate, batch size, and number of convolutional layers to improve feature extraction accuracy and convergence speed.

[0052] The model's performance was evaluated using a validation set, with evaluation metrics including segmentation accuracy (mIoU), classification accuracy (Accuracy), and regression prediction error (MAE, RMSE) to ensure that the model can accurately capture cargo damage characteristics.

[0053] In the output layer, the model concatenates the protective film integrity features, the surface contamination coverage features, and the packaging deformation features into a damage feature vector, and normalizes it to between 0 and 1 using the Sigmoid function to form standardized damage feature values. Based on this damage feature vector, the model outputs a customs clearance damage feature set for subsequent cross-border logistics task analysis.

[0054] After training, the generalization ability of the model is verified using a test set to ensure that it can accurately process unseen transit cargo image data and correctly identify and quantify cargo damage features. Finally, the trained model parameters are saved for use in real-time cargo damage detection and assessment for subsequent cross-border logistics tasks.

[0055] In this implementation scheme, preprocessing the image data of the customs clearance operation effectively improves image clarity and ensures that potential damage details are fully preserved in subsequent analysis. Secondly, by using edge detection and connected component analysis, the model can accurately separate the outer contour of the goods from the local damage area, thereby avoiding misjudgments caused by background interference or shooting noise. Through multi-dimensional feature extraction, the damage to the goods can be quantified into standardized feature values. At the same time, the pre-trained model completes multi-task learning on a large-scale labeled image dataset, giving it strong generalization ability, thus maintaining stable recognition accuracy under different scenarios and diverse goods conditions. Finally, through damage feature vector concatenation and Sigmoid normalization, the damage features are uniformly mapped to the 0-1 interval, thereby ensuring the comparability of feature values, significantly improving the accuracy of damage recognition, and enhancing the logistics risk monitoring and quality assurance capabilities.

[0056] Specifically, the visualization mapping process involves the following steps: Normalizing the task execution risk characteristic value and task safety and completeness characteristic value for each cross-border logistics task, and constructing a two-dimensional task feature mapping space (i.e., using the normalized task execution risk characteristic value as the horizontal axis and the task safety and completeness characteristic value as the vertical axis); mapping the corresponding cross-border logistics task to the two-dimensional task feature mapping space based on the normalized task execution risk characteristic value and task safety and completeness characteristic value, as follows: The two midlines in the first quadrant of the two-dimensional mapping space for task features (i.e., the two lines with an abscissa of 0.5 and an ordinate of 0.5) are used to divide the space into four regions: Region A (task execution risk feature value < 0.5, task safety and completeness feature value ≥ 0.5), low risk, high completeness; Region B (task execution risk feature value ≥ 0.5, task safety and completeness feature value ≥ 0.5), high risk, high completeness; Region C (task execution risk feature value < 0.5, task safety and completeness feature value < 0.5), low risk, low completeness; Region D (task execution risk feature value ≥ 0.5, task safety and completeness feature value < 0.5), high risk, low completeness. Then, based on the normalized task execution risk feature value and task safety and completeness feature value of each cross-border logistics task, it is mapped to the corresponding region, with each cross-border logistics task serving as a mapping point. Furthermore, the color of each mapping point gradually changes from green to yellow to red, increasing with the normalized task execution risk characteristic value (i.e., when the normalized task execution risk characteristic value = 0, the mapping point color is pure green; when the normalized task execution risk characteristic value = 0.5, the mapping point color is yellow; when the normalized task execution risk characteristic value = 1, the mapping point color is pure red; when the normalized task execution risk characteristic value < 0.5, the mapping point color changes linearly from green to yellow, and its RGB component values ​​are calculated through linear interpolation; when 0.5 < normalized task execution risk characteristic value < 1, the mapping point color changes linearly from yellow to red, and its RGB component values ​​are calculated through linear interpolation). The transparency is set by increasing with (1 - normalized task security completeness characteristic value) (i.e., the less complete, the less transparent).

[0057] In this implementation plan, by normalizing and mapping the task execution risk characteristic values ​​and task safety integrity characteristic values ​​to a unified two-dimensional feature space, different tasks can be intuitively presented on the coordinate plane in a point-like form. This avoids the information fragmentation and comprehension difficulties caused by relying solely on tables or text data. Secondly, through regional division, tasks can be automatically divided into four categories, allowing managers to quickly identify the status of tasks and significantly improve the efficiency of risk situation perception. At the same time, the introduction of color gradients gradually transitions the risk level from green to red as the value increases, intuitively reinforcing the urgency of the risk. Transparency is linked to safety integrity, making tasks with insufficient integrity more prominent in the interface, thus helping managers focus their attention on high-risk and incomplete key links. Finally, this visualization process not only improves the distinguishability of task status but also upgrades the risk monitoring and safety control of cross-border logistics from numerical to graphical, thereby helping managers quickly identify problems in complex task groups and improve the targeting of management.

[0058] Please see Figure 4This invention provides a technical solution: a cross-border logistics information visualization management system based on big data, comprising: a data acquisition module for acquiring customs declaration information and transit operation data for several cross-border logistics tasks; an execution risk analysis module for analyzing the task execution risk characteristic value of each cross-border logistics task based on a pre-trained natural language processing model and combined with the customs declaration information of each cross-border logistics task; a transit operation analysis module for analyzing the task safety and completeness characteristic value of each cross-border logistics task based on the transit operation data of each cross-border logistics task; a visualization mapping module for performing visualization mapping processing on the task execution risk characteristic value and task safety and completeness characteristic value of each cross-border logistics task; and a joint management feedback module for jointly managing the corresponding cross-border logistics tasks based on the visualization mapping processing results.

[0059] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention.

[0060] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A cross-border logistics information visualization management method based on big data, characterized in that: Includes the following steps: Obtain customs declaration information and transit operation data for several cross-border logistics tasks; Based on a pre-trained natural language processing model and combined with the customs declaration information of each cross-border logistics task, the task execution risk feature value of each cross-border logistics task is analyzed. Specifically, the natural language processing model performs customs semantic information mining and evaluation feature extraction on the customs declaration information to obtain a cross-border customs evaluation feature set including declaration compliance feature value and customs operation complexity feature value, and comprehensively analyzes to obtain the task execution risk feature value of each cross-border logistics task. Based on the customs clearance operation data for each cross-border logistics task, including customs clearance discrepancy rate, transshipment load ratio, transshipment completion rate, and customs clearance operation image data, the task safety and integrity characteristic values ​​for each cross-border logistics task are analyzed, specifically as follows: Based on the transit difference rate, transshipment load ratio, and transshipment completion rate of each cross-border logistics task, the characteristic value of the cross-border logistics operation completion of each cross-border logistics task is analyzed. Among them, the transit difference rate is the proportion of the difference between the actual quantity of goods and the expected total quantity of goods during the transit operation. The transshipment load ratio is the ratio between the actual weight of the loaded cargo and the maximum load capacity of the transport vehicle during the transshipment operation. Based on a pre-trained cargo damage identification model and combined with the transit operation image data of each cross-border logistics task, the cargo damage feature value of each cross-border logistics task is analyzed. Specifically, the cargo damage identification model performs multi-dimensional damage analysis and quantitative extraction processing on the transit operation image data to obtain a transit damage feature set including the feature value of the protective film integrity, the feature value of the outer surface contamination coverage, and the feature value of the packaging deformation, and analyzes the cargo damage feature value of each cross-border logistics task. Based on the cross-border logistics operation completion characteristic value and cargo damage characteristic value of each cross-border logistics task, analyze the task safety integrity characteristic value of each cross-border logistics task. Visual mapping is performed on the task execution risk characteristic value and task safety integrity characteristic value of each cross-border logistics task; Joint management of corresponding cross-border logistics tasks is carried out based on the results of visualization mapping.

2. The cross-border logistics information visualization management method based on big data according to claim 1, characterized in that, The natural language processing model includes an input layer, a semantic recognition layer, a feature extraction layer, and an output layer.

3. The cross-border logistics information visualization management method based on big data according to claim 2, characterized in that, The specific steps for analyzing the cross-border customs assessment feature set for each cross-border logistics task are as follows: In the input layer of the natural language processing model, customs declaration information for each cross-border logistics task is received and preprocessed. In the semantic recognition layer of the natural language processing model, semantic processing is performed on the customs declaration information of each preprocessed cross-border logistics task. In the feature extraction layer of the natural language processing model, based on the customs declaration information of each cross-border logistics task after semantic processing, the cross-border customs feature vector of each cross-border logistics task is extracted. In the output layer of the natural language processing model, based on the cross-border customs feature vector of each cross-border logistics task, the cross-border customs evaluation feature set of each cross-border logistics task is output.

4. The cross-border logistics information visualization management method based on big data according to claim 1, characterized in that, The customs clearance operation image data specifically includes the pixel value and two-dimensional coordinates of each pixel in the customs clearance operation image, and the cargo damage recognition model includes a transfer input layer, a damage extraction layer, and a damage output layer.

5. The cross-border logistics information visualization management method based on big data according to claim 4, characterized in that, The specific steps for analyzing the transit damage feature set of each cross-border logistics task are as follows: In the transshipment input layer of the cargo damage identification model, the image data of the customs clearance operation for each cross-border logistics task is received and the images are preprocessed. In the damage extraction layer of the cargo damage identification model, the damage feature vector of each cross-border logistics task is extracted based on the preprocessed customs clearance operation image data of each cross-border logistics task. In the damage output layer of the cargo damage identification model, the transit damage feature set of each cross-border logistics task is output based on the damage feature vector of each cross-border logistics task.

6. The cross-border logistics information visualization management method based on big data according to claim 1, characterized in that, The specific steps for visual mapping processing are as follows: The task execution risk characteristic value and task safety and integrity characteristic value of each cross-border logistics task are normalized, and a two-dimensional mapping space of task characteristics is constructed. Based on the normalized task execution risk characteristic value and task safety and integrity characteristic value, the corresponding cross-border logistics tasks are mapped to the two-dimensional task characteristic mapping space.

7. A cross-border logistics information visualization management system based on big data, employing the cross-border logistics information visualization management method based on big data as described in any one of claims 1-6, characterized in that, include: The data acquisition module is used to obtain customs declaration information and transit operation data for several cross-border logistics tasks; Execute the risk analysis module; This is used to analyze the task execution risk characteristics of each cross-border logistics task based on a pre-trained natural language processing model and combined with the customs declaration information of each cross-border logistics task. The transit operation analysis module is used to analyze the task safety and integrity characteristics of each cross-border logistics task based on the transit operation data of each cross-border logistics task. The visualization mapping module is used to perform visualization mapping processing on the task execution risk characteristic value and task safety integrity characteristic value of each cross-border logistics task; The joint management feedback module is used to jointly manage corresponding cross-border logistics tasks based on the results of visual mapping processing.

Images