A supply chain full-link risk early warning platform and method based on big data

Abstract

The present invention discloses a supply chain full-link risk warning platform and method based on big data. The platform includes a supply chain division module, a data collection module, a risk assessment module, an alternative supplier module, a risk path transmission module and a risk plan module. By dividing the supply chain information according to the link nodes and then collecting multi-source data, combining supplier data, consumer behavior data and public opinion data to generate supplier quality information and bind suppliers, dynamically adjusting the number of alternative suppliers based on real-time volatility, constructing a risk transmission path containing path nodes of different risk levels, and finally generating a risk plan adapted to the link scenario. Through multi-dimensional data fusion and a dynamic threshold mechanism, the present invention strengthens the ability to capture the cascading effect of supply chain risk transmission, and achieves a coordinated improvement in the timeliness of risk warnings and the accuracy of plans.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a big data-based full-link supply chain risk early warning platform and method. Background Art

[0002] Based on the globalization of the economy and the iterative updates of science and technology, risk warning operations for the supply chain have been born to help companies control risks in more critical supply demands. However, existing risk predictions still have the following limitations: the supply chain has multiple categories of data throughout the process, and the current risk prediction can only realize the processing and prediction of data of a unified category, resulting in the singleness of risk assessment; the risk prediction of the supply chain relies on the processing of historical data and does not take into account temporary and sudden situations. In some areas with timeliness requirements, especially the e-commerce field, historical data is not referenceable due to sudden situations such as marketing and current affairs, making risk predictions less accurate. Summary of the Invention

[0003] In view of the above problems, the present invention provides a full-link supply chain risk warning platform and method based on big data, which solves the problems of existing risk prediction in application, such as single risk assessment and inability to accurately predict emergencies.

[0004] To achieve the above objectives, in the first aspect, the present application provides a supply chain full-link risk early warning platform based on big data, including a supply chain segmentation module, a data collection module, a risk assessment module, an alternative supplier module, a risk path transmission module and a risk plan module.

[0005] The supply chain partitioning module is used to obtain supply chain information and divide the supply chain information according to link nodes to obtain multiple link node information;

[0006] The data collection module is used to obtain multi-source data for each link node, including supplier data, consumer behavior data, logistics timeliness data, regional data, and public opinion data;

[0007] The risk assessment module is used to generate risk information for each link node based on multi-source data, and to generate supplier quality information based on consumer behavior data and public opinion data, and to bind the supplier quality information to the supplier;

[0008] The alternative supplier module is used to screen a preset number of alternative suppliers in the database based on risk information and link node information, and update the alternative suppliers to the link node information. The preset number is configured to be generated based on the real-time volatility of the current link node;

[0009] The risk path transmission module is used to generate a risk transmission path based on risk information and supply chain information. The risk transmission path includes multiple path nodes of different risk levels.

[0010] The risk plan module is used to generate risk plans based on risk information, risk transmission paths and updated link node information.

[0011] In some embodiments, the data acquisition module includes a data standardization unit, a spatiotemporal correlation unit, and a real-time update unit. The data standardization unit is used to standardize multi-source data to obtain standardized data, including:

[0012] Extract the first key field from the supplier data and convert the first key field into a first structured data entry;

[0013] Process the return records in the consumer behavior data to obtain the category return rate, and perform NLP sentiment analysis on the text reviews in the consumer behavior data to generate sentiment polarity scores;

[0014] Construct a regional supplier mapping table by combining regional data, supplier data, and regional policies;

[0015] Extracting the second key field from the public opinion data and converting the second key field into a second structured data entry;

[0016] Convert logistics timeliness data into third-party structured data entries;

[0017] The spatiotemporal association unit is used to associate normalized data with link nodes;

[0018] The real-time update unit is used to dynamically update multi-source data through the stream processing framework.

[0019] In some embodiments, the risk path transmission module includes a node association unit, a risk quantification unit, and a path adjustment unit. The node association unit is used to construct a logical dependency relationship between multiple link nodes to obtain a composite topology structure, including:

[0020] Generate an initial topology structure based on the node connection rules in the supply chain information;

[0021] Calculate the hierarchical association between each link node, build multi-directional association lines on the initial topology structure, and generate a composite topology structure;

[0022] Initializing risk transfer coefficients for multiple link nodes in a composite topology;

[0023] The risk quantification unit is used to quantify the risks of multiple link nodes based on the risk information and the composite topology structure, obtain the risk increment of each link node, and generate the final risk weight of each link node;

[0024] The path adjustment unit is used to generate risk transmission paths with different risk levels based on the final risk weight and composite topological structure.

[0025] In a second aspect, the present invention further provides a supply chain full-link risk warning method based on big data, applicable to the risk warning platform of the first aspect, the method comprising:

[0026] Obtain supply chain information, divide the supply chain information according to link nodes, and obtain information of multiple link nodes;

[0027] Obtain multi-source data of each link node information regularly according to the preset acquisition frequency and pre-process it, and map the pre-processed multi-source data with the link node and store it;

[0028] Generate risk information for each link node based on pre-processed multi-source data, generate supplier quality information based on consumer behavior data and public opinion data, and bind supplier quality information to suppliers;

[0029] Based on the risk information and link node information, a preset number of alternative suppliers are screened in the database, and the alternative suppliers are updated to the link node information. The preset number is configured to be generated based on the real-time volatility of the current link node;

[0030] Generate a risk transmission path based on risk information and supply chain information. The risk transmission path includes multiple path nodes with different risk levels.

[0031] Generate risk plans based on risk information, risk transmission paths and updated link node information.

[0032] In some embodiments, obtaining and preprocessing multi-source data of each link node information, and mapping and storing the pre-processed multi-source data with the link nodes includes:

[0033] Extract the first key field from the supplier data and convert the first key field into a first structured data entry;

[0034] Process the return records in the consumer behavior data to obtain the category return rate, and perform NLP sentiment analysis on the text reviews in the consumer behavior data to generate sentiment polarity scores;

[0035] Construct a regional supplier mapping table by combining regional data, supplier data, and regional policies;

[0036] Extracting the second key field from the public opinion data and converting the second key field into a second structured data entry;

[0037] Convert logistics timeliness data into third-party structured data entries;

[0038] Generate node data information according to a preset format based on the first structured data entry, the second structured data entry, the third structured data entry, the regional supplier mapping table, and the sentiment polarity score;

[0039] Mapping and storing node data information and link node information;

[0040] Furthermore, it is evaluated whether the node data information meets the preset fluctuation condition. If not, the preset collection frequency of the multi-source data is adjusted.

[0041] In some embodiments, generating risk information for each link node information based on pre-processed multi-source data includes:

[0042] According to the link node information, the pre-processed multi-source data associated with it is sorted by the influence coefficient according to the AHP algorithm to obtain the influence coefficient of each data type in the multi-source data;

[0043] The risk weight of each data type is converted using a linear normalization function based on the impact coefficient and recorded as the initial risk weight;

[0044] Calculate the risk probability of each data type. The risk probability is configured to be calculated using the Poisson distribution model based on the number of risk occurrences in historical data information of the same data type at the same link node.

[0045] Merge the risk probability with the initial risk weight to form the basic risk weight;

[0046] The CUSUM algorithm is used to extract the sudden information from multi-source data. The sudden information includes policy sudden features, promotion sudden features, and regional sudden features.

[0047] Convert the emergency information into emergency risk weights using fuzzy logic algorithm;

[0048] The basic risk weight and the sudden risk weight are integrated to obtain risk information.

[0049] In some embodiments, the pre-processed multi-source data associated with the link node information is sorted by influence coefficient according to the AHP algorithm, and the influence coefficient of each data type in the multi-source data is obtained, including:

[0050] Constructing a judgment matrix corresponding to the node type of the current link node, wherein the matrix elements of the judgment matrix are generated by the relative importance of the data types of the multi-source data;

[0051] Calculate the maximum eigenvalue of the judgment matrix and its associated eigenvector, where each component of the eigenvector corresponds to a data type;

[0052] Normalize the eigenvectors to obtain the influence coefficients of multiple data types;

[0053] Calculate the risk probability of each data type. The risk probability is configured to be calculated using the Poisson distribution model using the number of risk occurrences in the historical data information of the same data type at the same link node, including:

[0054] Count the average number of risk events of the current data type per unit time based on historical data information;

[0055] The probability of at least one risk event occurring in the current time window is calculated using the Poisson distribution probability function. This is expressed as the risk probability using formula (1). Formula (1) is as follows:

[0056] ;

[0057] In formula (1), is the risk probability, is the number of risk events that occur, is the base of natural logarithms, is the average number of risk events occurring per unit time;

[0058] The CUSUM algorithm is used to extract the burst information from the multi-source data. The burst information includes policy burst characteristics, promotion burst characteristics, and regional burst characteristics.

[0059] Cumulative statistics are constructed for the time series data stream in multi-source data. When the cumulative statistics exceed the preset control limit, it is marked as a mutation point, which is expressed by formula (2). Formula (2) is as follows:

[0060] ;

[0061] In formula (2), For Cumulative statistics at time, For Cumulative statistics at time, For The observed value at time, is the benchmark mean, is the permissible offset parameter;

[0062] Generating at least one of the policy burst feature, promotion burst feature, or regional burst feature according to the data type classification corresponding to the mutation point;

[0063] The conversion of emergency information into emergency risk weights using fuzzy logic algorithms includes:

[0064] Establishing a fuzzy rule base, wherein the input variables of the fuzzy rule base are the intensity of the burst feature and the duration of the burst feature, and the output variable of the fuzzy rule base is the burst risk weight;

[0065] The fuzzy reasoning results are converted into numerical sudden risk weights through defuzzification operations.

[0066] In some embodiments, generating supplier quality information based on consumer behavior data and public opinion data, and binding the supplier quality information to the supplier includes:

[0067] Process the return records in consumer behavior data and calculate the return rate of each supplier's category;

[0068] Extract textual evaluations from consumer behavior data for NLP sentiment analysis to generate supplier quality satisfaction scores;

[0069] Extract key fields from public opinion data to identify supplier-related quality incidents;

[0070] Calculate the public opinion risk value based on the frequency and spread of quality incidents;

[0071] The product category return rate, quality satisfaction score and public opinion risk value are weighted and integrated to generate a comprehensive quality score;

[0072] Divide the quality grade according to the comprehensive quality score and obtain the supplier quality information;

[0073] Associate and store supplier quality information with corresponding suppliers;

[0074] and, triggering a screening mechanism for alternative suppliers when the quality level falls below a preset level threshold;

[0075] In addition, supplier quality information will be updated according to preset cycles based on newly added consumer behavior data and public opinion data.

[0076] In some embodiments, a preset number of candidate suppliers are screened in a database based on risk information and link node information, and the candidate suppliers are updated to the link node information. The preset number is configured to be generated based on the real-time volatility of the current link node, including:

[0077] Calculate the real-time volatility of the current link node. The real-time volatility is obtained by converting the degree of change of key indicators in the link node information.

[0078] Dynamically adjust the preset quantity according to the real-time volatility;

[0079] Prioritize suppliers in the database based on pre-set screening criteria, including geographic coverage match, supplier quality rating, historical on-time delivery rate, and cost premium;

[0080] A preset number of suppliers to be replaced with high rankings are selected and recorded as candidate suppliers, and are updated to the link node information.

[0081] In some embodiments, a risk transmission path is generated based on risk information and supply chain information. The risk transmission path includes multiple path nodes of different risk levels, including:

[0082] Generate an initial topology structure based on the node connection rules in the supply chain information;

[0083] Calculating the hierarchical correlation between each link node, where the hierarchical correlation is configured to be generated based on the physical connection strength, data interaction frequency, and service dependency between multiple link nodes;

[0084] Construct multi-directional association lines on the initial topology structure according to the hierarchical association degree to generate a composite topology structure;

[0085] Initializing risk conduction coefficients for multiple link nodes in a composite topology structure, the risk conduction coefficients are configured to be obtained by the following steps:

[0086] Calculate the conduction influence weights of the preceding link node and the subsequent link node at the current link node;

[0087] Also, calculate the collaborative risk weights of other link nodes at the same level as the current link node;

[0088] and, calculating the indirect conduction attenuation weights between multiple link nodes across the hierarchy and the current link node;

[0089] The risk transmission coefficient is obtained by integrating the transmission impact weight, the synergistic risk weight and the indirect transmission attenuation weight;

[0090] Based on the risk information and the composite topology, the risks of multiple link nodes are quantified to obtain the risk increment of each link node, and the final risk weight of each link node is generated, which is expressed by formula (3). Formula (3) is as follows:

[0091] ;

[0092] In formula (3), is the final risk weight of the current link node, is the initial risk value of the current link node, is the risk transmission coefficient, For risk increment;

[0093] Based on the final risk weight and composite topological structure, risk transmission paths with different risk levels are generated.

[0094] Different from the existing technology, the above technical solution provides a full-link supply chain risk warning platform and method based on big data. The platform includes a supply chain division module, a data collection module, a risk assessment module, an alternative supplier module, a risk path transmission module and a risk plan module. After dividing the supply chain information according to the link nodes, multi-source data is collected, and supplier quality information is generated and bound to the supplier in combination with supplier data, consumer behavior data and public opinion data. The number of alternative suppliers is dynamically adjusted based on real-time volatility, and a risk transmission path containing path nodes of different risk levels is constructed, and finally a risk plan adapted to the link scenario is generated. Through multi-dimensional data fusion and a dynamic threshold mechanism, the above technical solution strengthens the ability to capture the cascading effect of supply chain risk transmission, and achieves a coordinated improvement in the timeliness of risk warnings and the accuracy of plans.

[0095] The above-mentioned records related to the content of the invention are only an overview of the technical solution of this application. In order to enable ordinary technicians in this field to understand the technical solution of this application more clearly, and then implement it according to the text of the specification and the contents recorded in the drawings, and to make the above-mentioned purposes and other purposes, features and advantages of this application easier to understand, the following is an explanation in combination with the specific implementation methods and drawings of this application. BRIEF DESCRIPTION OF THE DRAWINGS

[0096] The drawings are only used to illustrate the principles, implementation methods, applications, features and effects of the specific embodiments of the present invention and other related contents, and are not to be considered as limiting the present application.

[0097] In the drawings of the specification:

[0098] Figure 1 This is a schematic diagram of the structure of the risk warning platform described in the specific implementation method;

[0099] Figure 2 This is a method step diagram of steps S101 to S106 of the risk warning method described in the specific implementation method;

[0100] Figure 3 This is a method step diagram of steps S201 to S207 of the risk warning method described in the specific implementation method;

[0101] Figure 4 A method step diagram of steps S301 to S307 of the risk warning method described in the specific implementation method;

[0102] Figure 5 This is a method step diagram of steps S401 to S407 of the risk warning method described in the specific implementation method.

[0103] The reference numerals in the above drawings are described as follows:

[0104] 1. Risk early warning platform; 11. Supply chain segmentation module; 12. Data collection module; 13. Risk assessment module; 14. Alternative supplier module; 15. Risk path transmission module; 16. Risk contingency plan module. DETAILED DESCRIPTION

[0105] In order to explain in detail the possible application scenarios, technical principles, specific solutions that can be implemented, and the purpose and effects of this application, the following is a detailed description of the specific embodiments listed in conjunction with the accompanying drawings. The embodiments described herein are only used to more clearly illustrate the technical solutions of this application and are therefore only examples and are not intended to limit the scope of protection of this application.

[0106] References to "embodiments" herein mean that the specific features, structures, or characteristics described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the word "embodiment" in various places in the specification does not necessarily refer to the same embodiment, nor does it particularly limit its independence or relevance to other embodiments. In principle, in this application, as long as there are no technical contradictions or conflicts, the various technical features mentioned in the embodiments can be combined in any manner to form a corresponding implementable technical solution.

[0107] Unless otherwise defined, the technical terms used herein have the same meanings as those generally understood by those skilled in the art to which this application belongs; the use of relevant terms herein is only for describing specific embodiments and is not intended to limit this application.

[0108] In the description of this application, the term "and / or" is used to describe a logical relationship between objects, indicating that three relationships can exist. For example, A and / or B means: A exists, B exists, and both A and B exist. In addition, the character " / " in this document generally indicates that the objects before and after are in a logical "or" relationship.

[0109] In this application, terms such as "first" and "second" are merely used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual quantity, priority or sequence relationship between these entities or operations.

[0110] Without further limitations, in this application, the words "include", "comprise", "have" or other similar open-ended expressions used in sentences are intended to cover non-exclusive inclusion. These expressions do not exclude the presence of additional elements in the process, method or product that includes the elements, so that the process, method or product that includes a series of elements may include not only those defined elements, but also other elements that are not explicitly listed, or also include elements inherent to such process, method or product.

[0111] Consistent with the understanding in the Examination Guidelines, in this application, expressions such as "greater than," "less than," and "exceed" are understood to exclude the number itself; expressions such as "above," "below," and "within" are understood to include the number itself. Furthermore, in the description of the embodiments of this application, "multiple" means two or more (including two), and similar expressions related to "multiple," such as "multiple groups" and "multiple times," are also understood in this manner, unless otherwise specifically defined.

[0112] In the description of the embodiments of the present application, the space-related expressions used, such as "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "vertical", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc., indicate the orientation or position relationship based on the orientation or position relationship shown in the specific embodiments or drawings, and are only for the convenience of describing the specific embodiments of the present application or facilitating the reader's understanding, and do not indicate or imply that the device or component referred to must have a specific position, a specific orientation, or be constructed or operated in a specific orientation. Therefore, it should not be understood as a limitation on the embodiments of the present application.

[0113] The processor described in the embodiments of the present application can be implemented by hardware, firmware, software or a combination thereof, and can use a circuit, a single or multiple application-specific integrated circuits (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field programmable gate array (FPGA), a central processing unit (CPU), a controller, a microcontroller, at least one of a microprocessor, and also includes other physical, biological or chemical structures that can achieve similar or equivalent functions to the processors listed above, such as biological neurons, quantum computing units, DNA computing units, etc., so that the processor can execute some or all of the steps in the computer program or method involved in the various embodiments of the present application, or any combination of the steps mentioned therein.

[0114] The computer program involved in the embodiment can be stored in a computer device readable storage medium, which includes but is not limited to a disk, a tape, a magnetic card, a floppy disk, a flash memory, an optical disc, an optical card, a read-only memory (ROM), a random access memory (RAM), an erasable programmable ROM (EPROM) and an electrically erasable programmable ROM (EEPROM), etc., and also includes other biological, physical or chemical structures that can achieve similar or equivalent functions to the storage media listed above, such as DNA, RNA, protein and other units with information storage capabilities. In a specific embodiment, the storage medium involved can be one of the above-mentioned media types or a combination of the above-mentioned media types. In different embodiments, the computer program involved in the embodiment can be stored in a single medium in a centralized manner or in a distributed manner on multiple media. The memory containing the computer device readable storage medium can be a non-volatile memory or a random access memory. These computer device readable storage media can be built into the device or connected to the device involved in the embodiment as an external device or part of an external device. In some embodiments, a memory having a computer-readable storage medium is deployed locally; in other embodiments, a solution of deploying the memory away from the processor may also be adopted, such as a network-attached memory accessed via an RF circuit or an external port and a communication network, wherein the communication network may be the Internet, one or more intranets, a local area network (LAN), a wide area network (WLAN), a storage area network (SAN), etc., or a suitable combination thereof, as long as the computer device can access the memory. In addition, the computer program involved in the embodiment can be stored in plaintext / ciphertext form, or can be designed as training data, which can be integrated and reorganized through model training and implicitly stored in the parameter state of a deep neural network or other machine learning model.

[0115] See also Figure 1 In a first aspect, this embodiment provides a supply chain full-link risk early warning platform 1 based on big data, including a supply chain segmentation module 11, a data collection module 12, a risk assessment module 13, an alternative supplier module 14, a risk path transmission module 15, and a risk plan module 16.

[0116] The supply chain partitioning module 11 is used to obtain supply chain information and partition the supply chain information according to link nodes to obtain a plurality of link node information;

[0117] The data collection module 12 is used to obtain multi-source data of each link node information, including supplier data, consumer behavior data, logistics timeliness data, regional data and public opinion data;

[0118] The risk assessment module 13 is used to generate risk information for each link node based on multi-source data, and to generate supplier quality information based on consumer behavior data and public opinion data, and to bind the supplier quality information to the supplier;

[0119] The candidate supplier module 14 is used to screen a preset number of candidate suppliers in the database based on risk information and link node information, and update the candidate suppliers to the link node information. The preset number is configured to be generated based on the real-time volatility of the current link node;

[0120] The risk path transmission module 15 is used to generate a risk transmission path based on risk information and supply chain information. The risk transmission path includes multiple path nodes of different risk levels.

[0121] The risk plan module 16 is used to generate a risk plan based on risk information, risk transmission paths and updated link node information.

[0122] In the supply chain segmentation module 11, link node information includes merchant management nodes, procurement and inventory nodes, order flow nodes, warehousing and sorting nodes, logistics and distribution nodes, after-sales nodes, and consumer feedback nodes. Risks are segmented by link node information to achieve refined early warning. The consumer feedback node is used to collect user evaluation text and return records to generate supplier quality information.

[0123] In the risk assessment module 13, preferably, the supplier data in the multi-source data includes qualification documents, delivery punctuality rate and quality sampling results; consumer behavior data refers to data such as consumer purchasing style, purchase quantity, purchase age, transaction success rate, etc.; public opinion data can be divided into three categories: current affairs public opinion, activity public opinion and negative review public opinion. Among them, current affairs public opinion refers to sudden events, such as events caused by local policies, public panic caused by consumer accidents and reduction of this demand, etc. Activity public opinion is such as a certain promotional event, and negative review public opinion refers to consumer negative review data, including but not limited to negative logistics reviews, negative product quality reviews, etc., that is, supplier quality information. Preferably, regional data can construct a regional supplier mapping table through the mapping relationship between the supplier's location and the business coverage area.

[0124] In the alternative supplier module 14, the preset quantity is dynamically adjusted based on real-time fluctuations. This means that the module 14 automatically calculates the required number of alternative suppliers based on the real-time fluctuations of the current link node (e.g., inventory turnover deviation, logistics time fluctuations). The higher the real-time fluctuations, the larger the preset quantity. Preferably, the selection criteria for alternative suppliers include geographic coverage matching, quality grade, and historical on-time delivery performance. Geographic coverage matching is calculated by the overlap between the supplier's service area and the current node's business area.

[0125] In the risk path transmission module 15, the generation of the risk transmission path preferably relies on the logical dependency relationship between the link node information of the composite topology modeling. Furthermore, when the supplier quality information fails to meet the requirements, the alternative supplier screening mechanism is triggered, realizing a dynamic risk path and improving the response capability.

[0126] Risk contingency plan module 16 generates response measures based on the interrelated impacts of nodes at different levels in the risk transmission path. For example, if the risk information indicates that a popular product is running low on inventory during Prime Day and that alternative suppliers cannot restock within seven days, the procurement node will predict demand based on historical flash sale data and preemptively trigger alternative suppliers (e.g., those with ready stock in overseas warehouses). If the risk information indicates that a product faces removal due to a surge in negative reviews related to "battery fires," the consumer feedback node will monitor keywords in real time and trigger quality risk contingency plans (e.g., suspending purchases, contacting suppliers for inspection). If further focus is needed on a specific type of e-commerce (e.g., cross-border e-commerce / DTC brands), the corresponding nodes (e.g., customs clearance, overseas warehouses) can be refined.

[0127] The full-link supply chain risk early warning platform 1 provided in this embodiment utilizes a modular architecture to achieve closed-loop control of risk identification, transmission modeling, and dynamic response. The supply chain segmentation module 11 divides supply chain information into independent link nodes (e.g., procurement nodes and logistics nodes) based on business segments, with each node associated with a specific data source and risk indicator. The data acquisition module 12 integrates multi-source data and constructs structured data entries through standardized processing to provide input for risk assessment. The risk assessment module 13 generates risk information for link nodes based on multi-source data fusion and extracts quality characteristics from consumer behavior and public opinion data to generate supplier quality information, dynamically binding suppliers to risk indicators. The alternative supplier module 14 dynamically adjusts the number of alternative suppliers based on the current node's real-time volatility and selects the optimal alternative resources based on conditions such as geographic coverage match and quality grade. The risk path transmission module 15 models the logical dependencies between nodes (e.g., the business dependency strength between procurement nodes and warehousing nodes) through a composite topological structure, quantifies the cross-node risk transmission effect, and generates transmission paths with different risk levels. The risk contingency plan module 16 outputs multi-level response strategies, such as supplier switching and coordinated inventory transfer, based on the risk weights of nodes in the transmission path and information about alternative suppliers.

[0128] Furthermore, the above technical solution can be understood in conjunction with the following examples: in cross-border e-commerce scenarios (such as Amazon, Taobao, etc.), when a customs clearance node triggers risks due to sudden policy changes, the regional supplier mapping table screens alternative suppliers that comply with the new policy, and the risk path transmission module 15 calculates the indirect transmission attenuation weight of the event on the overseas warehouse node, generating a cross-level risk transmission path; the risk plan module 16 combines the inventory turnover data of the warehousing node and the replenishment time of the alternative supplier, and outputs a composite plan including customs clearance material updates, overseas warehouse allocation and supplier switching.

[0129] This embodiment achieves multi-dimensional risk warning and dynamic response through a modular architecture. The supply chain segmentation module 11 breaks down the supply chain into independent link nodes according to business links, achieving refined positioning of risk sources; the data acquisition module 12 integrates multi-source data to improve the comprehensiveness of risk assessment dimensions; the risk assessment module 13 integrates multi-source data to generate link node risk information and dynamically binds suppliers to risk indicators to enhance risk identification accuracy; the alternative supplier module 14 dynamically adjusts the number of alternative suppliers and further screens the optimal alternative resources to improve the scenario adaptability of supplier resources; the risk path transmission module 15 accurately captures the cascading risk diffusion path; and the risk plan module 16 ensures that the risk plan is highly consistent with the real-time business scenario, effectively improving the accuracy of risk identification and achieving coordinated optimization of the timeliness of risk warnings and the accuracy of risk plans.

[0130] In some embodiments, the data acquisition module 12 includes a data standardization unit, a spatiotemporal correlation unit, and a real-time update unit. The data standardization unit is used to perform standardization processing on multi-source data to obtain standardized data, including:

[0131] Extract the first key field from the supplier data and convert the first key field into a first structured data entry;

[0132] Process the return records in the consumer behavior data to obtain the category return rate, and perform NLP sentiment analysis on the text reviews in the consumer behavior data to generate sentiment polarity scores;

[0133] Construct a regional supplier mapping table by combining regional data, supplier data, and regional policies;

[0134] Extracting the second key field from the public opinion data and converting the second key field into a second structured data entry;

[0135] Convert logistics timeliness data into third-party structured data entries;

[0136] The spatiotemporal association unit is used to associate normalized data with link nodes;

[0137] The real-time update unit is used to dynamically update multi-source data through the stream processing framework.

[0138] In this embodiment, the data standardization unit refers to a functional module that performs unified format conversion and key information extraction on multi-source heterogeneous data. The first key field extraction is to convert the unstructured content in the supplier data into structured data entries, such as parsing the test report in PDF format into a data table containing the test date and results. The category return rate is obtained by statistically calculating the return records in the consumer behavior data by commodity category, such as counting the proportion of returns due to "logistics damage" under the "household appliances" category; NLP sentiment analysis of text evaluation refers to semantic parsing of user review texts and generating sentiment polarity scores (such as numerical scores in the range of -1 to 1) to quantify consumer satisfaction.

[0139] The regional supplier mapping table is constructed by associating suppliers' geographical locations, business coverage areas, and regional policies. For example, suppliers that meet the import qualifications of a certain country are bound to the target clearing node.

[0140] Preferably, the second key field extraction is structured storage of hot event descriptions, dissemination scope and related entities (such as brand names) in the public opinion data.

[0141] The third structured data entry converts the original trajectory information in the logistics timeliness data (such as "collected" and "in transit") into unified fields (such as node status and estimated delay days) to facilitate cross-node timeliness comparison and analysis.

[0142] The spatiotemporal association unit is used to dynamically bind standardized data to link nodes according to business logic, for example, associating a supplier's on-time delivery data to a procurement node, or mapping a region's logistics delay data to the corresponding distribution node.

[0143] The real-time update unit realizes dynamic data synchronization through the stream processing framework. For example, when the supplier qualification document changes, the weight recalculation of the risk assessment module 13 is triggered in real time; or when the keyword "new customs policy" appears in the public opinion data, the regional supplier mapping table is immediately updated and the transmission risk of the associated nodes is marked.

[0144] This embodiment can be understood as: eliminating format barriers of multi-source heterogeneous data through data standardization, and building a unified risk assessment input; time-space association ensures accurate matching of data and business nodes; and a real-time update mechanism ensures the timeliness of risk indicators. For example, when a supplier's quality inspection results change due to an update of the quality inspection report, the data standardization unit extracts the latest pass rate and updates the structured entries, the real-time update unit triggers the risk reassessment of the procurement node, and the time-space association unit synchronizes the updated data to the relevant link nodes, ultimately supporting the dynamic response of risk warnings. The data standardization unit eliminates data integration barriers, the time-space association unit achieves accurate mapping of business links and data sources, and the real-time update unit ensures the timeliness of risk assessment indicators, thereby improving the real-time response accuracy of risk warnings and the efficiency of cross-node data collaboration, and strengthening the dynamic management and control capabilities of supply chain risks.

[0145] In some embodiments, the risk path transmission module 15 includes a node association unit, a risk quantification unit, and a path adjustment unit. The node association unit is used to construct a logical dependency relationship between multiple link nodes to obtain a composite topology structure, including:

[0146] Generate an initial topology structure based on the node connection rules in the supply chain information;

[0147] Calculate the hierarchical association between each link node, build multi-directional association lines on the initial topology structure, and generate a composite topology structure;

[0148] Initializing risk transfer coefficients for multiple link nodes in a composite topology;

[0149] The risk quantification unit is used to quantify the risks of multiple link nodes based on the risk information and the composite topology structure, obtain the risk increment of each link node, and generate the final risk weight of each link node;

[0150] The path adjustment unit is used to generate risk transmission paths with different risk levels based on the final risk weight and composite topological structure.

[0151] In this embodiment, the initial topological structure represents the basic business flow between nodes; preferably, calculating the hierarchical association between each link node can be understood as quantifying the transmission impact potential between nodes by analyzing the physical connection strength between link nodes (such as the number of transportation routes between logistics nodes and distribution nodes), data interaction frequency (such as the number of inventory synchronization times between order nodes and warehousing nodes), and business dependence (such as the proportion of raw material supply from supplier nodes to production nodes); constructing multi-directional association lines on the initial topological structure can be understood as superimposing cross-level and cross-business functional connection paths on the initial topological structure (such as the reverse logistics channel of the after-sales node directly connected to the supplier node) to form a composite topological structure to support multi-dimensional risk transmission modeling.

[0152] The risk transmission coefficient is a parameter that quantifies the risk transmission ability between nodes in a composite topological structure. Preferably, it is obtained by integrating the transmission impact weight of the previous node on the current node (such as the impact ratio of procurement delays on warehouse supply interruptions), the collaborative risk weight of nodes at the same level (such as the cumulative effect of simultaneous delays in multiple logistics nodes), and the indirect transmission attenuation weight across levels (such as the attenuation rate of the cross-regional impact of sudden changes in customs clearance policies on overseas warehouse nodes).

[0153] The risk increment is based on the risk information of the current node (such as the probability of supplier delivery delay) and the risk transmission coefficient in the composite topology structure to calculate the risk diffusion value of the node to the associated nodes; the final risk weight can be used to divide the risk level of the node.

[0154] The path adjustment unit generates a risk transmission path based on the final risk weight and the associated lines in the composite topology structure. For example, it marks high-risk nodes whose weights exceed the threshold as key nodes of the path, and traces the transmission chain along the multi-directional associated lines to form a visual path containing nodes of different risk levels.

[0155] This embodiment can be understood as: modeling the multi-dimensional relationship between nodes through a composite topological structure, breaking through the limitations of the one-way transmission model; the risk quantification unit combines static risk values with dynamic transmission effects to accurately depict the cascading risk diffusion path; the path adjustment unit dynamically optimizes the transmission path based on weight thresholds to improve the targeted risk positioning and intervention. For example, in a cross-border e-commerce scenario, when a customs clearance node triggers a risk due to policy adjustments, the node association unit constructs a cross-level association line with the overseas warehouse node, the risk quantification unit calculates the transmission increment of the customs clearance delay on the warehouse inventory, and the path adjustment unit generates a transmission path including customs clearance, warehousing, and logistics nodes, providing a basis for the risk plan module 16. This embodiment uses a composite topological structure to multi-dimensionally model the logical dependency and transmission effect between link nodes, dynamically quantifies the node cascade risk weights based on the risk increment, and generates a multi-level risk transmission path based on the weight threshold, achieving accurate positioning and dynamic intervention of the risk transmission link, comprehensively improving the real-time nature of the supply chain cascade risk warning and the cross-node collaborative prevention and control capabilities.

[0156] See also Figure 2 In a second aspect, this embodiment further provides a supply chain full-link risk warning method based on big data, applicable to the risk warning platform of the first aspect, and comprising:

[0157] S101, obtaining supply chain information, dividing the supply chain information according to link nodes, and obtaining information of multiple link nodes;

[0158] S102, regularly acquiring multi-source data of each link node information according to a preset acquisition frequency and preprocessing it, mapping the preprocessed multi-source data with the link nodes and storing them;

[0159] S103: Generate risk information for each link node based on the pre-processed multi-source data, generate supplier quality information based on consumer behavior data and public opinion data, and bind the supplier quality information to the supplier;

[0160] S104. Screen a preset number of candidate suppliers from the database based on the risk information and link node information, and update the candidate suppliers to the link node information. The preset number is configured to be generated based on the real-time volatility of the current link node.

[0161] S105. Generate a risk transmission path based on the risk information and supply chain information. The risk transmission path includes multiple path nodes of different risk levels.

[0162] S106. Generate a risk plan based on the risk information, risk transmission path, and updated link node information.

[0163] In step S101, the link node division is completed through preset business rules. The preset business rules can be the upstream and downstream dependencies between the order flow nodes and the warehousing and sorting nodes, such as dividing the clearance nodes and overseas warehouse nodes of the Amazon platform as independent units.

[0164] In step S102, preprocessing eliminates differences in heterogeneous data formats. Furthermore, multi-source data can be standardized to generate unified structured data entries, ensuring input consistency and computational accuracy for the risk assessment module. The preprocessed multi-source data is then bound and stored to corresponding link nodes using the spatiotemporal association unit described in the previous embodiment. For example, a supplier's on-time delivery rate can be mapped to a procurement node.

[0165] In step S103, risk information is a quantitative indicator of node risk calculated based on multi-source data. The process of generating risk information can be understood as calculating the risk probability of link node information based on pre-processed multi-source data, fusing historical risk patterns with real-time emergency characteristics through weights, and outputting a dynamic risk value. For example, when the timeliness deviation rate of a logistics node exceeds a threshold due to sudden weather, a high risk level is generated in combination with the historical delay probability. Preferably, the process of generating and binding supplier quality information can be understood as integrating the return rate, evaluation sentiment polarity and public opinion propagation range in consumer behavior data, weightedly calculating a comprehensive quality score, dividing the levels according to the threshold, and binding to the supplier's unique identifier. For example, a supplier's negative review sentiment score is lower than -0.5 and is spread more than 100,000 times, triggering a quality level downgrade and binding to the supplier's file.

[0166] In step S104, higher real-time volatility indicates poorer node business stability, necessitating an increase in the number of potential suppliers to mitigate potential disruptions and enhance supply chain resilience. Preferred screening criteria include geographic coverage (e.g., overlap between the supplier's service area and the node's business area), quality rating (e.g., ISO-certified suppliers are prioritized), and historical on-time delivery performance (e.g., 95% or higher fulfillment rate over the past 90 days).

[0167] In step S105, the process of generating a risk transmission path can be understood as analyzing the business relevance between nodes in the path (e.g., the impact of a supply disruption at a procurement node on a warehousing node), combining it with risk information (e.g., the probability of supplier delays), and simulating the direction of risk diffusion to form a transmission chain from the risk source to the affected nodes. Preferably, each node in the risk transmission path is classified according to its final risk weight, for example, into high-risk, medium-risk, and low-risk categories, with high-risk nodes being directly impacted and medium-risk and low-risk nodes being indirectly affected. This allows the identification of core risk links and the development of prioritized intervention strategies.

[0168] In step S106, a risk contingency plan is generated based on the risk levels of nodes in the risk transmission path and information about potential suppliers. For example, if a customs clearance node experiences high risk due to a sudden policy change, the contingency plan module combines the potential supplier's replenishment timelines and inventory data at the storage node to generate a coordinated plan for updating customs clearance materials, transferring funds from overseas warehouses, and switching suppliers.

[0169] This embodiment achieves full-chain supply chain risk control through a collaborative mechanism that dynamically divides link nodes, models cross-node risk transmission, and dynamically updates alternative suppliers. The specific steps can be understood as follows: The supply chain is divided into independent nodes according to business stages, with each node associated with supplier quality information (such as quality inspection rates and policy compliance) and alternative suppliers in the database. In supplier management, a regional supplier mapping table is constructed using regional and public opinion data to account for marginal policy influences (such as import and export restrictions caused by trade wars) and dynamically adjust supplier qualification matching strategies. Alternative supplier screening dynamically determines a preset number based on real-time volatility and selects the best suppliers based on demand and price-performance ratio. For example, in the event of a sudden change in tariff policy, regional suppliers that comply with the new regulations can be deployed to replace existing large partners. The risk transmission path integrates public opinion risk (such as the popularity of policy keywords), enterprise operational risk (such as supplier financial crisis), and quality risk (such as the sampling failure rate). A cascading diffusion model is constructed based on the strength of business dependencies between nodes (such as the inventory supply relationship between procurement and warehousing nodes). Nodes are classified as high-risk, medium-risk, and low-risk, and the core risk sources are identified. The final plan module generates short-term emergency plans (such as rapid updates of customs clearance materials and emergency replenishment of regional suppliers) based on the transmission path priority and alternative supplier resources, achieving dynamic adaptation of risk response and business scenarios.

[0170] See also Figure 3 In some embodiments, obtaining multi-source data of each link node information and preprocessing it, and mapping and storing the preprocessed multi-source data with the link node includes:

[0171] S201. Extract a first key field from supplier data and convert the first key field into a first structured data entry;

[0172] S202: Processing the return records in the consumer behavior data to obtain the category return rate, and performing NLP sentiment analysis on the textual evaluations in the consumer behavior data to generate sentiment polarity scores;

[0173] S203. Constructing a regional supplier mapping table by combining regional data, supplier data, and regional policies;

[0174] S204: extract the second key field from the public opinion data and convert the second key field into a second structured data entry;

[0175] S205: Convert the logistics timeliness data into a third structured data entry;

[0176] S206: Generate node data information based on the first structured data entry, the second structured data entry, the third structured data entry, the regional supplier mapping table, and the sentiment polarity score according to a preset format;

[0177] S207, mapping and storing the node data information and the link node information;

[0178] Furthermore, it is evaluated whether the node data information meets the preset fluctuation condition. If not, the preset collection frequency of the multi-source data is adjusted.

[0179] In step S201, first key field extraction involves identifying and extracting core attributes relevant to risk assessment from supplier data, such as the certification validity period in supplier qualification documents, the percentage of non-conforming items in quality inspection results, and the historical fluctuation range of on-time delivery rates. First key fields are converted into first structured data entries by parsing unstructured documents. For example, the "on-time delivery rate" field is standardized into a numerical variable to facilitate quantitative analysis in subsequent modules.

[0180] In step S202, the category return rate is calculated by counting the proportion of return records for a specific product category to total sales orders. Preferably, NLP sentiment analysis of text reviews uses a pre-trained model to identify the sentiment of the review text and output a sentiment polarity score (e.g., -0.8 indicates strong negative sentiment) to reflect consumer satisfaction with the product or service.

[0181] In step S203, preferably, the regional supplier mapping table is constructed by associating the supplier registration location, service coverage area and regional policy, for example, binding suppliers with EU CE certification with the compliance requirements of the target clearing node to ensure rapid matching of alternative resources when regional policies change.

[0182] In step S204, the second key field extraction focuses on the event type, dissemination heat and related entities in the public opinion data, and converts them into a second structured data entry, for example, marking the "trade war" public opinion as a "current affairs public opinion" type and associating it with a list of affected suppliers.

[0183] In step S205, the original state in the logistics timeliness data is converted into a third structured data entry, for example, the deviation days between the actual transportation time and the promised timeliness is calculated based on the logistics node time sequence, which is used to evaluate the timeliness risk of the logistics node.

[0184] In step S206, the preset format requires that structured entries from different data sources be integrated into a unified JSON or database table structure, for example, supplier qualifications, category return rates, and logistics deviation days are aligned by field names to eliminate data heterogeneity.

[0185] In step S207, the mapping storage associates and stores node data information in the form of key-value pairs by assigning a unique identifier to each link node. For example, logistics time deviation data is associated with the distribution node ID. Preferably, when evaluating the preset fluctuation condition, the preset fluctuation condition is configured as the information entropy of the node data or the time series fluctuation amplitude threshold. For example, when the standard deviation of the sentiment polarity score collected for a node exceeds 0.5 for three consecutive times, it is determined to be a fluctuation anomaly, and the trigger collection frequency is adjusted from once per hour to once every 15 minutes to improve the real-time performance of the data.

[0186] It's important to note that within each link node of the full supply chain, supplier data covers multiple types of supply resources corresponding to each link. For example, within the procurement and inventory nodes, supplier data includes multiple optional suppliers for the same material, ensuring redundancy in raw material supply. Within the logistics and distribution nodes, supplier data involves multiple logistics carriers in different regions, enabling flexible scheduling of transportation routes. Within the warehousing and sorting nodes, supplier data can be subdivided into self-operated warehousing facilities and third-party warehousing service providers, with third-party service providers further managed in different levels based on operational scale and automation level. Within the after-sales node, supplier data includes self-operated service teams and external cooperative service outlets, with external outlets categorized and configured based on service qualifications (e.g., brand-authorized repairers, regional comprehensive service points). By managing supplier data at each node in layers based on resource type and service capabilities, the alternative supplier module can be used to quickly match optimal alternatives in risk scenarios.

[0187] This embodiment improves the real-time nature and data consistency of risk warnings through standardized processing of multi-source heterogeneous data and dynamic collection frequency adjustment: First, structured data entries ensure that core indicators such as supplier qualifications and logistics timeliness can be quantified and analyzed; regional supplier mapping tables enhance cross-regional risk response capabilities; the mapping of node data information and link nodes ensures deep coupling between business scenarios and risk indicators, supporting risk transmission path modeling; preset fluctuation conditions trigger dynamic optimization of collection strategies to ensure timely capture of abnormal fluctuations. The above technical solution achieves accurate screening of alternative suppliers and scenario adaptability of plan generation through hierarchical management of supplier resources at each node and multi-dimensional data integration, effectively improving the agility and reliability of risk prevention and control throughout the entire chain.

[0188] See also Figure 4 In some embodiments, generating risk information for each link node information based on pre-processed multi-source data includes:

[0189] S301, sorting the influence coefficients of the pre-processed multi-source data associated with the link node information according to the AHP algorithm to obtain the influence coefficient of each data type in the multi-source data;

[0190] S302. Calculate the risk weight of each data type using a linear normalization function based on the impact coefficient, and record it as the initial risk weight;

[0191] S303: Calculate the risk probability of each data type, where the risk probability is calculated using a Poisson distribution model based on the number of risk occurrences in historical data of the same data type at the same link node.

[0192] S304. Merge the risk probability with the initial risk weight to form a basic risk weight;

[0193] S305, extracting burst information from multi-source data using a CUSUM algorithm, where the burst information includes policy burst characteristics, promotion burst characteristics, and regional burst characteristics;

[0194] S306, converting the emergency information into emergency risk weights using a fuzzy logic algorithm;

[0195] S307. Combine the basic risk weight and the sudden risk weight to obtain risk information.

[0196] In step S301, preferably, the AHP algorithm quantifies the degree of influence of multi-source data on link node risks by constructing a hierarchical model, wherein the link node serves as the target layer and its associated multi-source data type serves as the criterion layer. A judgment matrix is constructed through expert experience and historical data to calculate the influence coefficient of each data type.

[0197] In step S302, a linear normalization function is used to map the impact coefficient to a standardized interval and ensure that the sum of the weights is a fixed value, thereby forming an initial risk weight.

[0198] In step S303, preferably, the risk probability can be obtained by calculating the frequency of historical risk occurrence using a Poisson distribution model, for example, the average occurrence rate of past risk events of a certain data type under the same link node is used as the basis for estimating the current risk probability.

[0199] In step S304, the risk probability and the initial risk weight can be fused by weighted geometric mean, Euclidean distance weighting, attention mechanism fusion or neural network model fusion. Preferably, the weighted geometric mean method is adopted to dynamically integrate the initial risk weight and the risk probability according to a preset ratio. Specifically, the initial risk weight reflects the static impact of the data type on the node, while the risk probability characterizes the dynamic frequency of historical risk occurrence. By setting the weighting coefficient of the basic weight ratio and the risk probability ratio, the geometric mean is calculated as the basic risk weight, which not only retains the benchmark weight assigned by expert experience, but also strengthens the decision-making priority of high-frequency risk data.

[0200] In step S305, the CUSUM algorithm detects sudden fluctuations by monitoring the cumulative deviations of time series data, such as a sharp increase in order volume during a promotion or an unusual jump in the frequency of policy API calls. This sudden information includes policy, promotion, and geographic characteristics. The policy characteristics monitor the frequency of text changes in the customs policy API; the promotion characteristics detect sudden changes in the second-order derivative of the sales time series, accurately capturing flash sales; and the geographic characteristics analyze the spatiotemporal clustering of natural disaster warning signals.

[0201] In step S306, preferably, a fuzzy logic algorithm converts unstructured emergency features (such as the frequency of policy text updates) into membership scores, which are mapped into emergency risk weights based on a preset rule base.

[0202] In step S307, the basic risk weight and the sudden risk weight are fused by weighted geometric mean, Euclidean distance weighting, attention mechanism fusion or neural network model fusion. Preferably, the attention mechanism is adopted to dynamically allocate the fusion ratio of the basic and sudden weights according to the intensity of the sudden feature. Specifically, by constructing an attention scoring model, analyzing the characteristics such as the frequency of policy mutations, promotion scale or disaster warning level in the sudden information, calculating the attention coefficient of the sudden risk weight, and performing weighted summation of the two types of weights according to the attention coefficient. For example, when the policy suddenly changes, the proportion of the sudden weight in the final risk information is increased to achieve adaptive adjustment of the risk weight according to the scenario.

[0203] The technical solution of this embodiment can be understood as follows: risk information reflects the multidimensional risk characteristics of multi-source data in link nodes, including each risk probability corresponding to the multi-source data, which can be understood as the risk weight of the data type, and is evaluated based on the link node information. The basic risk weight is obtained through a hierarchical assessment of the impact of the multi-source data types associated with the link node. First, the AHP algorithm is used to sort the data types by their influence coefficients and quantify the static weights. Then, the historical risk probability is statistically analyzed using the Poisson distribution model, and a weighted geometric average is used to generate a comprehensive weight.

[0204] The sudden risk weights are designed for scenarios such as policy changes, promotional peaks, or regional disasters. The CUSUM algorithm extracts sudden risk features, which are then converted into dynamic weights through fuzzy logic. The final risk information is dynamically adjusted through an attention mechanism to combine the two weights. For example, the sudden risk weight is increased in the event of a policy change, achieving an adaptive balance between normal risk patterns and responses to sudden scenarios. For example, at the procurement node, the basic weight of logistics timeliness data reflects the risk of conventional transportation efficiency. When the sudden risk weight of tariff policy changes is added, a composite risk indicator is generated that includes supplier switching priorities and emergency customs clearance strategies, representing the risk information.

[0205] This embodiment improves the accuracy and scenario adaptability of risk information generation through hierarchical assessment and dynamic emergency response mechanisms. The AHP algorithm and Poisson distribution are combined to quantify the historical impact weights of multi-source data, strengthening the decision-making priority of high-frequency risk factors. The CUSUM algorithm and fuzzy logic accurately capture sudden changes in policies, promotional peaks, and other sudden events, converting them into dynamic risk weights. The attention mechanism adaptively adjusts the fusion ratio of basic and sudden event weights based on the intensity of the sudden event, achieving a dynamic balance between normal risk patterns and sudden event response. Ultimately, this generates multi-dimensional risk information to support the precise triggering of risk plans.

[0206] In some embodiments, the pre-processed multi-source data associated with the link node information is sorted by influence coefficient according to the AHP algorithm, and the influence coefficient of each data type in the multi-source data is obtained, including:

[0207] Constructing a judgment matrix corresponding to the node type of the current link node, wherein the matrix elements of the judgment matrix are generated by the relative importance of the data types of the multi-source data;

[0208] Calculate the maximum eigenvalue of the judgment matrix and its associated eigenvector, where each component of the eigenvector corresponds to a data type;

[0209] Normalize the eigenvectors to obtain the influence coefficients of multiple data types;

[0210] Calculate the risk probability of each data type. The risk probability is configured to be calculated using the Poisson distribution model using the number of risk occurrences in the historical data information of the same data type at the same link node, including:

[0211] Count the average number of risk events of the current data type per unit time based on historical data information;

[0212] The probability of at least one risk event occurring in the current time window is calculated using the Poisson distribution probability function. This is expressed as the risk probability using formula (1). Formula (1) is as follows:

[0213] ;

[0214] In formula (1), is the risk probability, is the number of risk events that occur, is the base of natural logarithms, is the average number of risk events occurring per unit time;

[0215] The CUSUM algorithm is used to extract sudden information from multi-source data. The sudden information includes policy sudden features, promotion sudden features, and regional sudden features.

[0216] Cumulative statistics are constructed for the time series data stream in multi-source data. When the cumulative statistics exceed the preset control limit, it is marked as a mutation point, which is expressed by formula (2). Formula (2) is as follows:

[0217] ;

[0218] In formula (2), For Cumulative statistics at time, For Cumulative statistics at time, For The observed value at time, is the benchmark mean, is the permissible offset parameter;

[0219] Generating at least one of a policy burst feature, a promotion burst feature, or a regional burst feature according to the data type classification corresponding to the mutation point;

[0220] The conversion of emergency information into emergency risk weights using fuzzy logic algorithms includes:

[0221] Establishing a fuzzy rule base, wherein the input variables of the fuzzy rule base are the intensity of the burst feature and the duration of the burst feature, and the output variable of the fuzzy rule base is the burst risk weight;

[0222] The fuzzy reasoning results are converted into numerical sudden risk weights through defuzzification operations.

[0223] In this embodiment, when the influence coefficients are sorted according to the AHP algorithm, the matrix elements of the judgment matrix are generated by the relative importance of the data types of the multi-source data, where the relative importance is determined according to the business characteristics of the current link node. For example, the comparison scale of the logistics timeliness data and the supplier data in the procurement node is higher than that in the warehousing node.

[0224] The normalization of the eigenvector is achieved by dividing each component by the vector sum so that the sum of the influence coefficients is a fixed value.

[0225] When calculating the risk probability of each data type, the Poisson distribution model The parameters are determined by statistically analyzing the average number of risk events within a unit time window in historical data information, such as the daily average number of delivery delays at logistics nodes in the past 30 days.

[0226] When extracting burst information from multi-source data using the CUSUM algorithm, the allowable offset parameter in the CUSUM algorithm Preset according to business scenarios, such as fluctuations in logistics timeliness Set to 1.5 times the daily average deviation. The sliding window mean of historical data, for example, the moving average of the observations in the previous 24 hours is taken as the current The value of .

[0227] When converting emergency information into emergency risk weights using a fuzzy logic algorithm, the fuzzy rule base utilizes a Mamdani-type fuzzy inference system. The membership functions of the input variables are triangular. For example, a "high" intensity policy emergency corresponds to a triangular function with a vertex of 10 daily updates and a base span of 5-15 updates. Defuzzification is performed using the centroid method to calculate the centroid of the fuzzy output region and convert it into an emergency risk weight.

[0228] This embodiment improves the accuracy and scenario adaptability of risk factor calculations through hierarchical impact assessment and dynamic burst feature extraction. The AHP algorithm constructs a judgment matrix based on node characteristics to quantify the static impact weights of multi-source data. The Poisson distribution model statistically analyzes the frequency of historical risk occurrences and strengthens the probability weights of high-frequency events. The CUSUM algorithm combines a sliding window mean and offset parameters to accurately capture sudden policy changes and other sudden events. The Mamdani fuzzy rule base uses triangular membership functions and the centroid method to defuzzify, enabling the reliable conversion of unstructured sudden information into risk weights.

[0229] See also Figure 5In some embodiments, generating supplier quality information based on consumer behavior data and public opinion data, and binding the supplier quality information to the supplier includes:

[0230] S401. Process the return records in the consumer behavior data and calculate the return rate of each supplier's category;

[0231] S402: extracting textual evaluations from consumer behavior data, performing NLP sentiment analysis, and generating supplier quality satisfaction scores;

[0232] S403. Extract key fields from public opinion data to identify supplier-related quality events;

[0233] S404. Calculate the public opinion risk value based on the frequency and spread of quality incidents;

[0234] S405. Weighted integration of the product category return rate, quality satisfaction score, and public opinion risk value to generate a comprehensive quality score;

[0235] S406. Classify the quality grades according to the comprehensive quality scores and obtain supplier quality information;

[0236] S407, associating the supplier quality information with the corresponding supplier and storing it;

[0237] and, triggering a screening mechanism for alternative suppliers when the quality level falls below a preset level threshold;

[0238] In addition, supplier quality information will be updated according to preset cycles based on newly added consumer behavior data and public opinion data.

[0239] In step S401, the category return rate is calculated by counting the proportion of returns to total sales for each supplier in a specific product category, reflecting the actual problem rate of the product. For example, the return rate of diapers for a maternity and baby supplier can be quantified by the ratio of monthly returns to total shipments.

[0240] In step S402, the quality satisfaction score can be generated by performing sentiment analysis on the consumer review text using a pre-trained NLP model. The model outputs the sentiment polarity score and linearly maps it to a preset score range to quantify the consumer's subjective satisfaction with the product quality.

[0241] In step S403, key field extraction identifies the quality event type and propagation scope associated with the supplier from the public opinion data, for example, by capturing the keyword "battery failure" through text matching technology and associating it with the corresponding supplier.

[0242] In step S404, the public opinion risk value is calculated based on the frequency of occurrence of quality events per unit time, the breadth of dissemination and the severity of the event type. The breadth of dissemination is quantified by the logarithmic transformation of the forwarding volume or the comment volume, and the severity of the event is adjusted by the preset coefficient of the quality problem or service problem type.

[0243] In step S405, the comprehensive quality score is linearly integrated with the category return rate, quality satisfaction score and public opinion risk value through preset weights, where the return rate and satisfaction score are used as negative indicators to deduct points, and the public opinion risk value is used as a positive indicator to increase points, and finally a comprehensive quality score of 0-100 points is generated. The lower the comprehensive quality score, the worse the quality.

[0244] In step S406, quality tiers are assigned based on preset comprehensive quality score thresholds. Suppliers with comprehensive quality scores below the minimum threshold are assigned to a lower quality tier, triggering the alternative supplier screening mechanism described in the aforementioned embodiment. Preferably, the preset comprehensive quality score thresholds are determined based on industry standards, historical supplier score distributions, and expert experience, for example, by combining historical data quantile thresholds to ensure that the tiers match actual business risk tolerances.

[0245] In step S407, supplier quality information is stored in conjunction with the supplier's unique identifier and dynamically updated at a preset interval. During updates, the return rate, satisfaction score, and public opinion risk value are recalculated to ensure data timeliness.

[0246] Optionally, the screening mechanism for candidate suppliers may include the following steps:

[0247] When the supplier quality level is lower than the preset level threshold, a candidate set of alternative suppliers that meet the business requirements of the current link node is retrieved from the database. The screening criteria include:

[0248] 1) The quality level is not lower than that of the current supplier and has passed the qualification review;

[0249] 2) The geographic coverage matching degree is calculated as the overlap ratio between the supplier’s service scope and the current node’s business area, and the matching degree must be ≥ the preset threshold;

[0250] 3) The historical on-time delivery rate is calculated based on historical order fulfillment data, and suppliers ranked in the top 20% in terms of on-time delivery rate are given priority;

[0251] The preset quantity is dynamically adjusted based on the real-time volatility of the current node. For every 10% increase in the real-time volatility, the preset quantity increases by 1.

[0252] Finally, a weighted scoring model is constructed based on quality grade, geographical matching and punctuality. A preset number of alternative suppliers with the highest ranking are selected by total score and updated to the link node information.

[0253] This embodiment achieves accurate supplier quality assessment and risk response through the integration of multi-dimensional indicators and a dynamic update mechanism. Category return rates quantify the actual product usage problem rate, NLP sentiment analysis analyzes consumer subjective reviews, and public opinion risk values capture the impact of event transmission. Weighted fusion generates a comprehensive quality score. Preset thresholds divide quality levels into triggers for alternative supplier screening, and a dynamic update mechanism ensures data timeliness. For example, when the score falls below the threshold, alternative solution matching is initiated. Combined with periodic reassessments, quality status is synchronized in real time, improving the efficiency of supply chain risk response.

[0254] In some embodiments, a preset number of candidate suppliers are screened in a database based on risk information and link node information, and the candidate suppliers are updated to the link node information. The preset number is configured to be generated based on the real-time volatility of the current link node, including:

[0255] Calculate the real-time volatility of the current link node. The real-time volatility is obtained by converting the degree of change of key indicators in the link node information.

[0256] Dynamically adjust the preset quantity according to the real-time volatility;

[0257] Prioritize suppliers in the database based on pre-set screening criteria, including geographic coverage match, supplier quality rating, historical on-time delivery rate, and cost premium;

[0258] A preset number of suppliers to be replaced with high rankings are selected and recorded as candidate suppliers, and are updated to the link node information.

[0259] In this embodiment, the real-time volatility is calculated by statistically analyzing the temporal variation of key indicators of link nodes (such as order volume and inventory consumption rate). Specifically, the volatility can be quantified based on the ratio of the standard deviation to the mean of the indicator within the moving time window.

[0260] The preset quantity is dynamically adjusted according to the real-time volatility. The greater the real-time volatility, the greater the preset quantity.

[0261] In the database, the alternative suppliers are prioritized according to the preset screening conditions. The screening conditions include: the degree of match with the geographical coverage of the current link node; the supplier quality grade is not lower than the preset requirements; the historical delivery on-time rate reaches the preset standard; the cost premium does not exceed the preset ratio. Among them, the geographical coverage match is calculated by the overlap rate between the supplier's service area and the current node's business coverage. For example, if a logistics node covers the East China region, the supplier whose service scope includes this region will be screened; the cost premium refers to the excess ratio of the alternative supplier's quotation relative to the current supplier, which must be lower than the preset upper limit. Priority sorting is achieved through a weighted comprehensive score, for example, the geographical match weight is 40%, the quality grade weight is 30%, the delivery on-time rate weight is 20%, and the cost premium weight is 10%. The one with the highest total score is given priority.

[0262] Preferably, the alternative supplier switching mechanism is triggered when the basic risk weight and the sudden risk weight in the risk information exceed a dynamic threshold. The dynamic threshold is adjusted inversely according to the real-time volatility, that is, when the real-time volatility increases, the dynamic threshold is correspondingly reduced. Specifically, the alternative supplier switching mechanism is triggered when the weighted sum of the basic risk weight and the sudden risk weight exceeds the threshold.

[0263] This embodiment improves the accuracy and responsiveness of supplier selection through a dynamic control mechanism driven by real-time volatility. The number of supplier candidates is dynamically adjusted based on the volatility of key indicators, ensuring resource redundancy in high-volatility scenarios. Multi-dimensional screening criteria combined with weighted priority sorting ensure a balanced optimization of geographic coverage, quality level, and cost control. Dynamic thresholds adjust inversely with real-time volatility, triggering rapid switching when the total risk weight exceeds the threshold, enhancing the supply chain's risk resilience.

[0264] In some embodiments, a risk transmission path is generated based on risk information and supply chain information. The risk transmission path includes multiple path nodes of different risk levels, including:

[0265] Generate an initial topology structure based on the node connection rules in the supply chain information;

[0266] Calculating the hierarchical correlation between each link node, where the hierarchical correlation is configured to be generated based on the physical connection strength, data interaction frequency, and service dependency between multiple link nodes;

[0267] Construct multi-directional association lines on the initial topology structure according to the hierarchical association degree to generate a composite topology structure;

[0268] Initializing risk conduction coefficients for multiple link nodes in a composite topology structure, the risk conduction coefficients are configured to be obtained by the following steps:

[0269] Calculate the conduction influence weights of the preceding link node and the subsequent link node at the current link node;

[0270] Also, calculate the collaborative risk weights of other link nodes at the same level as the current link node;

[0271] and, calculating the indirect conduction attenuation weights between multiple link nodes across the hierarchy and the current link node;

[0272] The risk transmission coefficient is obtained by integrating the transmission impact weight, the synergistic risk weight and the indirect transmission attenuation weight;

[0273] Based on the risk information and the composite topology, the risks of multiple link nodes are quantified to obtain the risk increment of each link node, and the final risk weight of each link node is generated, which is expressed by formula (3). Formula (3) is as follows:

[0274] ;

[0275] In formula (3), is the final risk weight of the current link node, is the initial risk value of the current link node, is the risk transmission coefficient, For risk increment;

[0276] Based on the final risk weight and composite topological structure, risk transmission paths with different risk levels are generated.

[0277] In this embodiment, hierarchical correlation refers to the closeness of the business relationship between link nodes, which is obtained by weighted calculation of the physical connection strength (such as the number of transportation routes between logistics nodes and warehousing nodes), data interaction frequency (such as the number of inventory synchronization times between order nodes and warehousing nodes), and business dependence (such as the proportion of raw material supply from supplier nodes to production nodes).

[0278] Constructing multi-directional association lines can be understood as superimposing cross-level and cross-business functional connection paths (such as reverse logistics channels that directly connect after-sales nodes to supplier nodes) on the basis of the initial topological structure, ultimately forming a composite topological structure that supports multi-dimensional risk transmission.

[0279] In the process of obtaining the risk transmission coefficient, the transmission impact weight represents the ability of the previous node to transfer risk to the current node, such as the probability of warehouse out-of-stock due to procurement delays. The collaborative risk weight quantifies the cumulative effect of risks occurring simultaneously at nodes on the same level, such as the cascading impact of delays at multiple logistics nodes. The indirect transmission attenuation weight measures the risk attenuation rate of nodes across multiple levels, such as the indirect impact of sudden customs clearance policies on overseas warehouse nodes. The risk transmission coefficient is obtained by weighted fusion of the transmission impact weight, collaborative risk weight, and indirect transmission attenuation weight, and is used to quantify the overall strength of risk transmission between link nodes.

[0280] Incremental risk Based on the risk information of the current link node (such as the probability of supplier delivery delay) and the risk transmission coefficient The product of is calculated to reflect the risk diffusion value of the associated link nodes. The final risk weight The initial risk value of the current link node With incremental risk Weighted sum generation.

[0281] When risk information or supply chain information is monitored to change, the risk transmission path is updated.

[0282] This embodiment achieves accurate identification and real-time updating of supply chain risk transmission paths through composite topology modeling and a dynamic transmission calculation mechanism. Based on the physical connection strength, data interaction frequency, and business dependency, hierarchical correlation is calculated, and multi-directional correlation lines are constructed to form a composite topology. Risk transmission coefficients are generated by integrating transmission impact weights, collaborative risk weights, and indirect transmission attenuation weights to quantify the intensity of risk diffusion between nodes. Final risk weights are calculated by combining initial risk values and risk increments, dynamically generating multi-level transmission paths. Dynamic updates of topology structures and transmission coefficients improve the timeliness of cascading risk warnings and the accuracy of risk response plans.

[0283] By adopting the above technical solution, the present invention is different from the existing technology and has the following beneficial effects:

[0284] The present invention realizes full-dimensional monitoring of supply chain risks by dividing supply chain information according to link nodes and collecting multi-source data; combines supplier data, consumer behavior data and public opinion data to generate dynamically bound supplier quality information, thereby improving the objectivity and real-time nature of supplier evaluation; dynamically adjusts the number of alternative suppliers based on real-time volatility, thereby enhancing the elastic response capabilities of supply chain nodes; accurately captures the transmission path and cascade effect of risks between supply chain nodes by constructing a composite topological structure and a risk transmission coefficient calculation mechanism; utilizes a risk quantification model to fuse basic risk weights and emergency risk weights, thereby ensuring the comprehensiveness of risk assessment and the sensitivity of response to emergencies; and finally generates risk plans adapted to link scenarios through multi-dimensional data fusion, thereby realizing closed-loop management of the entire process from risk identification, transmission analysis to plan formulation, effectively improving the timeliness of supply chain risk warnings and the accuracy of disposal strategies.

[0285] Finally, it should be noted that although the above embodiments have been described in the specification and drawings of this application, this does not limit the scope of patent protection of this application. All technical solutions generated by replacing or modifying equivalent structures or equivalent processes based on the essential concepts of this application using the contents recorded in the specification and drawings of this application, as well as directly or indirectly implementing the technical solutions of the above embodiments in other related technical fields, are included in the scope of patent protection of this application.

Claims

1. A supply chain full-link risk early warning platform based on big data, characterized by: include: A supply chain partitioning module is used to obtain supply chain information and partition the supply chain information according to link nodes to obtain a plurality of link node information; A data collection module is used to obtain multi-source data of each link node information, wherein the multi-source data includes supplier data, consumer behavior data, logistics timeliness data, regional data and public opinion data; A risk assessment module is configured to generate risk information for each link node based on the multi-source data, generate supplier quality information based on the consumer behavior data and public opinion data, and bind the supplier quality information to the supplier, including: Process the return records in the consumer behavior data and calculate the return rate of each supplier's category; Extracting textual evaluations from the consumer behavior data for NLP sentiment analysis to generate supplier quality satisfaction scores; Extract key fields from the public opinion data to identify supplier-related quality events; Calculate the public opinion risk value based on the frequency and spread of the quality incident; The return rate, quality satisfaction score and public opinion risk value of the product category are weighted and integrated to generate a comprehensive quality score; Divide the quality grades according to the comprehensive quality score to obtain the supplier quality information; Associating and storing the supplier quality information with the corresponding supplier; and, when the quality level is lower than a preset level threshold, triggering a screening mechanism for alternative suppliers; and, updating the supplier quality information according to a preset period based on newly added consumer behavior data and public opinion data; an alternative supplier module, configured to screen a preset number of alternative suppliers from a database based on the risk information and the link node information, and update the alternative suppliers to the link node information, wherein the preset number is configured to be generated based on the real-time volatility of the current link node; A risk path transmission module, configured to generate a risk transmission path based on the risk information and supply chain information, wherein the risk transmission path includes a plurality of path nodes of different risk levels; A risk plan module, configured to generate a risk plan based on the risk information, the risk transmission path, and the updated link node information; The risk assessment module is further configured to perform the following steps: Acquire multi-source data of each link node information and perform preprocessing, and map and store the preprocessed multi-source data with the link node; Generating risk information of each link node information according to the pre-processed multi-source data includes: According to the link node information, the pre-processed multi-source data associated therewith are sorted by influence coefficient according to the AHP algorithm to obtain the influence coefficient of each data type in the multi-source data; Calculate the risk weight of each data type using a linear normalization function according to the impact coefficient, and record it as the initial risk weight; Calculating the risk probability of each data type, wherein the risk probability is configured to be obtained by calculating the number of risk occurrences in historical data information of the same data type at the same link node using a Poisson distribution model; Merging the risk probability with the initial risk weight to form a basic risk weight; Extracting burst information from the multi-source data using a CUSUM algorithm, wherein the burst information includes policy burst characteristics, promotion burst characteristics, and regional burst characteristics; Converting the emergency information into emergency risk weights using a fuzzy logic algorithm; The basic risk weight is combined with the sudden risk weight to obtain the risk information.

2. The big data-based supply chain full-link risk early warning platform according to claim 1 is characterized in that: The data acquisition module includes: The data standardization unit is used to perform standardization processing on the multi-source data to obtain standardized data, including: Extracting a first key field from the supplier data and converting the first key field into a first structured data entry; Processing the return records in the consumer behavior data to obtain the category return rate, and performing NLP sentiment analysis on the text reviews in the consumer behavior data to generate sentiment polarity scores; Constructing a regional supplier mapping table by combining the regional data with supplier data and regional policies; Extracting a second key field from the public opinion data and converting the second key field into a second structured data entry; Convert logistics timeliness data into third-party structured data entries; a spatiotemporal association unit for associating normalized data with link nodes; The real-time update unit is used to dynamically update multi-source data through the stream processing framework.

3. The big data-based supply chain full-link risk early warning platform according to claim 1 is characterized in that: The risk path transmission module includes: The node association unit is used to build the logical dependency relationship of multiple link nodes to obtain a composite topology structure, including: Generate an initial topology structure based on the node connection rules in the supply chain information; Calculating the hierarchical association between each of the link nodes, constructing multi-directional association lines on the initial topology structure, and generating a composite topology structure; Initializing risk transfer coefficients for a plurality of link nodes in the composite topology structure; a risk quantification unit, configured to quantify the risks of a plurality of link nodes according to the risk information and the composite topology structure, obtain a risk increment for each link node, and generate a final risk weight for each link node; The path adjustment unit is used to generate risk transmission paths with different risk levels according to the final risk weight and the composite topological structure.

4. A supply chain full-link risk early warning method based on big data, characterized by: Applicable to the risk warning platform according to any one of claims 1 to 3, the method comprising: Acquire supply chain information, divide the supply chain information according to link nodes, and obtain information of multiple link nodes; Regularly acquiring multi-source data of each link node information according to a preset acquisition frequency and preprocessing it, and mapping and storing the preprocessed multi-source data with the link node; Generating risk information of each link node information based on the pre-processed multi-source data, generating supplier quality information based on the consumer behavior data and public opinion data, and binding the supplier quality information to the supplier, including: Process the return records in the consumer behavior data and calculate the return rate of each supplier's category; Extracting textual evaluations from the consumer behavior data for NLP sentiment analysis to generate supplier quality satisfaction scores; Extract key fields from the public opinion data to identify supplier-related quality events; Calculate the public opinion risk value based on the frequency and spread of the quality incident; The return rate, quality satisfaction score and public opinion risk value of the product category are weighted and integrated to generate a comprehensive quality score; Divide the quality grades according to the comprehensive quality score to obtain the supplier quality information; Associating and storing the supplier quality information with the corresponding supplier; and, when the quality level is lower than a preset level threshold, triggering a screening mechanism for the alternative suppliers; and, updating the supplier quality information according to a preset period based on newly added consumer behavior data and public opinion data; Screening a preset number of candidate suppliers in a database based on the risk information and the link node information, and updating the candidate suppliers to the link node information, wherein the preset number is configured to be generated based on the real-time volatility of the current link node; generating a risk transmission path based on the risk information and supply chain information, wherein the risk transmission path includes a plurality of path nodes of different risk levels; Generate a risk plan based on the risk information, the risk transmission path, and the updated link node information; The method further comprises: Acquire multi-source data of each link node information and perform preprocessing, and map and store the preprocessed multi-source data with the link node; Generating risk information of each link node information according to the pre-processed multi-source data includes: According to the link node information, the pre-processed multi-source data associated therewith are sorted by influence coefficient according to the AHP algorithm to obtain the influence coefficient of each data type in the multi-source data; Calculate the risk weight of each data type using a linear normalization function according to the impact coefficient, and record it as the initial risk weight; Calculating the risk probability of each data type, wherein the risk probability is configured to be obtained by calculating the number of risk occurrences in historical data information of the same data type at the same link node using a Poisson distribution model; Merging the risk probability with the initial risk weight to form a basic risk weight; Extracting burst information from the multi-source data using a CUSUM algorithm, wherein the burst information includes policy burst characteristics, promotion burst characteristics, and regional burst characteristics; Converting the emergency information into emergency risk weights using a fuzzy logic algorithm; The basic risk weight is combined with the sudden risk weight to obtain the risk information.

5. The big data-based supply chain full-link risk early warning method according to claim 4 is characterized in that: Acquiring multi-source data of each link node information and preprocessing it, and mapping and storing the preprocessed multi-source data with the link node includes: Extracting a first key field from the supplier data and converting the first key field into a first structured data entry; Processing the return records in the consumer behavior data to obtain the category return rate, and performing NLP sentiment analysis on the text reviews in the consumer behavior data to generate sentiment polarity scores; Constructing a regional supplier mapping table by combining the regional data with supplier data and regional policies; Extracting a second key field from the public opinion data and converting the second key field into a second structured data entry; Convert logistics timeliness data into third-party structured data entries; Generate node data information according to a preset format based on the first structured data entry, the second structured data entry, the third structured data entry, the regional supplier mapping table, and the sentiment polarity score; Mapping and storing the node data information and the link node information; Furthermore, it is evaluated whether the node data information satisfies a preset fluctuation condition, and if not, the preset collection frequency of the multi-source data is adjusted.

6. The big data-based supply chain full-link risk early warning method according to claim 4 is characterized in that: According to the link node information, the pre-processed multi-source data associated therewith is sorted by influence coefficient according to the AHP algorithm, and the influence coefficient of each data type in the multi-source data is obtained, including: Constructing a judgment matrix corresponding to the node type of the current link node, wherein the matrix elements of the judgment matrix are generated according to the relative importance of the data types of the multi-source data; Calculating the maximum eigenvalue of the judgment matrix and its associated eigenvector, where each component of the eigenvector corresponds to a data type; Normalizing the characteristic vectors to obtain influence coefficients of multiple data types; Calculating the risk probability of each data type, wherein the risk probability is configured to be obtained by calculating the number of risk occurrences in historical data information of the same data type at the same link node using a Poisson distribution model, including: Counting the average number of risk events of the current data type within a unit time based on the historical data information; The probability of at least one risk event occurring within the current time window is calculated using the Poisson distribution probability function, and is expressed as the risk probability using formula (1). Formula (1) is as follows: ； In formula (1), is the risk probability, is the number of risk events that occur, is the base of natural logarithms, is the average number of risk events occurring per unit time; The CUSUM algorithm is used to extract the burst information from the multi-source data, wherein the burst information includes policy burst characteristics, promotion burst characteristics, and regional burst characteristics. A cumulative statistic is constructed for the time series data stream in the multi-source data. When the cumulative statistic exceeds a preset control limit, it is marked as a mutation point, which is expressed by formula (2). Formula (2) is as follows: ； In formula (2), For Cumulative statistics at time, For Cumulative statistics at time, For The observed value at time, is the benchmark mean, is the permissible offset parameter; Generating at least one of the policy burst feature, the promotion burst feature, or the regional burst feature according to the data type classification corresponding to the mutation point; Converting the emergency information into emergency risk weights using a fuzzy logic algorithm includes: Establishing a fuzzy rule base, wherein the input variables of the fuzzy rule base are the intensity of the burst feature and the duration of the burst feature, and the output variable of the fuzzy rule base is the burst risk weight; The fuzzy reasoning result is converted into the sudden risk weight in numerical form through defuzzification operation.

7. The big data-based supply chain full-link risk early warning method according to claim 4 is characterized in that: Screening a preset number of candidate suppliers in a database based on the risk information and the link node information, and updating the candidate suppliers to the link node information, wherein the preset number is configured to be generated based on the real-time volatility of the current link node, includes: Calculating the real-time volatility of the current link node, where the real-time volatility is obtained by converting the degree of change of the key indicators in the link node information; Dynamically adjust the preset amount according to the size of the real-time volatility; Prioritizing the suppliers to be replaced in the database according to preset screening criteria, wherein the preset screening criteria include geographic coverage matching, supplier quality level, historical on-time delivery rate, and cost premium; A preset number of suppliers to be replaced that are ranked high are selected and recorded as candidate suppliers, and are updated to the link node information.

8. The big data-based supply chain full-link risk early warning method according to claim 4 is characterized in that: A risk transmission path is generated based on the risk information and supply chain information. The risk transmission path includes multiple path nodes of different risk levels, including: Generate an initial topology structure based on the node connection rules in the supply chain information; Calculating a hierarchical correlation between each of the link nodes, wherein the hierarchical correlation is configured to be generated by physical connection strength, data interaction frequency, and service dependency between the plurality of link nodes; Constructing multi-directional association lines on the initial topological structure according to the hierarchical association degree to generate a composite topological structure; Initializing risk conduction coefficients for a plurality of link nodes in the composite topology structure, wherein the risk conduction coefficients are configured to be obtained by: Calculate the conduction influence weights of the preceding link node and the subsequent link node at the current link node; Also, calculate the collaborative risk weights of other link nodes at the same level as the current link node; and, calculating the indirect conduction attenuation weights between multiple link nodes across the hierarchy and the current link node; The risk transmission coefficient is obtained by fusing the transmission impact weight, the collaborative risk weight and the indirect transmission attenuation weight; The risks of multiple link nodes are quantified according to the risk information and the composite topology structure to obtain the risk increment of each link node, and the final risk weight of each link node is generated, which is expressed by formula (3). Formula (3) is as follows: ； In formula (3), is the final risk weight of the current link node, is the initial risk value of the current link node, is the risk transmission coefficient, For risk increment; A risk transmission path containing different risk levels is generated according to the final risk weight and the composite topological structure.

Images