Knowledge graph-based supply chain intelligent decision optimization method and system

Abstract

The invention relates to the technical field of supply chain decision, and discloses a supply chain intelligent decision optimization method and system based on a knowledge graph, and the method comprises the steps: collecting the multi-source heterogeneous data of a supply chain, and constructing the knowledge graph of the supply chain through the multi-source heterogeneous data; identifying order fluctuation characteristics in the logistics dynamic data, performing emotion quantification processing on the market feedback information to obtain a supply chain emotion index, and performing knowledge enhancement on the knowledge graph to obtain an enhanced knowledge graph; carrying out supply toughness marking on the enhanced knowledge graph to obtain a toughness marking graph, analyzing a chain scission risk conduction path of the toughness marking graph, and carrying out link risk grade marking on the toughness marking graph to obtain a link marking graph; and analyzing the elastic coefficient of the supply chain, carrying out compression resistance level quantification on the toughness marker map to obtain a target knowledge map, and carrying out intelligent decision making on the supply chain to obtain an intelligent decision report. According to the invention, the risk pre-judgment capability of the supply chain can be improved.

Description

TECHNICAL FIELD

[0001] The application relates to a knowledge graph-based supply chain intelligent decision optimization method and system, and belongs to the technical field of supply chain decision. BACKGROUND

[0002] A supply chain refers to a functional network chain structure mode in which a supplier, a manufacturer, a distributor, a retailer and an end user are connected into an entirety, starting from the procurement of raw materials, through the control of information flow, logistics and capital flow, to the manufacture of intermediate products and final products, and finally to the delivery of products to consumers by a sales network. Through intelligent decision optimization of the supply chain, reliable data support can be provided for enterprise planning.

[0003] At present, intelligent decision optimization of the supply chain mainly relies on big data model analysis. Through the collection of massive data generated in each link of the supply chain, the data are analyzed by using an algorithm, so that the optimization of decisions such as demand prediction, inventory management or supplier selection is realized. However, the big data model has insufficient adaptability to dynamically changing supply chain scenarios and cannot identify weak links of the supply chain, so that the risk prediction ability of the supply chain is insufficient. SUMMARY

[0004] The application provides a knowledge graph-based supply chain intelligent decision optimization method and system, and the main purpose is to improve the risk prediction ability of the supply chain.

[0005] To achieve the above purpose, the knowledge graph-based supply chain intelligent decision optimization method provided by the application comprises the following steps: Collecting multi-source heterogeneous data of a supply chain, wherein the multi-source heterogeneous data comprises supplier operation data, logistics dynamic data and market feedback information, and using the multi-source heterogeneous data to construct a knowledge graph of the supply chain; Identifying order fluctuation characteristics in the logistics dynamic data, and performing emotion quantification processing on the market feedback information to obtain a supply chain emotion index, using the order fluctuation characteristics and the supply chain emotion index to perform knowledge enhancement on the knowledge graph to obtain an enhanced knowledge graph; Collecting historical operation data and supply chain event information of the supply chain, performing supply resilience marking on the enhanced knowledge graph based on the historical operation data to obtain a resilience marked graph, analyzing a broken chain risk transmission path of the resilience marked graph according to the supply chain event information, performing link risk level annotation on the resilience marked graph based on the broken chain risk transmission path to obtain a link annotated graph; Using the link labeling graph, the resilience coefficient of the supply chain is analyzed. Based on the resilience coefficient, the stress resistance level of the supply network in the resilience labeling graph is quantified to obtain a target knowledge graph. Based on the target knowledge graph, intelligent decision-making is performed on the supply chain to obtain an intelligent decision report.

[0006] Optionally, based on the supply chain event information, the transmission path of supply chain disruption risk in the resilience marker map is analyzed, including: Identify the event characteristics of the supply chain event information, and quantify the event characteristics to obtain event parameters; Based on the event parameters, calculate the supply chain event information and the impact intensity of the corresponding event; Query the node association relationships of the resilience marker map, and construct the link transmission network of the resilience marker map based on the node association relationships; Based on the intensity of the impact and the link transmission network, the transmission path of the chain breakage risk in the resilience marker map is analyzed.

[0007] Optionally, the step of analyzing the chain break risk transmission path of the resilience marker map based on the influence intensity and the link transmission network includes: Using the influence strength, the network nodes in the link transmission network are labeled with influence strength to obtain an influence-labeled network; Calculate the propagation probability of adjacent network nodes in the intensity-labeled network; Based on the propagation probability, the node links of the intensity labeling network are constructed; Analyze the total risk value of the impact of the node links, and based on the total risk value, determine the transmission path of the chain break risk in the resilience marker map.

[0008] Optionally, the market feedback information is subjected to sentiment quantification processing to obtain a supply chain sentiment index, including: The market feedback information is processed into text to obtain the feedback text. The feedback text is subjected to semantic framing processing to obtain semantically framed text; Based on the semantically segmented text, the sentiment score of the supply chain is calculated; The emotion score is fused with the supply entity corresponding to the supply chain to obtain a bound emotion vector; Based on the bound sentiment vector, the sentiment index of the supply chain is calculated.

[0009] Optionally, the knowledge graph can be augmented using the order fluctuation characteristics and the supply chain sentiment index to obtain an augmented knowledge graph, including: The order fluctuation characteristics are decomposed into fluctuation pattern identifiers and fluctuation intensity values; The supply chain sentiment index is decomposed into a triplet to obtain the decomposed sentiment index. Based on the fluctuation pattern identifier, the logistics nodes in the knowledge graph are updated, and the transportation relationships between logistics entities in the knowledge graph are updated using the fluctuation intensity value, resulting in a preliminary enhanced graph. Using the decomposed sentiment index, the credibility attribute of the preliminary enhanced graph is constructed to obtain the enhanced knowledge graph.

[0010] Optionally, the resilience coefficient of the supply chain can be analyzed using the link annotation map, including: Using the aforementioned link annotation map, the response resilience, recovery resilience, and redundancy resilience of the supply chain are analyzed to obtain three-dimensional resilience information; Based on the three-dimensional elasticity information, the elasticity index of the supply chain is quantified to obtain the elasticity quantification value; The elastic quantization value is dynamically weighted to obtain the elastic weight value; The elasticity coefficient of the supply chain is calculated based on the elasticity quantification value and the elasticity weight value.

[0011] Optionally, based on the historical operational data, the enhanced knowledge graph is subjected to supply resilience labeling to obtain a resilience labeling graph, including: Identify the node fulfillment rate, supply delay standard deviation, quality fluctuation coefficient, and interruption recovery time of the historical supply chain corresponding to the historical operational data; The node resilience value of the historical supply chain is calculated based on the node fulfillment rate, the standard deviation of supply delay, the quality fluctuation coefficient, and the interruption recovery time. Based on the node resilience values, the enhanced knowledge graph is labeled with resilience to obtain a resilience-labeled graph.

[0012] Optionally, based on the elasticity coefficient, the supply network in the toughness marker map is quantified for compressive strength to obtain a target knowledge graph, including: The node toughness values ​​of the toughness marker spectrum are queried, and a weighted logarithmic operation is performed on the elastic coefficient and the node toughness values ​​to obtain the node dynamic bearing value; The dynamic load-bearing value of the node is segmented and thresholded to obtain discrete compressive strength level labels; The discrete compressive strength grade labels are numerically converted to obtain quantitative values ​​of the compressive strength grade. Using the quantified value of the compressive strength grade, attribute values ​​are assigned to the toughness marker spectrum to obtain the target knowledge graph.

[0013] Optionally, using the multi-source heterogeneous data, a knowledge graph of the supply chain is constructed, including: Identify supply chain entities in the multi-source heterogeneous data and define the entity relationships of the supply chain entities; Analyze the three-dimensional attribute information in the multi-source heterogeneous data, and use the three-dimensional attribute information to perform attribute association on the entity relationship to obtain attribute-associated entity relationship; A knowledge graph of the supply chain is constructed using the supply chain entities as nodes and the attribute-related entity relationships as edges.

[0014] To address the aforementioned problems, this invention also provides a knowledge graph-based intelligent supply chain decision optimization system, the system comprising: The knowledge graph construction module is used to collect multi-source heterogeneous data from the supply chain. The multi-source heterogeneous data includes supplier operation data, logistics dynamic data, and market feedback information. The knowledge graph of the supply chain is constructed using the multi-source heterogeneous data. The knowledge graph enhancement module is used to identify order fluctuation characteristics in the logistics dynamic data and perform sentiment quantification processing on the market feedback information to obtain a supply chain sentiment index. The knowledge graph is then enhanced using the order fluctuation characteristics and the supply chain sentiment index to obtain an enhanced knowledge graph. The graph annotation module is used to collect historical operation data and supply chain event information of the supply chain. Based on the historical operation data, the augmented knowledge graph is marked with supply resilience to obtain a resilience-marked graph. Based on the supply chain event information, the transmission path of the risk of chain breakage in the resilience-marked graph is analyzed. Based on the transmission path of the risk of chain breakage, the link risk level of the resilience-marked graph is marked to obtain a link-marked graph. The intelligent analysis module is used to analyze the resilience coefficient of the supply chain using the link labeling graph, quantify the stress resistance level of the supply network in the resilience labeling graph based on the resilience coefficient, obtain a target knowledge graph, and make intelligent decisions on the supply chain based on the target knowledge graph to obtain an intelligent decision report.

[0015] This invention first constructs a supply chain knowledge graph by collecting heterogeneous data from multiple sources, including supplier operational data, logistics dynamic data, and market feedback information. This unified modeling of dispersed entities (such as suppliers and logistics nodes), relationships (such as supply relationships and transportation relationships), and attributes (such as capacity and timeliness) achieves structured integration of supply chain data. Then, by identifying order fluctuation characteristics in logistics dynamic data and quantifying market feedback information to obtain a supply chain sentiment index, this invention enhances the knowledge graph, forming an augmented knowledge graph. This augmented knowledge graph not only includes basic entity relationships but also reflects real-time dynamic fluctuations and external sentiment guidance within the supply chain, thus overcoming the static limitations of traditional knowledge graphs and improving its adaptability to complex supply chain scenarios. Furthermore, this invention uses historical operational data to label the augmented knowledge graph with supply chain resilience data. By combining supply chain event information to analyze the transmission path of supply chain disruption risks and labeling their levels, a link labeling graph is obtained. This combines historical experience with real-time risks, visualizing weak links and risk diffusion patterns in the supply chain, and achieving a precise mapping between supply chain resilience and risk. Furthermore, this invention utilizes link annotation graph analysis to determine resilience coefficients, quantifying the supply network's resilience level and forming a target knowledge graph. This provides a multi-dimensional basis for decision-making, and ultimately, based on this graph, an intelligent decision-making report can output specific solutions for different scenarios. Therefore, this invention can improve the risk prediction capability of the supply chain. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a knowledge graph-based intelligent decision-making optimization method for the supply chain according to an embodiment of the present invention. Figure 2 This is a schematic diagram of a module for implementing a knowledge graph-based intelligent decision-making optimization method for the supply chain, as provided in an embodiment of the present invention.

[0017] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] This application provides a knowledge graph-based intelligent supply chain decision optimization method. The executing entity of the knowledge graph-based intelligent supply chain decision optimization method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the knowledge graph-based intelligent supply chain decision optimization method can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0020] Reference Figure 1 The diagram shown is a flowchart illustrating a knowledge graph-based intelligent supply chain decision optimization method according to an embodiment of the present invention. In this embodiment, the knowledge graph-based intelligent supply chain decision optimization method includes: S1. Collect multi-source heterogeneous data from the supply chain, wherein the multi-source heterogeneous data includes supplier operation data, logistics dynamic data and market feedback information, and construct a knowledge graph of the supply chain using the multi-source heterogeneous data.

[0021] This invention can collect real information from each part of the supply chain by collecting multi-source heterogeneous data, such as supplier production reports, logistics information in transit, and customer feedback forms, so that users can clearly understand the real situation of each link in the supply chain.

[0022] The multi-source heterogeneous data includes supplier operation data, logistics dynamic data, and market feedback information. The supplier operation data refers to various data generated by suppliers in their daily operations, such as daily production line output, raw material inventory balance, on-time delivery rate in the past six months, and equipment maintenance records. The logistics dynamic data refers to information on real-time changes in the logistics process, such as the real-time location of transport vehicles, the immediate inventory quantity in the warehouse, the time for loading and unloading goods, and the congestion status of transportation routes. The market feedback information refers to various feedback from the market, such as reports of slow-moving products submitted by distributors, suggestions from end customers for improving product packaging, and notifications of sudden increases in regional market demand.

[0023] Optionally, the multi-source heterogeneous data can be obtained through, for example, by connecting to the supplier's ERP system to obtain production data, by collecting transportation trajectories through the GPS terminals of logistics vehicles, and by capturing market opinions through forms or comments on customer feedback platforms.

[0024] Furthermore, by utilizing the multi-source heterogeneous data, the embodiments of the present invention can construct a knowledge graph of the supply chain, which can connect scattered information and allow users to see the connections between various parts of the supply chain. For example, by connecting data such as supplier capacity, logistics transportation routes, and customer needs, the graph can intuitively show who supplies whom, how the goods are transported, and what needs need to be met, thereby clarifying the structure of the entire supply chain.

[0025] The knowledge graph refers to a network that connects entities in the supply chain, such as suppliers, manufacturers, logistics points, and customers, as well as their relationships in terms of supply, transportation, and demand, and attributes such as the capacity, transportation timeliness, and demand quantity of each entity, in a structured way.

[0026] As an embodiment of the present invention, a knowledge graph of the supply chain is constructed using the multi-source heterogeneous data, including: Identify supply chain entities in the multi-source heterogeneous data and define the entity relationships of the supply chain entities; Analyze the three-dimensional attribute information in the multi-source heterogeneous data, and use the three-dimensional attribute information to perform attribute association on the entity relationship to obtain attribute-associated entity relationship; A knowledge graph of the supply chain is constructed using the supply chain entities as nodes and the attribute-related entity relationships as edges.

[0027] The three-dimensional attribute information refers to the information attributes in multi-source heterogeneous data regarding supplier operation data, logistics dynamic data, and market feedback information. For example, the attribute-associated entity relationship refers to the attribute-based entity relationship formed by associating the three-dimensional attribute information with the basic relationship between entities (such as "supply" and "delivery"). For example, "Supplier A → 2024-06-10 → 300 boxes → Quality inspection qualified → Supply → Retailer B" is an example where the three-dimensional attribute information is bound to the entity relationship of "supply", making the relationship description more specific.

[0028] In practical implementation, for multi-source heterogeneous data (such as Excel spreadsheets from order systems, JSON logs from logistics systems, and PDF texts of supplier contracts), Named Entity Recognition (NER) technology is used for entity recognition. Pre-trained models (such as BERT) are used to extract supply chain entities such as suppliers, warehouses, and products. Simultaneously, dependency parsing technology is used to identify relationships between entities such as "supply," "delivery," and "storage" from the data. For example, a "storage" relationship is identified between "Shanghai warehouse" and "Huawei Mate 60 mobile phone" from logistics records. In extracting the three-dimensional attribute information of entity relationships from multi-source heterogeneous data (such as transaction date in the time dimension, number of transported boxes in the quantity dimension, and quality inspection pass rate in the status dimension), information extraction techniques (such as regular expression matching) can be used to extract specific values. For example, the three-dimensional attributes of "Beijing supplier" supplying goods to "Guangzhou warehouse" can be extracted from order data as "2023-10-01," "200 boxes," and "98%." Then, attribute mapping algorithms are used to associate these three-dimensional attributes with the corresponding entity relationships, forming attributed associations. A knowledge graph is constructed using a graph database (such as Neo4j). Identified supply chain entities (such as "Shenzhen factory" and "iPhone 15") are used as graph nodes, and entity relationships with three-dimensional attributes (such as "Shenzhen factory → 2023-09-15 → 3000 units → 99% pass rate → supplier → Shanghai distributor") are used as edges connecting the nodes. The node ID, edge type, and attribute key-value pairs are defined through the graph data model, and finally a visualized supply chain knowledge graph is generated.

[0029] S2. Identify the order fluctuation characteristics in the logistics dynamic data, and perform sentiment quantification processing on the market feedback information to obtain the supply chain sentiment index. Use the order fluctuation characteristics and the supply chain sentiment index to enhance the knowledge graph to obtain an enhanced knowledge graph.

[0030] This invention can identify abnormal increases or decreases in order volume and frequency changes in logistics records by recognizing order fluctuation characteristics in the logistics dynamic data. For example, in the fresh food supply chain, logistics system data can be used to discover that orders in a certain area have decreased by 20% for three consecutive days compared to previous periods, thereby understanding the short-term demand changes in that area.

[0031] The order fluctuation characteristics refer to the changes in order quantity, frequency, and time intervals in logistics dynamic data that differ from the historical normal state.

[0032] Optionally, the order fluctuation characteristics can be calculated using data statistics tools to determine the historical mean and standard deviation of order volume, order frequency, and order interval duration, thereby defining the normal range. Then, the real-time collected logistics dynamic data can be compared with this range to identify fluctuation characteristics such as sudden increases / decreases in order volume and abnormal frequency.

[0033] Furthermore, this embodiment of the invention quantifies the market feedback information to obtain a supply chain sentiment index, which can transform scattered customer evaluations and distributor feedback into measurable indicators. For example, in the apparel supply chain, the proportion of negative feedback such as "poor quality" and "slow logistics" increased from 10% to 30%, quantifying a decrease in the sentiment index and intuitively reflecting the market's dissatisfaction with the supply chain.

[0034] The supply chain sentiment index refers to a quantitative indicator that can intuitively reflect the positive or negative sentiment and degree of the market towards each link of the supply chain (such as product quality, logistics timeliness, etc.).

[0035] As an embodiment of the present invention, the market feedback information is subjected to sentiment quantification processing to obtain a supply chain sentiment index, including: The market feedback information is processed into text to obtain the feedback text. The feedback text is subjected to semantic framing processing to obtain semantically framed text; Based on the semantically segmented text, the sentiment score of the supply chain is calculated; The emotion score is fused with the supply entity corresponding to the supply chain to obtain a bound emotion vector; Based on the bound sentiment vector, the sentiment index of the supply chain is calculated.

[0036] The semantic framed text refers to the set of semantic units divided according to supply chain links (such as logistics, quality, and price) after semantic segmentation of the feedback text. The sentiment score refers to the numerical value obtained by quantifying the sentiment polarity (positive, negative, and neutral) of each semantic frame based on the semantic framed text through sentiment analysis technology. The bound sentiment vector refers to the feature vector formed by combining the sentiment score with the corresponding entity in the supply chain (such as supplier and logistics provider) through vector fusion technology.

[0037] In practical implementation, for market feedback information (such as customer voice reviews, video comments, or paper questionnaire feedback), speech-to-text (ASR) technology can be used to convert speech into text, and OCR technology can be used to recognize the text in images or paper documents, processing them uniformly into structured feedback text. For example, a customer's voice message "The delivery of this batch of parts is severely delayed" can be converted into text, and the handwritten "Packaging intact, thumbs up" in a product review image can be recognized as text. Word segmentation tools (such as Jieba) are used to segment the feedback text, and semantic role labeling technology is combined to identify the supply chain links (such as logistics, quality, and price) involved in the text, splitting it into semantic frames according to semantic units. For example, "Slow logistics delivery, poor product quality" can be split into two semantic frames: "[Logistics link: slow delivery]" and "[Quality link: poor product]". A pre-trained sentiment analysis model (such as TextCNN) is used to determine the sentiment polarity of each semantic frame. Positive sentiment scores are set to [0.1, 1], negative to [-1, -0.1], and neutral to 0. The sentiment score for each frame is calculated by combining the frame's importance weights (e.g., 0.4 for quality-related frames, 0.3 for logistics). These scores are then accumulated to obtain the initial sentiment score for the supply chain. For example, "good quality" (positive, 0.8 × 0.4 = 0.32) and "slow delivery" (negative, -0.6 × 0.3 = -0.18) have an initial score of 0.14. Entity linking technology is used to match entities in the semantic frame text (e.g., "Yuantong Logistics" and "Supplier A") with entities in the supply chain knowledge graph. A vector fusion algorithm is used to combine the sentiment score with the feature vector of the corresponding entity to generate a bound sentiment vector. For example, the sentiment score of -0.7 for "Yuantong Logistics slow delivery" is fused with the entity vector of "Yuantong Logistics" to obtain the bound vector (Yuantong Logistics, -0.7). Based on the weight of each supply entity in the supply chain (e.g., core supplier 0.5, logistics provider 0.3, distributor 0.2), a weighted sum of all bound sentiment vectors is obtained to obtain the supply chain sentiment index, with the index range set to [-1, 1]. For example, if the core supplier sentiment vector is 0.6 (weight 0.5), the logistics provider is -0.4 (0.3), and the distributor is 0.2 (0.2), the index = 0.6×0.5+(-0.4)×0.3+0.2×0.2=0.3-0.12+0.04=0.22, representing an overall positive sentiment.

[0038] Furthermore, this embodiment of the invention utilizes the order fluctuation characteristics and the supply chain sentiment index to enhance the knowledge graph, resulting in an enhanced knowledge graph that incorporates dynamic changes and market sentiment dimensions into the original graph that only reflects entity relationships. For example, in a retail supply chain, the graph can be linked to the fluctuation characteristic of a 50% weekly increase in order volume for a certain product, as well as the sentiment index of a 20% decrease in customer satisfaction, making the real-time status of entities such as suppliers and logistics providers more intuitive.

[0039] The enhanced knowledge graph refers to a knowledge graph that incorporates information such as order fluctuation characteristics (e.g., sudden increases or decreases in order volume) and supply chain sentiment index (e.g., positive or negative market sentiment) on the basis of the original knowledge graph, thereby more comprehensively reflecting the dynamic relationship between supply chain entities and market feedback.

[0040] As an embodiment of the present invention, the knowledge graph is augmented using the order fluctuation characteristics and the supply chain sentiment index to obtain an augmented knowledge graph, including: The order fluctuation characteristics are decomposed into fluctuation pattern identifiers and fluctuation intensity values; The supply chain sentiment index is decomposed into a triplet to obtain the decomposed sentiment index. Based on the fluctuation pattern identifier, the logistics nodes in the knowledge graph are updated, and the transportation relationships between logistics entities in the knowledge graph are updated using the fluctuation intensity value, resulting in a preliminary enhanced graph. Using the decomposed sentiment index, the credibility attribute of the preliminary enhanced graph is constructed to obtain the enhanced knowledge graph.

[0041] The fluctuation pattern identifier refers to a symbol or code that identifies the type of order fluctuation characteristics and is used to distinguish different fluctuation patterns (such as periodicity, suddenness, trend, etc.). The fluctuation intensity value refers to a quantitative indicator of the order fluctuation amplitude, which usually ranges from 0 to 1. The larger the value, the more intense the fluctuation. The decomposed sentiment index refers to the sentiment quantification value that is bound to specific entity pairs after the overall sentiment index of the supply chain is decomposed according to the relationship between entities in the knowledge graph (such as supply, transportation, collaboration, etc.).

[0042] In practical implementation, time series decomposition algorithms (such as STL decomposition) can be used to break down order fluctuation characteristics, extract fluctuation patterns (such as periodic fluctuations and sudden fluctuations), and assign identifiers (such as "P - periodic" and "S - sudden"). Simultaneously, the fluctuation intensity value (range 0-1) is determined by calculating the ratio of the fluctuation amplitude to the mean. For example, the quarterly order volume for a certain electronic product, "20% monthly growth and a 50% surge before the Spring Festival," can be decomposed into pattern identifiers "P - quarterly + S - holiday," intensity values ​​of 0.4 (regular fluctuation) and 0.8 (holiday fluctuation). The supply chain sentiment index (such as 0.6) can be decomposed into triplets in the form of (entity A, relationship R, entity B, sentiment value) according to entity influence weights. Entity linking technology is used to match corresponding entities. For example, the overall sentiment index of 0.6 can be decomposed into decomposed sentiment indices such as (supplier X, supply, factory Y, 0.7) and (factory Y, production, product Z, 0.5), with weights allocated according to entity correlation (such as a core supplier weight of 0.3). Based on the fluctuation pattern identifier, the graph update rules are invoked to expand the attributes of logistics nodes. For example, in the "P-Quarter" mode, the attribute "Quarterly reserve threshold 8000 pieces" is added to the Guangzhou warehouse node. A weighted correction algorithm is used to integrate the fluctuation intensity value (e.g., 0.4) into the transportation relationship attribute. For example, the transportation relationship "Guangzhou warehouse → Wuhan distributor" is updated to "Transportation volume 500±200 pieces / week (fluctuation intensity 0.4)", forming a preliminary enhanced graph. The decomposed sentiment index is mapped to a credibility attribute value (range 0-1, sentiment value 1 corresponds to credibility 1, -1 corresponds to 0.1). Through graph attribute fusion technology, it is associated with the corresponding entities and relationships in the preliminary enhanced graph. For example, the attribute "credibility 0.85" is added to the relationship (supplier X, supplier, factory Y), and the attribute "cooperation credibility 0.9" is added to the supplier X node. Finally, an enhanced knowledge graph with credibility annotations is generated.

[0043] S3. Collect historical operation data and supply chain event information of the supply chain. Based on the historical operation data, perform supply resilience labeling on the enhanced knowledge graph to obtain a resilience labeling graph. According to the supply chain event information, analyze the risk transmission path of the supply chain disruption in the resilience labeling graph. Based on the risk transmission path of the supply chain disruption, label the link risk level of the resilience labeling graph to obtain a link labeling graph.

[0044] This invention, through the collection of historical operational data and event information from the supply chain, can accumulate records of the supply chain's past performance and key events. For example, in the electronic component supply chain, it can collect data on production qualification rates and inventory turnover rates (historical operational data) from the past five years, as well as past chip supply disruptions and logistics strikes (supply chain event information), providing a basis for understanding the past state of the supply chain.

[0045] The supply chain event information refers to information related to various key events that occur in the operation of the supply chain and may affect its normal flow, such as the specific details of events such as supplier supply disruptions, logistics interruptions, and policy adjustments. It is an important basis for reflecting past disturbances in the supply chain.

[0046] In practice, operational data such as historical order data and logistics records can be automatically collected through sensors and management system interfaces deployed at various stages of the supply chain.

[0047] This invention, through embodiments thereof, uses historical operational data to perform supply chain resilience labeling on the enhanced knowledge graph, resulting in a resilience-labeled graph. This graph quantifies and labels the resilience characteristics of each entity in the supply chain, such as its ability to withstand risks and its recovery speed during past disturbances, allowing the graph to intuitively reflect the resilience level of each part. For example, in the food supply chain, logistics providers that resumed delivery within 3 days during the pandemic can be labeled "high resilience," making the resilience-labeled graph clearly show their ability to cope with emergencies.

[0048] The resilience marker map refers to a map formed by marking the supply resilience characteristics (such as risk resistance and recovery speed) of each entity (such as suppliers, logistics nodes, etc.) on the basis of an augmented knowledge graph and combined with historical supply chain operation data.

[0049] As an embodiment of the present invention, based on the historical operational data, the enhanced knowledge graph is subjected to supply resilience labeling to obtain a resilience labeling graph, including: Identify the node fulfillment rate, supply delay standard deviation, quality fluctuation coefficient, and interruption recovery time of the historical supply chain corresponding to the historical operational data; The node resilience value of the historical supply chain is calculated based on the node fulfillment rate, the standard deviation of supply delay, the quality fluctuation coefficient, and the interruption recovery time. Based on the node resilience values, the enhanced knowledge graph is labeled with resilience to obtain a resilience-labeled graph.

[0050] The node fulfillment rate refers to the ratio of the number of orders actually fulfilled by a node in the supply chain (such as a supplier, warehouse, or logistics provider) to the total number of orders to be fulfilled within a certain period. The standard deviation of supply delay refers to a statistic calculated based on multiple supply delay times (such as delay days) of a node, used to measure the dispersion of delay time; the smaller the value, the more stable the delay situation. The quality fluctuation coefficient refers to the ratio of the standard deviation of the quality non-conformity rate of the products or services provided by a node to the average non-conformity rate, used to measure quality stability; the lower the value, the more stable the quality. The interruption recovery time refers to the time from the occurrence of the interruption to the restoration of normal supply after a supply interruption at a node in the supply chain due to a sudden event (such as a natural disaster or equipment failure), usually measured in hours or days. The node resilience value is a quantitative value obtained by weighting the node fulfillment rate, standard deviation of supply delay, quality fluctuation coefficient, and interruption recovery time, used to measure the node's ability to resist risks and maintain stable supply.

[0051] In practice, historical order records, logistics tracking data, and quality inspection reports are retrieved from historical operational data. Then, data cleaning and statistical analysis techniques are used to extract indicators: node fulfillment rate (actual fulfilled orders / total orders × 100%), such as a supplier's fulfillment rate of 92% within six months; standard deviation of supply delay (calculated based on 30 delay days), such as a logistics node delay day standard deviation of 1.5 days; quality fluctuation coefficient (standard deviation of quality non-conformance rate / average non-conformance rate), such as a component quality fluctuation coefficient of 0.2; interruption recovery time (hours from supply interruption to normal operation), such as a warehouse recovering 48 hours after being interrupted by rainstorm. Set weights for each indicator (compliance rate 0.3, delay standard deviation 0.2, quality fluctuation coefficient 0.2, recovery time 0.3), and normalize them (e.g., compliance rate 92% → 0.92, recovery time 48 hours is mapped to 0.5 according to "≤24h → 1, 24-72h → 0.5, >72h → 0.2"). Substitute these values ​​into the formula: Node resilience value = (compliance rate × 0.3) + (1 - delay standard deviation / 5 × 0.2) + (1 - quality fluctuation coefficient × 0.2) + (recovery time mapped value × 0.3). For example, the calculated resilience value of a certain node is 0.78. Set resilience labeling thresholds (≥0.8 for "high resilience", 0.5-0.8 for "medium resilience", and <0.5 for "low resilience"). Use graph database attribute update technology to add the label corresponding to the node resilience value to the corresponding node attribute of the augmented knowledge graph. For example, label the supplier node with a resilience value of 0.78 as "medium resilience", forming a graph with resilience labels.

[0052] This invention, through analyzing the supply chain event information and the resilience marker map, can clarify the specific chain of risk diffusion from a certain link (such as supplier issues) to other related links. For example, in the automotive supply chain, by combining past battery supplier delay events, the transmission path of "battery supply delay → vehicle assembly stagnation → dealer delivery default" can be analyzed in the map, making the diffusion logic of supply chain disruption risk clear at a glance.

[0053] The chain break risk transmission path refers to the map formed by marking the risk level (such as high, medium, and low) of each path based on the resilience marker map and according to the chain break risk transmission path, which can intuitively present the risk level of different transmission links.

[0054] As an embodiment of the present invention, analyzing the chain break risk transmission path of the resilience marker map based on the supply chain event information includes: Identify the event characteristics of the supply chain event information, and quantify the event characteristics to obtain event parameters; Based on the event parameters, calculate the supply chain event information and the impact intensity of the corresponding event; Query the node association relationships of the resilience marker map, and construct the link transmission network of the resilience marker map based on the node association relationships; Based on the intensity of the impact and the link transmission network, the transmission path of the chain breakage risk in the resilience marker map is analyzed.

[0055] The event parameters refer to the specific values ​​or code set obtained after quantifying the characteristics of supply chain events. The impact intensity refers to the quantitative indicator that reflects the degree of impact of the event on the supply chain, calculated based on the event parameters through a weighted formula. The link transmission network refers to the directed network model constructed based on the node association relationship in the resilience marker graph.

[0056] In practical implementation, natural language processing technologies (such as the BERT model) can be used to identify event characteristics from supply chain event information (such as news reports and system alarms), including event type (such as raw material supply disruptions or logistics obstructions), affected areas, and impacted entities. These characteristics are then converted into parameters using pre-defined quantification rules. For example, "snowstorm in North China caused a 3-day disruption to road transport" can be quantified as the event type code "WL-01," a regional impact score of 0.7 (out of 1), and a duration of 3 days, forming an event parameter set. When calculating the impact intensity based on these event parameters, the formula I = 0.4 × regional impact score + 0.3 × (duration / 7) + 0.3 × event type weight (e.g., 0.8 for natural disasters and 0.5 for human error) can be used. For example, for the aforementioned snowstorm event, I = 0.4 × 0.7 + 0.3 × (3 / 7) + 0.3 × 0.8 ≈ 0.28 + 0.13 + 0.24 = 0.65 (range 0-1, higher values ​​indicate stronger impact). The specific weights need to be set according to the actual application. The relationships between nodes in the resilience marker graph are extracted using a graph database query language (such as SPARQL) (e.g., “Supplier A - Supply - Factory B”, “Factory B - Transportation - Warehouse C”). A network topology building tool is used to generate a link transmission network by treating nodes as vertices and relationships as directed edges. The attributes of the edges include transmission frequency (e.g., 3 times per day) and dependency coefficient (e.g., Factory B's dependency coefficient on Supplier A is 0.9).

[0057] Furthermore, as an optional embodiment of the present invention, the step of analyzing the chain break risk transmission path of the resilience marker map based on the influence intensity and the link transmission network includes: Using the influence strength, the network nodes in the link transmission network are labeled with influence strength to obtain an influence-labeled network; Calculate the propagation probability of adjacent network nodes in the intensity-labeled network; Based on the propagation probability, the node links of the intensity labeling network are constructed; Analyze the total risk value of the impact of the node links, and based on the total risk value, determine the transmission path of the chain break risk in the resilience marker map.

[0058] In this context, the node link refers to a directed path formed by connecting adjacent nodes with a transmission probability ≥ a set threshold (e.g., 0.1) in the strength labeling network. It includes multiple sequentially associated nodes and the transmission relationships between nodes. The total impact risk value is a quantitative value obtained by weighting and summing the transmission probability of each link segment in the node link with the corresponding link importance weight. It is used to measure the degree of risk accumulation of the entire link.

[0059] In practical implementation, based on the impact intensity I (e.g., I=0.65 for a blizzard event), graph node attribute labeling technology can be used to mark the initial intensity value (equal to I) of the initial nodes directly related to the event in the link transmission network (e.g., logistics hub nodes in North China), and the indirect related nodes can be labeled according to distance attenuation (e.g., the intensity of a node at 1 hop distance = I×0.7, and the intensity of a node at 2 hop distance = I×0.49), forming an intensity labeling network. For example, "Logistics Hub A" is labeled with 0.65, and "Warehouse B" (1 hop distance) is labeled with 0.455. A transmission probability threshold of 0.1 is set, and a link filtering algorithm is used to retain adjacent node connections with P≥0.1, constructing node links. For example, links such as "Logistics Hub A→Warehouse B (P=0.125)" and "Warehouse B→Distributor C (P=0.11)" are retained, forming the initial node link "Logistics Hub A - Warehouse B - Distributor C". The total risk value of the node link impact is calculated using the formula: Total Risk Value = Σ (Transmission Probability of Each Link × Importance Weight of Link), where the core link has a weight of 0.6, and secondary links have a weight of 0.4. For example, in the above links, "A→B" (P=0.125, weight 0.6) and "B→C" (P=0.11, weight 0.4) have a total risk value of 0.125×0.6+0.11×0.4≈0.075+0.044=0.119; set a total risk threshold of 0.1, and identify links with a risk of chain break as transmission paths.

[0060] Furthermore, as another embodiment of the present invention, the formula for calculating the transmission probability is as follows: ; in, Indicates the propagation probability. This represents the influence coefficient corresponding to the influence strength, where i represents a network node in the strength labeling network, and j represents the network node adjacent to i. This represents the resilience value of node i. This represents the resilience value of node j. This represents the link distance between node i and node j.

[0061] It should be further explained that the above formula for calculating the transmission probability first determines the event intensity coefficient based on the event's impact range and duration, then combines this with the resilience level of the two nodes (the lower the resilience, the higher the base value of the transmission probability), and the physical distance between the nodes (the farther the distance, the more obvious the transmission attenuation). This comprehensively measures the likelihood of risk propagating from one node to another, reflecting the combined effect of event intensity, node resilience, and spatial distance on risk propagation. This formula can accurately quantify the risk transmission probability in different scenarios, making risk analysis more targeted. For example, for events with a wide impact range and long duration, it amplifies their initial transmission potential; while resilient nodes can effectively reduce the probability of risk propagation outwards, and distant nodes will also have a reduced chance of being affected due to the attenuation effect. Ultimately, the numerical results clearly distinguish the differences in risk transmission across different links, providing a reliable basis for identifying high-risk transmission paths.

[0062] Furthermore, this embodiment of the invention, by labeling the resilience marker map with link risk levels based on the chain break risk transmission path, obtains a link label map that clearly identifies the severity of risks in different transmission paths. For example, in the electronics supply chain, for the path "chip supplier delay → motherboard production disruption → complete machine assembly delay," the chip to motherboard segment is labeled as "high risk," and the motherboard to complete machine segment as "medium risk," allowing the link label map to intuitively present the severity of risks at each stage.

[0063] The link labeling map refers to a map formed by labeling the risk level (such as high, medium, and low) of each transmission link based on the resilience labeling map and the analysis of the chain break risk transmission path. It can intuitively reflect the severity of the risk of different transmission links.

[0064] In practice, based on the transmission path of chain break risks and the degree of risk exposure of each link in the transmission, the corresponding links in the resilience marking map can be labeled with high, medium and low risk levels to form a link labeling map.

[0065] S4. Using the link labeling graph, analyze the resilience coefficient of the supply chain. Based on the resilience coefficient, quantify the stress resistance level of the supply network in the resilience labeling graph to obtain a target knowledge graph. Based on the target knowledge graph, make intelligent decisions for the supply chain to obtain an intelligent decision report.

[0066] This invention utilizes the aforementioned link labeling map to analyze the resilience coefficient of the supply chain. This analysis, combined with the risk level of each link, quantifies the supply chain's resilience and adjustment capacity in the face of different risks. For example, in the fast-moving consumer goods (FMCG) supply chain, the risk level labeling of "raw material supply → warehousing → terminal" in the link labeling map yields a resilience coefficient of 0.6 (higher values ​​indicate stronger resilience), clearly demonstrating its resilience to fluctuations.

[0067] The resilience coefficient refers to an indicator derived from the analysis of the link labeling map, used to quantify the supply chain's resilience and adjustment space when facing different risks. The value reflects the strength of the resilience.

[0068] As an embodiment of the present invention, the resilience coefficient of the supply chain is analyzed using the link annotation map, including: Using the aforementioned link annotation map, the response resilience, recovery resilience, and redundancy resilience of the supply chain are analyzed to obtain three-dimensional resilience information; Based on the three-dimensional elasticity information, the elasticity index of the supply chain is quantified to obtain the elasticity quantification value; The elastic quantization value is dynamically weighted to obtain the elastic weight value; The elasticity coefficient of the supply chain is calculated based on the elasticity quantification value and the elasticity weight value.

[0069] The response resilience refers to the supply chain's ability to react quickly from the occurrence of a sudden disturbance (such as a surge in orders or a shortage of raw materials) to the initiation of initial adjustment measures (such as adjusting inventory or switching suppliers). The recovery resilience refers to the speed and extent to which the supply chain recovers to normal operating conditions (such as production capacity and delivery efficiency) after an interruption, and is usually measured by recovery ratio and recovery time. The redundancy resilience refers to the level of reserve resources (such as backup suppliers, redundant inventory, and alternative logistics channels) in the supply chain to buffer against risk shocks.

[0070] In practical implementation, data such as risk response time, recovery cycle, and reserve resource quantity for each link can be extracted from the link annotation map. Graph data analysis technology can then be used to calculate: response elasticity (the time from the occurrence of a disturbance to the first adjustment, e.g., ≤4 hours is optimal), recovery elasticity (the percentage of capacity restored to normal, e.g., 90% recovery within 24 hours is optimal), and redundancy elasticity (the ratio of the number of reserve nodes to the total number of nodes, e.g., ≥30% is optimal). For example, in the FMCG supply chain, a certain link has a response time of 2 hours, recovers 85% of its capacity within 24 hours, and has a reserve warehouse ratio of 25%, resulting in three-dimensional elasticity information (excellent response, good recovery, and moderate redundancy). A normalization algorithm is used to quantify the three-dimensional elasticity information into elasticity values ​​of 0-1. For response elasticity, the values ​​are set as follows: "≤2 hours → 1.0, 2-6 hours → 0.6, >6 hours → 0.2". For recovery elasticity, the values ​​are set as follows: "Recovery rate ≥90% → 1.0, 70%-90% → 0.7, <70% → 0.3". For redundancy elasticity, the values ​​are set as follows: "Percentage ≥30% → 1.0, 20%-30% → 0.5, <20% → 0.1". For example, the quantified values ​​for the above-mentioned FMCG supply chain are 0.6 for response, 0.7 for recovery, and 0.5 for redundancy. Based on the current weaknesses in the supply chain (e.g., increasing the recovery elasticity weight if historical data shows insufficient recovery capacity), a dynamic weighting model is used to allocate weights, setting the base weight for response elasticity to 0.2, recovery elasticity to 0.5, and redundancy elasticity to 0.3 (totaling 1). For example, if an electronics supply chain has recently experienced multiple recovery delays, the recovery elasticity weight is adjusted to 0.6, the response weight to 0.2, and the redundancy weight to 0.2.

[0071] Furthermore, as another embodiment of the present invention, the formula for the elastic coefficient is as follows: ; in, Represents the elastic coefficient. This represents the quantized value of the response elasticity. This indicates the recovery of the elasticity quantization value. This represents the redundancy elasticity quantization value. express The elastic weight value, express Elastic weight value, express The elastic weight value.

[0072] Furthermore, this embodiment of the invention quantifies the resilience level of the supply network in the resilience marker graph based on the elasticity coefficient, resulting in a target knowledge graph that can transform the supply network's ability to withstand risks into specific level indicators, making its resilience characteristics easier to measure. For example, in the home appliance supply chain, the "component supplier → assembly plant → distributor" network can be labeled with "resilience level A" based on the elasticity coefficient, allowing the target knowledge graph to intuitively present the network's risk resistance level.

[0073] The target knowledge graph refers to a graph formed by quantifying the resilience level of the supply network based on the elasticity coefficient on the basis of the resilience marker graph. It can intuitively present the risk resistance level of the supply network, such as the home appliance supply network marked with "resilience level A".

[0074] As an embodiment of the present invention, based on the elastic coefficient, the supply network in the toughness marker map is quantified for compressive strength level to obtain a target knowledge graph, including: The node toughness values ​​of the toughness marker spectrum are queried, and a weighted logarithmic operation is performed on the elastic coefficient and the node toughness values ​​to obtain the node dynamic bearing value; The dynamic load-bearing value of the node is segmented and thresholded to obtain discrete compressive strength level labels; The discrete compressive strength grade labels are numerically converted to obtain quantitative values ​​of the compressive strength grade. Using the quantified value of the compressive strength grade, attribute values ​​are assigned to the toughness marker spectrum to obtain the target knowledge graph.

[0075] The node dynamic load-bearing value refers to a quantitative indicator that reflects the node's ability to withstand pressure in a dynamic environment, obtained by weighted logarithmic calculation of the node's resilience value and supply chain elasticity coefficient. The discrete pressure resistance level label refers to a non-continuous level identifier that maps the node's dynamic load-bearing value to a segmented threshold, used to intuitively distinguish the level of the node's pressure resistance.

[0076] In practice, the resilience values ​​(e.g., 0.82, 0.65) of each node in the resilience marker map can be extracted using graph database query technology (such as Cypher statements). The elasticity coefficient is weighted at 0.6, and the node resilience value at 0.4. The dynamic carrying capacity value of a node is calculated using the formula: "Node dynamic carrying capacity value = 0.6 × ln (1 + elasticity coefficient) + 0.4 × ln (1 + node resilience value)". For example, when the elasticity coefficient is 0.7 and the node resilience value is 0.8, the dynamic carrying capacity value is approximately 0.6 × 0.53 + 0.4 × 0.59 ≈ 0.55. Segmentation thresholds are set (dynamic carrying capacity value ≥ 1.2 → Grade A, 0.8-1.2 → Grade B, 0.4-0.8 → Grade C, < 0.4 → Grade D). A threshold judgment algorithm is used to map the node dynamic carrying capacity value to discrete labels. For example, a node with a calculated value of 0.55 corresponds to the "Grade C" label. By using a grade-value mapping rule (Grade A = 4, Grade B = 3, Grade C = 2, Grade D = 1), discrete compressive strength grade labels are converted into quantified values ​​of 1-4. For example, "Grade C" corresponds to a quantified value of 2, thus achieving a numerical representation of the labels. Using a graph data update tool, the quantified compressive strength grade value is added as a new attribute to the corresponding node in the resilience labeling graph (e.g., "Compressive Strength Grade: 2" is added to the "Supplier A" node). After completing the attribute assignment, a target knowledge graph containing quantified compressive strength grade information is formed, such as the quantified grade labeling of chip supplier nodes in the electronics supply chain.

[0077] Furthermore, this embodiment of the invention enables intelligent decision-making on the supply chain based on the target knowledge graph. The resulting intelligent decision report can identify potential problems in advance and provide solutions by leveraging the connections and characteristics of each link in the graph. For example, in a fresh food supply chain, if the target knowledge graph reveals insufficient stability of the cold chain in a certain area, a decision report can be generated to prioritize the allocation of refrigerated trucks, thus helping the supply chain operate smoothly.

[0078] In practice, based on the quantified stress resistance level of each node in the target knowledge graph, intelligent decision-making algorithms (such as rule-based reasoning models) can be used to identify weak links in the supply chain, generate targeted optimization suggestions (such as adding backup resources for D-level nodes), and summarize them to form an intelligent decision-making report.

[0079] like Figure 2 The diagram shown is a functional block diagram of the supply chain intelligent decision optimization system based on knowledge graphs according to the present invention.

[0080] The knowledge graph-based intelligent supply chain decision optimization system 200 of this invention can be installed in an electronic device. Depending on the functions implemented, the knowledge graph-based intelligent supply chain decision optimization system may include a graph construction module 201, a graph enhancement module 202, a graph annotation module 203, and an intelligent analysis module 204. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0081] In this embodiment of the invention, the functions of each module / unit are as follows: The graph construction module 201 is used to collect multi-source heterogeneous data of the supply chain, wherein the multi-source heterogeneous data includes supplier operation data, logistics dynamic data and market feedback information, and to construct the knowledge graph of the supply chain using the multi-source heterogeneous data; The knowledge graph enhancement module 202 is used to identify order fluctuation characteristics in the logistics dynamic data, perform sentiment quantification processing on the market feedback information to obtain a supply chain sentiment index, and use the order fluctuation characteristics and the supply chain sentiment index to enhance the knowledge graph to obtain an enhanced knowledge graph. The graph annotation module 203 is used to collect historical operation data and supply chain event information of the supply chain, mark the supply resilience of the enhanced knowledge graph based on the historical operation data to obtain a resilience-marked graph, analyze the risk transmission path of the supply chain disruption of the resilience-marked graph based on the supply chain event information, and mark the link risk level of the resilience-marked graph based on the risk transmission path of the supply chain disruption to obtain a link-marked graph. The intelligent analysis module 204 is used to analyze the resilience coefficient of the supply chain using the link labeling graph, quantify the stress resistance level of the supply network in the resilience labeling graph based on the resilience coefficient, obtain a target knowledge graph, and make intelligent decisions on the supply chain based on the target knowledge graph to obtain an intelligent decision report.

[0082] In detail, the modules in the knowledge graph-based intelligent supply chain decision optimization system 200 described in this embodiment of the invention employ the same methods as described above. Figure 1 The method uses the same technical means as the knowledge graph-based intelligent decision-making optimization method for supply chains described in the article, and can produce the same technical effects, so it will not be elaborated here.

[0083] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0084] Finally, it should be noted that in the above embodiments, each embodiment can be combined with each other or independent. Deleting any one of them will not affect the technical implementation of other embodiments. The above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A knowledge graph-based intelligent decision-making optimization method for supply chains, characterized in that, The method includes: Collect multi-source heterogeneous data from the supply chain, including supplier operation data, logistics dynamic data, and market feedback information, and use the multi-source heterogeneous data to construct a knowledge graph of the supply chain; Identify the order fluctuation characteristics in the logistics dynamic data, perform sentiment quantification on the market feedback information to obtain the supply chain sentiment index, and use the order fluctuation characteristics and the supply chain sentiment index to enhance the knowledge graph to obtain an enhanced knowledge graph. Collect historical operation data and supply chain event information of the supply chain. Based on the historical operation data, mark the supply chain resilience of the enhanced knowledge graph to obtain a resilience marking graph. Analyze the risk transmission path of the supply chain disruption in the resilience marking graph according to the supply chain event information. Mark the link risk level of the resilience marking graph according to the risk transmission path of the supply chain disruption to obtain a link marking graph. Using the link labeling graph, the resilience coefficient of the supply chain is analyzed. Based on the resilience coefficient, the stress resistance level of the supply network in the resilience labeling graph is quantified to obtain a target knowledge graph. Based on the target knowledge graph, intelligent decision-making is performed on the supply chain to obtain an intelligent decision report.

2. The knowledge graph-based intelligent decision-making optimization method for supply chains as described in claim 1, characterized in that, Based on the supply chain event information, analyze the transmission path of supply chain disruption risk in the resilience marker map, including: Identify the event characteristics of the supply chain event information, and quantify the event characteristics to obtain event parameters; Based on the event parameters, calculate the supply chain event information and the impact intensity of the corresponding event; Query the node association relationships of the resilience marker map, and construct the link transmission network of the resilience marker map based on the node association relationships; Based on the intensity of the impact and the link transmission network, the transmission path of the chain breakage risk in the resilience marker map is analyzed.

3. The knowledge graph-based intelligent decision-making optimization method for supply chains as described in claim 2, characterized in that, The step of analyzing the chain break risk transmission path of the resilience marker map based on the influence intensity and the link transmission network includes: Using the influence strength, the network nodes in the link transmission network are labeled with influence strength to obtain an influence-labeled network; Calculate the propagation probability of adjacent network nodes in the intensity-labeled network; Based on the propagation probability, the node links of the intensity labeling network are constructed; Analyze the total risk value of the impact of the node links, and based on the total risk value, determine the transmission path of the chain break risk in the resilience marker map.

4. The knowledge graph-based intelligent decision-making optimization method for supply chains as described in claim 1, characterized in that, The market feedback information is subjected to sentiment quantification processing to obtain the supply chain sentiment index, which includes: The market feedback information is processed into text to obtain the feedback text. The feedback text is subjected to semantic framing processing to obtain semantically framed text; Based on the semantically segmented text, the sentiment score of the supply chain is calculated; The emotion score is fused with the supply entity corresponding to the supply chain to obtain a bound emotion vector; Based on the bound sentiment vector, the sentiment index of the supply chain is calculated.

5. The knowledge graph-based intelligent decision-making optimization method for supply chains as described in claim 1, characterized in that, Using the order fluctuation characteristics and the supply chain sentiment index, the knowledge graph is augmented to obtain an augmented knowledge graph, including: The order fluctuation characteristics are decomposed into fluctuation pattern identifiers and fluctuation intensity values; The supply chain sentiment index is decomposed into a triplet to obtain the decomposed sentiment index. Based on the fluctuation pattern identifier, the logistics nodes in the knowledge graph are updated, and the transportation relationships between logistics entities in the knowledge graph are updated using the fluctuation intensity value, resulting in a preliminary enhanced graph. Using the decomposed sentiment index, the credibility attribute of the preliminary enhanced graph is constructed to obtain the enhanced knowledge graph.

6. The knowledge graph-based intelligent decision-making optimization method for supply chains as described in claim 1, characterized in that, Using the aforementioned link annotation map, the resilience coefficient of the supply chain is analyzed, including: Using the aforementioned link annotation map, the response resilience, recovery resilience, and redundancy resilience of the supply chain are analyzed to obtain three-dimensional resilience information; Based on the three-dimensional elasticity information, the elasticity index of the supply chain is quantified to obtain the elasticity quantification value; The elastic quantization value is dynamically weighted to obtain the elastic weight value; The elasticity coefficient of the supply chain is calculated based on the elasticity quantification value and the elasticity weight value.

7. The knowledge graph-based intelligent decision-making optimization method for supply chains as described in claim 1, characterized in that, Based on the historical operational data, the enhanced knowledge graph is labeled with supply resilience to obtain a resilience-labeled graph, including: Identify the node fulfillment rate, supply delay standard deviation, quality fluctuation coefficient, and interruption recovery time of the historical supply chain corresponding to the historical operational data; The node resilience value of the historical supply chain is calculated based on the node fulfillment rate, the standard deviation of supply delay, the quality fluctuation coefficient, and the interruption recovery time. Based on the node resilience values, the enhanced knowledge graph is labeled with resilience to obtain a resilience-labeled graph.

8. The knowledge graph-based intelligent decision-making optimization method for supply chains as described in claim 1, characterized in that, Based on the elasticity coefficient, the supply network in the toughness marker map is quantified for compressive strength level to obtain a target knowledge graph, including: The node toughness values ​​of the toughness marker spectrum are queried, and a weighted logarithmic operation is performed on the elastic coefficient and the node toughness values ​​to obtain the node dynamic bearing value; The dynamic load-bearing value of the node is segmented and thresholded to obtain discrete compressive strength level labels; The discrete compressive strength grade labels are numerically converted to obtain quantitative values ​​of the compressive strength grade. Using the quantified value of the compressive strength grade, attribute values ​​are assigned to the toughness marker spectrum to obtain the target knowledge graph.

9. The knowledge graph-based intelligent decision-making optimization method for supply chains as described in claim 1, characterized in that, Using the aforementioned multi-source heterogeneous data, a knowledge graph of the supply chain is constructed, including: Identify supply chain entities in the multi-source heterogeneous data and define the entity relationships of the supply chain entities; Analyze the three-dimensional attribute information in the multi-source heterogeneous data, and use the three-dimensional attribute information to perform attribute association on the entity relationship to obtain attribute-associated entity relationship; A knowledge graph of the supply chain is constructed using the supply chain entities as nodes and the attribute-related entity relationships as edges.

10. A knowledge graph-based intelligent decision-making optimization system for supply chains, characterized in that: The system includes: The knowledge graph construction module is used to collect multi-source heterogeneous data from the supply chain. The multi-source heterogeneous data includes supplier operation data, logistics dynamic data, and market feedback information. The knowledge graph of the supply chain is constructed using the multi-source heterogeneous data. The knowledge graph enhancement module is used to identify order fluctuation characteristics in the logistics dynamic data and perform sentiment quantification processing on the market feedback information to obtain a supply chain sentiment index. The knowledge graph is then enhanced using the order fluctuation characteristics and the supply chain sentiment index to obtain an enhanced knowledge graph. The graph annotation module is used to collect historical operation data and supply chain event information of the supply chain. Based on the historical operation data, the augmented knowledge graph is marked with supply resilience to obtain a resilience-marked graph. Based on the supply chain event information, the transmission path of the risk of chain breakage in the resilience-marked graph is analyzed. Based on the transmission path of the risk of chain breakage, the link risk level of the resilience-marked graph is marked to obtain a link-marked graph. The intelligent analysis module is used to analyze the resilience coefficient of the supply chain using the link labeling graph, quantify the stress resistance level of the supply network in the resilience labeling graph based on the resilience coefficient, obtain a target knowledge graph, and make intelligent decisions on the supply chain based on the target knowledge graph to obtain an intelligent decision report.

Images