Real-time collaborative supply chain supervision system and method

Abstract

The invention relates to the technical field of supply chain supervision, in particular to a real-time collaborative supply chain supervision system, which comprises a multi-dimensional data atlas construction module, an atlas dynamic association response module and a cross-dimensional atlas node supplement module, calculating a correlation coefficient of each node in the plurality of multi-dimensional data maps; the graph dynamic association response module is used for unifying events in the same entity before the multi-dimensional data graph is constructed; the method comprises the following steps: acquiring a multi-dimensional data atlas initial event, defining a multi-dimensional data atlas name according to the multi-dimensional data atlas initial event, when a new event occurs, matching the multi-dimensional data atlas name with the same event type label as the new event, and outputting a new event reminding signal according to the relationship of each node in the multi-dimensional data atlas; and the cross-dimension atlas node supplementing module is used for supplementing the initial upper layer node in each multi-dimension data atlas and synchronously outputting a new event reminding signal.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of supply chain supervision, in particular to a real-time collaborative supply chain supervision system and method. BACKGROUND

[0002] A supply chain system is usually composed of multiple levels of participants around a core enterprise, including upstream raw material suppliers, component manufacturers, midstream finished product manufacturers, processing and assembly plants, downstream distributors, retailers, logistics service providers, and financial institutions, technical support providers throughout the entire chain, etc. Due to the significant differences in the functions of each link in the supply chain, for example, suppliers focus on the procurement, quality inspection and timely delivery of raw materials, manufacturers focus on production plan execution, capacity scheduling and quality control, logistics providers are responsible for warehouse management, transportation route planning and in-transit monitoring, and retailers focus on terminal sales data collection and inventory turnover optimization, in order to achieve efficient collaboration, accurate response to market demand, and smooth connection of information flow, logistics and capital flow, each link will output event information of different dimensions and granularities according to its business scenarios and management needs. The output of event information is essentially to break down the information barriers between each link and enable key information scattered in the supply chain network to flow and be shared. Through the integration of event information, a complete data portrait covering the entire process of procurement, production, warehousing, transportation and sales can be formed, so as to gain a comprehensive understanding of the operation status of the supply chain, including real-time operation dynamics of each link (such as whether raw materials are delivered on time, production lines are advancing according to the plan, and goods are successfully delivered to the terminal), and deep-seated problems and opportunities (such as the reasons for the inventory accumulation of a certain type of product, and which regions have rapidly growing market demand). When a participant outputs event information, due to the complex hierarchical structure and associated network characteristics of the supply chain, the association of some lower-level participants is hidden and difficult to identify through the surface business link. At the same time, the data of each participant is scattered in independent systems, and the event naming is heterogeneous and the format is not unified, making it difficult for the output participant to accurately match all associated lower-level participants, and thus unable to completely output event information to the associated participants. In view of this, we propose a real-time collaborative supply chain supervision system and method. SUMMARY

[0003] The purpose of this invention is to solve the problem that when participating entities output event information, due to the complexity of the supply chain hierarchy and the characteristics of the associated network, the relationships between some lower-level entities are hidden and difficult to identify by relying solely on the surface business links; and the data of each entity is scattered in independent systems, with heterogeneous event naming and inconsistent formats, making it difficult for the outputting entity to accurately match all associated lower-level participating entities and thus unable to fully output event information to associated entities.

[0004] To achieve the above objectives, this invention provides a real-time collaborative supply chain monitoring system, comprising a multi-dimensional data graph construction module, a dynamic graph association response module, and a cross-dimensional graph node supplementation module. The multi-dimensional data graph construction module constructs multiple multi-dimensional data graphs starting with each historical event and calculates the association coefficients of each node in the multiple multi-dimensional data graphs. The dynamic graph association response module unifies events within the same entity before constructing the multi-dimensional data graph; it defines the multi-dimensional data graph name using the starting event; when a new event occurs, it matches the multi-dimensional data graph name with the same event type label as the new event and outputs a new event alert signal based on the relationships between nodes in the multi-dimensional data graph. The cross-dimensional graph node supplementation module supplements the starting upper-level nodes in each multi-dimensional data graph and synchronously outputs new event alert signals.

[0005] As a further improvement to this technical solution, the multi-dimensional data graph construction module includes a graph node determination unit and a node correlation coefficient calculation unit; the graph node determination unit is used to calculate the correlation coverage between multiple historical events and determine each node in the multi-dimensional data graph; the node correlation coefficient calculation unit is used to calculate the correlation coefficient between each node and determine the edges of each node in the multi-dimensional data graph.

[0006] As a further improvement to this technical solution, the graph node determination unit senses multiple historical events in each participating subject, randomly retrieves event A from a certain participating subject, and calculates the association coverage rate of event A with historical events in other participating subjects one by one, setting an association threshold. If the association coverage rate between event A and event B in participating subject B is greater than the association threshold, then the causal relationship between event A and event B in participating subject B is determined to be event A→event B. Then, the association coverage rate between event B and event C in participating subject C is analyzed again until the association coverage rate is less than or equal to the association threshold. Starting with event A, events with causal relationships are identified one by one and defined as nodes in the multi-dimensional data graph. The causal relationship serves as the classification criterion for upper and lower levels in the multi-dimensional data graph: if the causal relationship between event A and event B is event A→event B, then event A is the upper-level node and event B is the lower-level node.

[0007] As a further improvement to this technical solution, the determination of the association coverage rate of the map node includes: sensing the timestamps of multiple occurrences of event A, sorting them into a time series according to the timestamps of event A, and setting a duration window; starting with the timestamp of each occurrence of event A, after retrieving the timestamp of each occurrence of event A, starting with the timestamp of occurrence, retrieving multiple events occurring in other participating subjects within the time window, and calculating the ratio between each event and the number of occurrences of event A, which is the association coverage rate.

[0008] The beneficial effects of the aforementioned further solutions are that, by sensing the historical events of participating entities, calculating the coverage rate of associations with other events starting from event A, and determining causal relationships based on association thresholds, multi-dimensional data graph nodes can be constructed, with causal relationships serving as the basis for upper and lower level classification. The association coverage rate is analyzed through timestamps and time window settings, thereby accurately quantifying the event association ratio, realizing supply chain event causal mining and multi-dimensional data graph construction, improving the accuracy of event association identification, providing support for insights into supply chain operation patterns and risk warnings, solving the problem of difficult event association identification in complex supply chains, and assisting in efficient supervision and decision-making.

[0009] Based on the above technical solution, the present invention can be further improved as follows: The node correlation coefficient calculation unit senses the historical data corresponding to each node in the multi-dimensional data map, constructs the corresponding historical data into a historical data set according to the time sequence, and then calculates the correlation coefficient between every two historical data based on causal relationships: The numerator for calculating the correlation coefficient is obtained by subtracting the mean of the corresponding historical dataset from a single historical data point of an event, multiplying it by the sum of the sum of the products of a single historical data point of another event and the mean of the historical dataset of another event, and then summing all the products. First, calculate the sum of squared deviations from the mean of one set of historical data for an event, then calculate the sum of squared deviations from the mean of another set of historical data for an event. Then multiply the two sums of squared deviations from the mean and take the square root of the product, which is the denominator. Divide the numerator by the denominator to get the correlation coefficient between each pair of historical data. The correlation coefficient is limited to the range of -1 and 1. The closer the absolute value of the correlation coefficient is to 1, the stronger the correlation between the corresponding nodes of the two historical data. When the correlation coefficient is >0, the two nodes are considered to be positively correlated. When the correlation coefficient is <0, the two nodes are considered to be negatively correlated.

[0010] The beneficial effects of the above-mentioned further solutions are that by calculating the correlation coefficient through the node correlation coefficient calculation unit, the correlation relationship between node data can be quantified, and potential correlation patterns between historical data can be discovered; it can assist in the diagnosis of operational status, identify key data linkage links by the strength and direction of correlation, accurately locate the abnormal transmission path, predict the direction and scope of risk diffusion in the data correlation network, enhance the resilience of the supply chain to cope with fluctuations, and promote the upgrading of supply chain management towards refinement and intelligence.

[0011] Based on the above technical solution, the present invention can be further improved as follows: The dynamic association response module of the graph includes a type label generation unit and a new event reminder output unit. Before the graph node determination unit calculates the association coverage, the type label generation unit sequentially analyzes the historical data of the events corresponding to each participating subject, sequentially retrieves two events existing in the same participating subject, calculates the similarity between the historical data corresponding to the two events, sets an event similarity threshold, and if the similarity between the historical data corresponding to the two events is ≤ the event similarity threshold, then it is determined that the two events are not of the same category; if the similarity is > the event similarity threshold, then it is determined that the two events are of the same category.

[0012] As a further improvement to this technical solution, the type label generation unit retrieves historical data of two events existing in the same participating subject and constructs them into historical data sets according to the chronological order of the events. The data in the two historical data sets are multiplied and summed to obtain the numerator for similarity calculation. Then, the data in each historical data set is squared, and the sum of all squared results is obtained. The square root of the sum is taken to obtain the norm of each data set. Finally, the two norms are multiplied to obtain the denominator for similarity calculation. The numerator divided by the denominator gives the similarity between the historical data corresponding to the two events. The new event reminder output unit defines the event type label at the beginning of the event as the name of the multi-dimensional data graph. When a new event occurs, the type label generation unit sets the event type label of the new event again, matches the multi-dimensional data graph name with the same event type label as the new event, retrieves the events in the lower-level participating subject based on the multi-dimensional data graph, and outputs a reminder signal. The reminder signal includes the correlation coefficient of each event in the multi-dimensional data graph and the correlation in the node correlation coefficient calculation unit. It also counts the number of times each event name appears in the same category of events, compares the number of times each event name appears, and defines the event name that appears most frequently as the event type label.

[0013] The beneficial effects of the above-mentioned further solutions are that the type label generation unit can accurately cluster events, unify naming standards, lay a solid foundation for graph construction and association analysis, and improve the efficiency of event identification and classification; the new event reminder output unit realizes dynamic association of events and graph adaptation, quickly locates the associated lower-level subject when a new event is triggered, and outputs reminders with quantified association information, which helps real-time monitoring of supply chain events, risk warning and linkage response, optimizes collaborative efficiency, enhances the supply chain's ability to perceive and handle events, and promotes the implementation of data-driven intelligent supply chain management.

[0014] Based on the above technical solution, the present invention can be further improved as follows: The cross-dimensional graph node supplementation module receives multiple multi-dimensional data graphs, sequentially retrieves the name of each multi-dimensional data graph, matches the nodes of each multi-dimensional data graph name in other dimension data graphs, retrieves the upper-level nodes in other dimension data graphs, and supplements the upper-level nodes to the multi-dimensional data graph. As a further improvement to this technical solution, after receiving a new event, the new event reminder output unit matches the name of the multi-dimensional data map that is the same as the event type label of the new event, and calls up the upper-level event node corresponding to the matching multi-dimensional data map to output the same reminder signal as the new event reminder output unit.

[0015] The beneficial effect of the above-mentioned further solution is that, by receiving multiple multi-dimensional data graphs through the cross-dimensional graph node supplementation module, retrieving the name of each graph, matching its nodes and upper-level nodes in other dimension graphs, and supplementing them to the original graph, it is possible to link cross-dimensional graph nodes and supplement upper-level nodes, break the isolation of dimensional data, build a more comprehensive and three-dimensional supply chain data association network, enable the connection of events in different dimensions, improve the coverage and characterization accuracy of data graphs on complex supply chain relationships, provide complete data support for cross-dimensional event analysis, full-chain risk tracing, etc., help to discover multi-dimensional collaborative patterns, optimize the overall decision-making and management of the supply chain, and enhance the adaptability and analytical depth to complex business scenarios.

[0016] A real-time collaborative supply chain supervision method includes the following steps: S1. Construct multiple multi-dimensional data graphs starting with each event, and calculate the correlation coefficients of each node in the multiple multi-dimensional data graphs; S2. Before constructing the multi-dimensional data graph, unify the events in the same entity; define the name of the multi-dimensional data graph; when a new event occurs, match the name of the multi-dimensional data graph with the same event type label as the new event, and output a new event reminder signal based on the relationship between the nodes in the multi-dimensional data graph. S3. Supplement the starting upper-level node in each multi-dimensional data graph and synchronously output new event reminder signals.

[0017] In addition to the objectives, features, and advantages described above, the present invention has other objectives, features, and advantages. The invention will now be described in further detail with reference to the figures. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the overall module working principle of the present invention; Figure 2 This is a schematic diagram illustrating the working principle of the multi-dimensional data map construction module in this invention; Figure 3 This is a schematic diagram illustrating the working principle of the dynamic correlation response module in the spectrum of the present invention. Figure 4 This is a schematic diagram illustrating the working principle of the cross-dimensional map node supplementation module in this invention for supplementing multi-dimensional data maps.

[0019] The meanings of the labels in the diagram are as follows: 100. Multi-dimensional data graph construction module; 110. Graph node determination unit; 120. Node correlation coefficient calculation unit; 200. Graph dynamic correlation response module; 210. Type label generation unit; 220. New event reminder output unit; 300. Cross-dimensional graph node supplementation module. Detailed Implementation

[0020] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] refer to Figures 1-4The system describes a real-time collaborative supply chain monitoring system, comprising a multi-dimensional data graph construction module 100, a dynamic graph association response module 200, and a cross-dimensional graph node supplementation module 300. The multi-dimensional data graph construction module 100 constructs multiple multi-dimensional data graphs starting from each historical event and calculates the association coefficients of each node in the multiple multi-dimensional data graphs. The dynamic graph association response module 200 unifies events within the same entity before constructing the multi-dimensional data graphs and defines the multi-dimensional data graph name using the starting event. When a new event occurs, it matches the multi-dimensional data graph name with the same event type label as the new event and outputs a new event alert signal based on the relationships between nodes in the multi-dimensional data graph. The cross-dimensional graph node supplementation module 300 supplements the starting upper-level nodes in each multi-dimensional data graph and synchronously outputs new event alert signals.

[0022] The multi-dimensional data graph construction module 100 includes a graph node determination unit 110 and a node correlation coefficient calculation unit 120. The graph node determination unit 110 senses multiple historical events (such as the supplier's "raw material delivery record" and the manufacturer's "production interruption record") in each participating entity, randomly selects event A in a certain participating entity, and analyzes the correlation coverage of event A with historical events in other participating entities one by one, starting with event A: senses the timestamps of multiple occurrences of event A, and sorts them into a time series according to the timestamps of event A. Window for setting duration ; Recall Event A Timestamp of the occurrence For duration window Starting with the timestamp of each occurrence of event A, retrieve the time window starting from that timestamp. For the other participating entities, calculate the ratio between the number of times each event occurs and the number of times event A occurs; this ratio is the association coverage rate. Set an association threshold. If the association coverage rate of event A and event B in participant B is greater than the association threshold, then determine the causal relationship between event A and event B in participant B as event A→event B. Then analyze the association coverage rate between event B and event C in participant C again until the association coverage rate is less than or equal to the association threshold. Starting with event A, events with causal relationships will be identified one by one and defined as nodes in the multi-dimensional data graph. The causal relationship will also serve as the classification criterion for upper and lower levels in the multi-dimensional data graph: if the causal relationship between event A and event B is event A→event B, then event A is the upper-level node and event B is the lower-level node.

[0023] The node correlation coefficient calculation unit 120 senses the historical data corresponding to each node in the multi-dimensional data map, constructs the corresponding historical data into a historical data set, and then calculates the correlation coefficient between the historical data based on causal relationships. The working principle is as follows: Subtract the mean of the corresponding historical data set from a single historical data point of an event, multiply by the sum of the sums ... , Then calculate the correlation coefficient between event A and event B: ,in For historical data The mean, For historical data The mean, The closer the absolute value is to 1, the stronger the correlation; if the correlation coefficient is close to 1, the stronger the correlation. If the correlation coefficient is greater than 0, it indicates that the two nodes are positively correlated. If the correlation coefficient is greater than 0, it indicates that the two nodes are negatively correlated. The correlation coefficient will be defined as the edge of the node in the multi-dimensional data graph. This will allow for the construction of multiple multi-dimensional data graphs for each participating entity based on different events, thereby structurally storing the correlation patterns of complex events in the supply chain. This will provide data support for long-term operational optimization and process improvement, thereby enhancing the resilience, responsiveness, and management sophistication of the supply chain as a whole, and shifting supply chain supervision from "passive response" to "proactive prediction and precise collaboration".

[0024] This invention further considers that, during the calculation of correlation coverage, due to the complex composition and diverse levels of participants in the supply chain system, the name of event A in participant A may be expressed in various ways (e.g., "delivery delay" on the supplier side and "raw material arrival delay" on the manufacturer side, which are actually the same supply chain transmission event). This can lead to heterogeneous naming of events that were originally the same, but are classified as independent events, causing omissions of causal relationships. This can further lead to inaccurate correlation coverage calculations by the graph node determination unit 110. The graph dynamic correlation response module 200 includes a type label generation unit 210 and a new event reminder output unit 220. Before calculating the correlation coverage of each event, the type label generation unit 210 analyzes the historical data of the corresponding events of each participant in turn, and calculates the similarity between the historical data: it sequentially retrieves two events existing in the same participant, as well as the historical data corresponding to the two events. The two events retrieved are event W and event E, and the historical data of event W is... The historical data for event E is Then the similarity between event W and event E is ; Set an event similarity threshold; if the similarity between event W and event E is... If the similarity is less than or equal to the event similarity threshold, then events W and E are determined not to be of the same category; if the similarity is less than or equal to the event similarity threshold, then events W and E are determined not to be of the same category. If the event similarity threshold is reached, then event W and event E are determined to be events of the same category. Count the number of times each event name appears in the same category of events, compare the number of times each event name appears, and define the event name that appears most frequently as the event type label.

[0025] In the supply chain ecosystem, although different participants (such as suppliers, manufacturers, and logistics providers) are related in their business processes, their historical data generated during operations may appear similar due to factors such as data collection dimensions (e.g., both record "time delay" data, but suppliers focus on delivery delays while manufacturers emphasize production stoppage delays), indicator definition methods (e.g., the criteria for judging "cost anomalies" differ between financial and operational perspectives), and semantic comprehension biases ("order anomalies" refer to order volume fluctuations for suppliers but order fulfillment issues for retailers). However, behind these similar historical data, due to differences in the participants' business scenarios, goals, and rules, the actual nature of the events they represent (whether it's a fulfillment issue, production failure, or market fluctuation, etc.) is drastically different. For example, a supplier's "delivery delay of 2 days" data reflects a fulfillment event in the logistics and distribution process, while a manufacturer's "production stoppage of 2 days" data may stem from a production event caused by equipment failure. There may be situations where two participating entities have similar historical data but express different events. Therefore, the type label generation unit 210 retrieves events existing within the same participating entity and calculates similarity. If the type label generation unit 210 directly calculates similarity across participating entities, the results may not accurately reflect the true relationship between events due to differences in business logic, data generation rules, and semantic understanding between different participating entities (such as different judgment criteria and focus dimensions for "delayed" events between suppliers and manufacturers). However, by focusing on the same participating entity, whose internal business processes are consistent and data features are self-consistent, retrieving two events within that entity to calculate similarity can avoid problems such as cross-entity semantic confusion and structural differences in feature dimensions. This effectively identifies events within the entity that are essentially the same but have different names or forms of expression, accurately solving the problem of heterogeneous event naming caused by the diversity of participating entities. This lays a reliable data foundation for subsequent correlation coverage calculation, event category merging, and correlation coefficient solving, ensuring the accuracy and effectiveness of supply chain event correlation analysis. Furthermore, when the type label generation unit 210 categorizes events with different names into the same category, the event data that was originally scattered and independent due to heterogeneous naming is integrated into a unified event category. This allows the node correlation coefficient calculation unit 120 to calculate correlation coefficients (such as analyzing the degree of linear correlation between events) so that event data that is classified into the same category will no longer be separated due to differences in name. This enables the data to be included in the calculation scope as valid data, effectively improving the richness of historical data and the reliability of analysis results, and making the correlation coefficient more accurately reflect the true correlation strength between events.

[0026] Since the graph node determination unit 110 constructs a multi-dimensional data graph starting with event A, to facilitate the rapid retrieval of the multi-dimensional data graph corresponding to a new event when a new event occurs in the participating entities, the new event alert output unit 220 defines the multi-dimensional data graph name as the event type label at the beginning of the event. When a new event occurs, the type label generation unit 210 sets the event type label of the new event again, matches the multi-dimensional data graph name with the same event type label as the new event, retrieves the event in the lower-level participating entities based on the multi-dimensional data graph, and outputs a new event alert signal. The new event alert signal includes the correlation coefficients of each event in the multi-dimensional data graph. And correlation (positive correlation, negative correlation); Because the type label generation unit 210 has grouped events with different names but the same essence into the same category and assigned them unique event type labels, it has achieved standardization and normalization of event naming. Therefore, the new event reminder output unit 220 can define the event type label at the beginning of the event as the name of the multi-dimensional data graph. By unifying the event type label, it eliminates the confusion caused by the heterogeneous naming of the graph node determination unit 110 in constructing multiple multi-dimensional data graphs, ensuring that each multi-dimensional data graph has a clear and unique identifier. Furthermore, by setting the event type label again through the type label generation unit 210, it can quickly match the multi-dimensional data graphs with the same name, avoiding graph search delays or errors caused by differences in event names or too many subject levels in the complex supply chain network. This allows for efficient retrieval of the related events of the lower-level participating subjects in the graph, as well as the corresponding correlation coefficients and correlation information, and outputs new event reminder signals.

[0027] To avoid incomplete causal chains in new events due to insufficient upstream coverage of a single multi-dimensional data graph, the cross-dimensional graph node supplementation module 300 receives multiple multi-dimensional data graphs, sequentially retrieves the name of each multi-dimensional data graph, matches the nodes of each multi-dimensional data graph name in other dimension data graphs, and retrieves the upper-level nodes (i.e., preceding related events) in the other dimension data graphs, supplementing the multi-dimensional data graph with these upper-level nodes. Upon receiving a new event, the new event alert output unit 220 matches the name of the multi-dimensional data graph with the same event type label as the new event, retrieves the upper-level event node corresponding to the matched multi-dimensional data graph, and outputs the same new event alert signal as the new event alert output unit 220. Because the event type label at the start of the event is defined as the name of the multi-dimensional data graph by the new event reminder output unit 220, the accuracy of matching the new event with the multi-dimensional data graph will not be further affected when the cross-dimensional graph node supplementation module 300 calls up the upper-level nodes of the new event in multiple multi-dimensional data graphs. At the same time, it breaks the information limitations of a single multi-dimensional data graph. By integrating the upstream nodes in multiple multi-dimensional data graphs, the causal relationship chain of the multi-dimensional data graph is improved, so that the participating entities can not only grasp the downstream impact that the new event may cause, but also clearly understand its potential upstream driving factors. This realizes the two-way flow of event information (both warning to the downstream and feedback to the upstream), promotes the collaborative response of cross-level participating entities, avoids the risk omission caused by information fragmentation, further enhances the depth and foresight of supply chain supervision, and provides more complete related data support for efficient decision-making across the entire chain.

[0028] A real-time collaborative supply chain supervision method includes the following steps: S1. Construct multiple multi-dimensional data graphs starting with each event, and calculate the correlation coefficients of each node in the multiple multi-dimensional data graphs; S2. Before constructing the multi-dimensional data graph, unify the events in the same entity; define the name of the multi-dimensional data graph; when a new event occurs, match the name of the multi-dimensional data graph with the same event type label as the new event, and output a new event reminder signal based on the relationship between the nodes in the multi-dimensional data graph. S3. Supplement the starting upper-level node in each multi-dimensional data graph and synchronously output new event reminder signals.

[0029] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A real-time collaborative supply chain monitoring system, characterized in that, The system includes a multi-dimensional data graph construction module (100), a dynamic graph association response module (200), and a cross-dimensional graph node supplementation module (300). Specifically: the multi-dimensional data graph construction module (100) constructs multiple multi-dimensional data graphs starting from each historical event, and calculates the association coefficients of each node in the multiple multi-dimensional data graphs; the dynamic graph association response module (200) unifies events within the same entity before constructing the multi-dimensional data graph; it defines the multi-dimensional data graph name using the starting event of the multi-dimensional data graph, and when a new event occurs, it matches the multi-dimensional data graph name with the same event type label as the new event, outputting a new event reminder signal based on the relationships between nodes in the multi-dimensional data graph; the cross-dimensional graph node supplementation module (300) supplements the starting upper-level nodes in each multi-dimensional data graph, synchronously outputting a new event reminder signal.

2. The real-time collaborative supply chain monitoring system according to claim 1, characterized in that: The multi-dimensional data graph construction module (100) includes a graph node determination unit (110) and a node correlation coefficient calculation unit (120); the graph node determination unit (110) is used to calculate the correlation coverage between multiple historical events and determine each node in the multi-dimensional data graph; the node correlation coefficient calculation unit (120) is used to calculate the correlation coefficient between each node and determine the edges of each node in the multi-dimensional data graph.

3. The real-time collaborative supply chain monitoring system according to claim 2, characterized in that: The graph node determination unit (110) senses multiple historical events in each participating subject, randomly retrieves event A in a certain participating subject, and calculates the association coverage rate of event A with historical events in other participating subjects one by one, starting with event A, and sets an association threshold. If the association coverage rate of event A and event B in participating subject B is greater than the association threshold, then the causal relationship between event A and event B in participating subject B is determined to be event A → event B. Then, the association coverage rate between event B and event C in participating subject C is analyzed again until the association coverage rate is less than or equal to the association threshold. Starting with event A, events with causal relationships are identified one by one and defined as nodes in the multi-dimensional data graph. The causal relationship serves as the classification criterion for upper and lower levels in the multi-dimensional data graph: if the causal relationship between event A and event B is event A→event B, then event A is the upper-level node and event B is the lower-level node.

4. The real-time collaborative supply chain monitoring system according to claim 3, characterized in that: The association coverage of the graph node determination unit (110) includes: sensing the timestamps of multiple occurrences of event A, sorting them into a time series according to the timestamps of event A, and setting a duration window; starting from the timestamp of each occurrence of event A, after retrieving the timestamp of each occurrence of event A, starting from the timestamp of occurrence, retrieving multiple events occurring in other participating subjects within the time window, and calculating the ratio between each event and the number of occurrences of event A, which is the association coverage.

5. The real-time collaborative supply chain monitoring system according to claim 2, characterized in that: The node correlation coefficient calculation unit (120) senses the historical data corresponding to each node in the multi-dimensional data map, constructs the corresponding historical data into a historical data set according to the time order, and then calculates the correlation coefficient between each pair of historical data based on causal relationships: The numerator for calculating the correlation coefficient is obtained by subtracting the mean of the corresponding historical dataset from a single historical data point of an event, multiplying it by the sum of the sum of the products of a single historical data point of another event and the mean of the historical dataset of another event, and then summing all the products. First, calculate the sum of squared deviations from the mean of one set of historical data for an event, then calculate the sum of squared deviations from the mean of another set of historical data for an event. Then multiply the two sums of squared deviations from the mean and take the square root of the product, which is the denominator. Divide the numerator by the denominator to get the correlation coefficient between each pair of historical data. The correlation coefficient is limited to the range of -1 and 1. The closer the absolute value of the correlation coefficient is to 1, the stronger the correlation between the corresponding nodes of the two historical data. When the correlation coefficient is >0, the two nodes are considered to be positively correlated. When the correlation coefficient is <0, the two nodes are considered to be negatively correlated.

6. The real-time collaborative supply chain monitoring system according to claim 2, characterized in that: The graph dynamic association response module (200) includes a type label generation unit (210) and a new event reminder output unit (220). Before the graph node determination unit (110) calculates the association coverage, the type label generation unit (210) sequentially analyzes the historical data of each participating subject's corresponding event, sequentially retrieves two events existing in the same participating subject, calculates the similarity between the historical data corresponding to the two events, sets an event similarity threshold, and if the similarity between the historical data corresponding to the two events is ≤ the event similarity threshold, then it is determined that the two events are not of the same category; if the similarity is > the event similarity threshold, then it is determined that the two events are of the same category.

7. The real-time collaborative supply chain monitoring system according to claim 6, characterized in that: The type label generation unit (210) retrieves the historical data of two events existing in the same participating subject and constructs them into a historical data set according to the time order of the events. The data in the two historical data sets are multiplied and summed to obtain the numerator for calculating similarity. Then, the data in each historical data set are squared, and the results of all squares are added together to obtain the sum. The square root of the sum is taken to obtain the norm of each data set. Finally, the two norms are multiplied to obtain the denominator for calculating similarity. The numerator divided by the denominator is the similarity between the historical data corresponding to the two events. The new event reminder output unit (220) defines the event type label at the beginning of the event as the name of the multi-dimensional data map. When a new event occurs, the type label generation unit (210) sets the event type label of the new event again, matches the multi-dimensional data map name with the same event type label as the new event, retrieves the events in the lower-level participating subject according to the multi-dimensional data map, and outputs a reminder signal. The reminder signal includes the correlation coefficient of each event in the multi-dimensional data map and the correlation in the node correlation coefficient calculation unit (120). It also counts the number of times each event name appears in the same category of events, compares the number of times each event name appears, and defines the event name that appears most frequently as the event type label.

8. The real-time collaborative supply chain monitoring system according to claim 2, characterized in that: The cross-dimensional graph node supplementation module (300) receives multiple multi-dimensional data graphs, sequentially retrieves the name of each multi-dimensional data graph, matches the nodes of each multi-dimensional data graph name in other dimension data graphs, retrieves the upper-level nodes in other dimension data graphs, and supplements the upper-level nodes to the multi-dimensional data graph.

9. The real-time collaborative supply chain monitoring system according to claim 7, characterized in that: After receiving a new event, the new event reminder output unit (220) matches the name of the multi-dimensional data map that is the same as the event type label of the new event, and calls up the upper-level event node corresponding to the matching multi-dimensional data map to output the same reminder signal as the new event reminder output unit (220).

10. A real-time collaborative supply chain monitoring method, applied to the real-time collaborative supply chain monitoring system according to any one of claims 1-9, characterized in that, Includes the following steps: S1. Construct multiple multi-dimensional data graphs starting with each event, and calculate the correlation coefficients of each node in the multiple multi-dimensional data graphs; S2. Before constructing a multi-dimensional data graph, unify the events within the same entity; Define a multi-dimensional data graph name. When a new event occurs, match the multi-dimensional data graph name that is the same as the event type label of the new event, and output a new event reminder signal based on the relationship between the nodes in the multi-dimensional data graph. S3. Supplement the starting upper-level node in each multi-dimensional data graph and synchronously output new event reminder signals.

Images