Industrial chain upstream and downstream cooperative relationship strength analysis method

Abstract

The invention belongs to the technical field of industrial collaboration analysis, and particularly discloses an industrial chain upstream and downstream collaboration relationship strength analysis method, which comprises the following steps: constructing a collaboration event sequence based on business serial number alignment, generating a collaboration event association link by using time sequence causal analysis, and combining delay identification and delay propagation mining to obtain an industrial chain upstream and downstream collaboration relationship strength analysis result. The method achieves the dynamic quantification of the intensity of the cooperation relation from the microcosmic event outbound, can accurately describe the upstream and downstream response dependence through the evaluation of the cooperation relation, remarkably improves the real-time performance, interpretability and operability of the perception of the cooperation state, and improves the detection precision of the cooperation state. Furthermore, after cooperation intensity dynamic quantification is carried out on a cooperation event association link based on delay identification and delay propagation mining, early warning is carried out by capturing a delay propagation section and carrying out risk feature identification on the delay propagation section, fine-grained early warning of the cooperation relationship in the spatial dimension is realized, and the accuracy of the early warning is improved. And the problems of coarse early warning granularity, weak guidance, response lag and the like are avoided to a certain extent.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of industrial cooperation analysis, and specifically discloses an upstream and downstream industrial chain cooperation relationship strength analysis method. BACKGROUND

[0002] With the continuous refinement of industrial division of labor, modern enterprises no longer operate as isolated economic units, but are deeply embedded in an industrial network composed of suppliers, manufacturers, distributors and other parties. In this context, the competition paradigm between enterprises has evolved from a single-party confrontation to overall competition between industrial chains. The synergy efficiency between upstream and downstream enterprises in the industrial chain has become a key factor in determining the overall value chain operation quality. Therefore, building dynamic perception capability of the cooperation state of the upstream and downstream of the industrial chain is crucial to improving the resilience of the entire industrial chain.

[0003] There are related researches on the perception of industrial chain synergy in the prior art, such as the enterprise synergy system based on industrial chain proposed in Chinese patent CN117311789A. This scheme builds a hierarchical architecture, including: a basic resource layer for realizing the connection and information interaction of upstream and downstream enterprises in the industrial chain, supporting industrial chain data fusion and decentralized data sharing based on blockchain; a service component layer providing pluggable and reusable service modules to support flexible deployment of cross-enterprise collaboration functions; and a service evolution layer that realizes dynamic response to the running state of the industrial chain through self-organizing collaborative development, scenario adaptive perception and service self-evolution mechanism.

[0004] Although the above scheme realizes the active perception and adaptive evolution capability of the industrial chain synergy system by introducing the crowd wisdom synergy mode and the service component self-evolution mechanism, the cooperation state perception mainly relies on macro-level data sharing records and service call logs, lacking fine-grained time sequence behavior analysis of specific business interaction events between upstream and downstream enterprises in the industrial chain, such as order creation, delivery notification, in-transit tracking, etc. The cooperation state evaluation can only rely on subjective evaluation, rather than objective business flow data analysis. However, the cooperation of the industrial chain in reality is essentially a time sequence link composed of a series of ordered and traceable micro events, and the change of its cooperation state has a clear behavior path. Therefore, the evaluation method that is independent of specific event sequences cannot accurately reflect the real-time, dynamic and explainability of the cooperation relationship, and the evaluation result is easily affected by model assumption bias, making it difficult to support precise synergy optimization and risk intervention.

[0005] Furthermore, under this macro-analysis paradigm, when potential risks are identified in collaborative relationships, it is impossible to pinpoint the anomaly to specific business processes or event nodes. This results in a lack of spatial operability in the generated early warning information, leading to problems such as coarse-grained warnings, weak guidance, and delayed responses, making it difficult to meet the needs for refined governance and rapid closed-loop handling of the supply chain's operational status. Summary of the Invention

[0006] Therefore, one objective of this application is to provide a method for analyzing the strength of collaborative relationships between upstream and downstream industries. By analyzing the strength of collaborative relationships and providing fine-grained early warnings based on real business interaction data and integrating temporal causality and delayed propagation, this method effectively solves the problems mentioned in the background art.

[0007] The objective of this invention can be achieved through the following technical solution: A method for analyzing the strength of upstream and downstream collaborative relationships in an industry chain, comprising the following steps: Step 1: Real-time acquisition of business interaction data from enterprise resource planning system and industry chain management system, extraction of collaborative events and event timestamps, and construction of a collaborative event sequence based on business serial number alignment, including participating enterprises and collaborative events.

[0008] Step 2: Based on the collaborative event sequence, the sliding window method is used to extract adjacent collaborative event pairs, and event pairs with significant causal correlation are screened through temporal causality test to construct a collaborative event association link with enterprises as nodes and business stages as edges.

[0009] Step 3: Perform response time deviation analysis on the collaborative event association links, identify delayed adjacent collaborative event pairs, and determine the collaborative relationship strength of the collaborative event association links based on the time-series propagation analysis of delayed adjacent collaborative event pairs.

[0010] Step 4: Locate the consecutively occurring delayed adjacent collaborative event segments in the collaborative event association link as the delayed propagation segment.

[0011] Step 5: Identify the risk characteristics of the delayed propagation segment and generate supply chain early warning information based on the risk characteristics of the delayed propagation segment.

[0012] Combining all the above technical solutions, the positive effects of this invention are as follows: 1. This invention generates collaborative event association links by constructing a collaborative event sequence based on business serial number alignment and using time-series causal analysis. Combined with delay identification and delay propagation mining, it realizes dynamic quantification of the strength of collaborative relationships from the micro-event flow. Its evaluation of collaborative relationships can accurately characterize the upstream and downstream response dependencies, significantly improving the real-time performance, interpretability, and operability of collaborative status perception.

[0013] 2、The application further captures the delay propagation section formed by continuous delay after realizing the dynamic quantification of cooperation strength through delay identification and propagation path mining on the basis of cooperation event correlation link, and performs risk feature labeling and semantic identification on the section, realizing spatial fine-grained risk early warning for key links. The mechanism breaks through the limitations of traditional macro early warning, significantly improves the positioning accuracy and interpretability of early warning, effectively alleviates the problems of coarse early warning granularity, insufficient guidance and response lag, and supports precise intervention and closed-loop disposal of bottleneck links in the industrial chain. BRIEF DESCRIPTION OF DRAWINGS

[0014] The application will be further described with the help of the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the application. For ordinary skilled in the art, other drawings can be obtained without creative labor on the basis of the following drawings.

[0015] Figure 1 The method embodiment of the application is illustrated by the flowchart.

[0016] Figure 2 The flowchart of the construction of the cooperation event correlation link in the application is illustrated.

[0017] Figure 3 The cooperation event correlation link in the application is illustrated by the schematic diagram. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the application will be described clearly and completely with the help of the accompanying drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by ordinary skilled in the art without creative labor are within the protection scope of the application.

[0019] Referring to Figure 1 The application proposes an upstream and downstream cooperation relationship strength analysis method in an industrial chain, including the following steps: Step 1: real-time acquisition of business interaction data from enterprise resource planning systems and industrial chain management systems, extraction of cooperation events and event time stamps, construction of cooperation event sequences containing participating enterprises and cooperation events based on business serial number alignment.

[0020] As an application of the above steps, the enterprise resource planning system refers to an information system for integrating and managing the core business processes of an enterprise, such as orders, production, inventory, etc. The typical functional modules include order management, which can provide order creation, change and completion events and their timestamps. The industry chain management system usually corresponds to the supply chain management system, which refers to a system that supports cross-enterprise logistics collaboration, covering logistics planning, transportation tracking and warehouse management functions. The logistics management interface can output delivery, in-transit, and signed events and corresponding timestamps.

[0021] In a heterogeneous system, the same business entity, such as an order, may have different representations in different systems, and cross-system data association needs to be achieved through a business flow number, such as an order number. Among them, the order management module of the enterprise resource planning system records upstream business events such as order creation, change and completion, reflecting demand initiation and execution status; the logistics management module of the industry chain management system records downstream execution events such as delivery, in-transit, and signed, embodying the physical delivery process. By aligning and associating the same business subject with the two types of events in timestamp sequence, a collaborative event stream covering the demand-delivery full link is constructed, realizing end-to-end traceable modeling and dynamic perception of the upstream and downstream collaborative process of the industry chain.

[0022] In the optimized implementation of the above steps, the collaborative event data set is specifically constructed as follows: order creation events, order change events and order completion events are extracted from the order management interface of the enterprise resource planning management system, and the timestamps of each event are recorded.

[0023] Delivery events, in-transit events and signed events are extracted from the logistics management interface of the industry chain management system, and the timestamps of each event are recorded.

[0024] Based on the business flow number, events from the two systems that are in the same business chain are merged and aligned by timestamp to form a collaborative event sequence identified by the business flow number.

[0025] Each collaborative event is labeled with the occurrence timestamp and the enterprise identifier participating in the event, and a collaborative event sequence is generated.

[0026] In the example of the above operation, the enterprises participating in the event can include suppliers, distributors, logistics service providers, retailers, etc.

[0027] Each collaborative event in the collaborative event sequence generated based on the business flow number alignment described above is a observable behavior record of the industry chain collaborative process, constituting a digital footprint reflecting cross-enterprise objective interaction. This footprint carries business actions, participating subjects and timing information in a structured event form, as a basic semantic unit, supporting the construction of subsequent collaborative association links, timing dependency analysis and quantitative evaluation of collaborative relationship strength, providing a data foundation for computable modeling of industry chain collaboration status.

[0028] Step2: extracting adjacent collaboration event pairs based on the sliding window method on the collaboration event sequence, and screening event pairs with significant causal correlation through temporal causality test to construct collaboration event correlation links with enterprises as nodes and business stages as edges.

[0029] Referring to Figure 2 In the optional implementation of the above steps, the specific construction process of the collaboration event correlation link is as follows: a sliding window with a step of 1 and a window size of 2 is used to traverse the sequence on the collaboration event sequence corresponding to a single business flow number to extract a continuous predecessor-successor event pair to form an adjacent collaboration event pair, which is used to depict the local interaction mode between two continuous business actions.

[0030] Applying the example to the above implementation, the collaboration event sequence is , and the adjacent collaboration event pair formed is , wherein respectively represent collaboration events.

[0031] Based on the combination of collaboration events and their context positions in the business process, the business stage of each adjacent event pair is defined.

[0032] Further applied to the above example, represent the collaboration events of order creation, order confirmation, order delivery, order transfer, order delivery, and order signing, wherein , , , , represent the business stages of demand initiation stage, commitment response stage, performance initiation stage, logistics delivery stage, and end delivery stage.

[0033] It should be noted that the business process in reality has a significant state transition feature, that is, the occurrence of the current collaboration behavior usually depends on the achievement of the previous state. Therefore, the present application extracts adjacent collaboration event pairs on the collaboration event sequence using a sliding window to construct a predecessor-successor correlation unit with a temporal dependence relationship, rather than using isolated collaboration events as the basic unit of analysis. This is because a single collaboration event is only a transient observation point on the time axis, lacks context information, cannot depict the response logic between enterprises, causal dependence, and process evolution dynamics, and is difficult to support accurate quantification of collaboration relationship strength and traceable analysis of propagation path.

[0034] Temporal causality test is performed on each adjacent collaboration event to obtain the causal strength, and the adjacent collaboration event pairs with a causal strength reaching the set threshold are selected according to the set causal strength threshold.

[0035] In a preferred implementation of the above scheme, the temporal causality test is implemented as follows: all structured collaboration event streams related to the target industry chain within a specific time period are extracted from the historical operation logs of the enterprise business system, each record containing a collaboration event, a participating enterprise node, a time stamp, and a business serial number, constituting a high-dimensional event sequence dataset.

[0036] The specific time period mentioned above can be 12 months, aiming to cover a long enough historical business cycle, which helps to improve the data abundance and representativeness of the effective response sample set, enhance the statistical significance and model robustness of the causality strength calculation, avoid evaluation bias caused by insufficient samples or single scenarios, and thus ensure that the collaboration event correlation link constructed has good real-world interpretability and temporal stability.

[0037] For a certain adjacent collaboration event pair, all ordered co-occurrence instances that meet the following conditions are identified in the high-dimensional event sequence dataset: 1) the two events occur in the same event sequence corresponding to the business serial number.

[0038] 2) The time stamp of the precursor event is earlier than that of the successor event.

[0039] 3) The successor event occurs within a reasonable business time window after the precursor event.

[0040] Exemplarily, the reasonable business time window mentioned above can be 72 hours.

[0041] By introducing a reasonable business time window, further logical rationality is ensured to avoid cross-cycle mis-matching.

[0042] The identified ordered co-occurrence instances constitute an effective response sample set of adjacent collaboration event pairs.

[0043] Understandably, the ordered co-occurrence instances identified above reflect the spatiotemporal correlation of the precursor event and the successor event in real business processes, providing a statistically significant response evidence set for the temporal causality test.

[0044] In the effective response sample set of adjacent collaboration event pairs, the total number of times the precursor event occurs in the historical data is counted, and the number of times the successor event is triggered within a set time window after the occurrence of the precursor event is counted, thereby defining the proportion of the number of times the successor event is triggered within the set time window after the occurrence of the precursor event to the total number of times the precursor event occurs in the historical data as the causality strength of the adjacent collaboration event pair.

[0045] The set time window mentioned above is the length of the reasonable business time window.

[0046] It should be noted that the above-mentioned time sequence causality test of adjacent cooperation event pairs is based on the Granger causality thought and the event co-occurrence consistency assumption: the cooperation between enterprises can be regarded as a stimulus-response mechanism, in which the precursor event is regarded as an external stimulus and the successor event is regarded as a system response. If a precursor event continuously and stably leads the occurrence of a successor event in historical data and meets the time proximity and business logic constraints, there is a potential functional dependence or process driving relationship between the two.

[0047] The causality strength defined in this framework is essentially a frequency school estimate of conditional probability, in which the numerator meets the time leading, same business flow, and time window constraint of the effective response sample number, reflecting the implementation degree of actual cooperation behavior, and the denominator is the total occurrence frequency of the precursor event in historical data, as a probability normalized benchmark. The causality strength measure reflects the possibility of the successor event responding as expected after the occurrence of the precursor event. The higher the value, the stronger the process coupling degree between the two links and the higher the cooperation stability.

[0048] For example, if order creation is followed by order confirmation in most cases, it indicates that this response behavior has become a standard operation process.

[0049] It should be noted that the causality strength threshold mentioned above reflects the functional dependence significance level of cooperation behavior between enterprises, and is used to distinguish substantive process driving from accidental event co-occurrence. Its setting should be based on the statistical distribution characteristics of historical data. Generally, the following methods can be used: by analyzing the mean value of the causality strength of various adjacent event pairs in the target industrial chain and the standard deviation , set the threshold value to to retain strong correlation event pairs that are significantly higher than the random level; or set an empirical threshold according to business rules, such as causality strength ≥ 0.7, to ensure that there is an effective causal relationship only when the successor event stably responds to the precursor event in most cases. This threshold mechanism enhances the robustness and business interpretability of the correlation link construction.

[0050] It should be explained that the causality test is performed on the adjacent cooperation event pairs in the cooperation event sequence, and only the event pairs with causality strength higher than the threshold are included in the cooperation event correlation link, rather than simply connecting all adjacent events into a chain. The core role is to filter noise and redundant connections from the perspective of statistical significance and business relevance, and to ensure that the correlation link constructed truly reflects the stable process dependence relationship between enterprises.

[0051] Specifically, fully connected approaches without causal testing often contain numerous accidental co-occurrences, asynchronous responses, or non-driving event pairs, leading to distorted link structures, causal confusion, and reduced analytical reliability. However, by using causal strength filtering, only strongly correlated events with high-frequency, stable, and time-consistent response patterns are retained. This effectively identifies the true stimulus-response paths, improving the accuracy, sparsity, and interpretability of the associated links, providing a high-fidelity topological foundation for subsequent delay propagation analysis, collaboration strength quantification, and risk warning.

[0052] The selected adjacent collaborative event pairs are arranged in chronological order to identify continuous segments with node overlap, i.e. and If they exist simultaneously, if they exist simultaneously Then, they can be spliced ​​into a longer ternary path. In this way, all uninterrupted continuous collaborative path segments can be extracted to ensure the temporal and logical continuity of the path and avoid breakage or jump.

[0053] Based on the identified continuous segments, network nodes are mapped to the enterprises participating in the events according to the time sequence, and edges are formed with the business stages to which adjacent collaborative events belong.

[0054] The collaborative event association links formed above are as follows: Figure 3 As shown.

[0055] The aforementioned collaborative event linkages reflect the temporal dependency structure and functional collaboration paths between upstream and downstream enterprises in the industry chain based on real business interactions. From a data-driven perspective, they characterize the core process framework of cross-enterprise collaboration, revealing the transmission paths of information flow, control flow, and execution flow. This provides a computable topological foundation and a semantic framework for dynamic analysis of subsequent collaborative relationship strength.

[0056] Step 3: Perform response time deviation analysis on the collaborative event association links, identify delayed adjacent collaborative event pairs, and determine the collaborative relationship strength of the collaborative event association links based on the time-series propagation analysis of delayed adjacent collaborative event pairs.

[0057] The above steps are implemented as follows: In the collaborative event association link, for each pair of adjacent collaborative events, the occurrence time of the successor event is calculated from the timestamp of the event, and the occurrence time of the predecessor event is subtracted from the occurrence time of the predecessor event as the actual response time.

[0058] For each pair of adjacent collaborative events, the historical average response time is calculated from the set of valid response samples as the baseline response time.

[0059] The historical average response time is used as the benchmark response time because it reflects the typical cycle of response to subsequent events after a preceding event is triggered under normal operating conditions. Its value is derived from a large number of valid historical response samples, demonstrating good representativeness and stability. Using this as a benchmark allows for the objective identification of abnormal delays that deviate from the normal rhythm. This decomposes fluctuations in actual response time into normal response and abnormal delay components, providing a quantifiable reference system for delay detection and ensuring that anomaly determinations are both business-reasonable and statistically significant.

[0060] The response delay deviation is defined as the positive difference between the actual response time and the reference response time.

[0061] The response delay deviation of each pair of adjacent collaborative events is compared with the preset allowable deviation. If the response delay deviation of a pair of adjacent collaborative events is greater than the allowable deviation, the event pair is marked as a delayed adjacent collaborative event, indicating that the collaborative link has significantly timed out and may have a blocking effect on subsequent processes.

[0062] The allowable deviation mentioned above reflects the business system's tolerance threshold for response latency, which can be set according to industry standards and contract service provisions.

[0063] The spatial distribution characteristics of all identified delayed adjacent collaborative events are statistically analyzed in the collaborative event association chain, and the propagation duration is calculated accordingly.

[0064] As a further implementation of the above scheme, the specific calculation process of propagation persistence is as follows: in the collaborative event association link, all adjacent collaborative event pairs are numbered according to the order of their occurrence time.

[0065] Identify all event pairs marked as delayed adjacent collaborative events from the associated links, extract their corresponding sequence numbers to form a delayed event index set, and sort the numbers in the set in ascending order.

[0066] The delay span is defined as the length of the maximum position interval covered by the delayed event in the process sequence.

[0067] As an example of the definition of the above delay span, all adjacent collaborative event pairs in a collaborative event association link are numbered in chronological order as shown in Table 1.

[0068] Table 1: Numbers of adjacent collaborative event pairs in the collaborative event association chain

[0069]

[0070] Event pairs with location numbers 2, 3, 4, 7, and 8 after delayed identification are marked as delayed adjacent cooperative events.

[0071] The set of delayed event indices is: {2, 3, 4, 7, 8}.

[0072] Delay span = Maximum number - Minimum number = 8 - 2 = 6.

[0073] This value indicates that the delayed behavior covers the entire interval from the 2nd to the 8th event pair, which is a large span, although no delay occurred at positions 5 and 6, indicating that the delay distribution is relatively dispersed.

[0074] Therefore, the delay span reflects the breadth of the time distribution of delay behavior throughout the entire process chain. If the delay is concentrated in a local interval, the span is small; if it is dispersed, the span is large.

[0075] The ratio of the number of delayed adjacent collaborative events to the delay span is used as the propagation duration.

[0076] Understandably, the number of delayed adjacent collaborative events represents the total number of event pairs with significant response delays, reflecting the breadth of the delay phenomenon; the delay span is the difference between the position numbers of the maximum and minimum delayed events, characterizing the length of the distribution interval of the delay in the process sequence. When comparing the number of delayed adjacent collaborative events with the delay span, if the delayed adjacent collaborative events are concentrated, for example, if the set of delayed event indices is {2, 3, 4}, then the delay span is 2 and the number of delayed adjacent collaborative events is 3. The propagation duration calculated from this is... When delayed adjacent collaborative events are scattered, such as when the delayed event index set is {2, 3, 4, 7, 8}, the delay span is 7, and the number of delayed adjacent collaborative events is 5. Therefore, the propagation duration is calculated as follows: .

[0077] Therefore, propagation persistence is essentially a measure of the temporal density of delayed events. A higher value indicates a high degree of clustering of delays within local process intervals, exhibiting significant temporal continuity and spatial clustering. This reflects a strong chain-like propagation tendency and cumulative effect of delayed behavior in the collaborative link, meaning that preceding delays easily trigger subsequent timeouts, revealing structural bottlenecks or collaboratively vulnerable sections in the system. Thus, a higher propagation persistence indicates a greater intensity of delay propagation and a greater risk of process blockage.

[0078] It should be noted that when only a single delayed adjacent collaborative event exists, the delay span is zero, indicating that the delayed behavior occurs in isolation, lacking the support of subsequent related delayed events, and cannot form a temporal propagation path or chain effect. In this case, the delayed behavior does not possess propagation attributes, only reflecting a local instantaneous anomaly, rather than a systemic collaborative blockage. Therefore, this invention stipulates that when the number of identified delayed adjacent collaborative events is 1, it is considered that there is no effective delay propagation, and no propagation duration is calculated, in order to avoid misjudging non-propagational anomalies and ensure the accuracy and semantic rationality of subsequent collaborative relationship strength analysis.

[0079] In the collaborative event association link, the collaborative relationship strength value of the collaborative event association link is obtained by combining the proportion of delayed adjacent collaborative events and the propagation duration of delayed adjacent collaborative events through negative correlation weighting.

[0080] It's important to understand that in a business context, the strength of collaborative relationships is a positive indicator. It directly reflects the health, stability, and efficiency of the collaborative relationship, while delays and delayed propagation are negative phenomena. Specifically, a high proportion of delays means that many links in the collaborative chain have experienced abnormal delays. This directly indicates poor smoothness and slow response throughout the entire chain, thus the overall strength of the collaborative relationship is undoubtedly low. The greater the duration of propagation, the more densely delayed events are in a time series, rather than being sporadic. This suggests that the problem is not accidental but persistent and systemic, possibly stemming from a serious failure in a certain link. This further implies that the collaborative relationship is in a state of sustained high risk and low strength.

[0081] In specific collaborative relationship strength analysis, the proportion of delayed adjacent collaborative events is used as a breadth feature to characterize the coverage of abnormal delays in the collaborative link; propagation persistence is used as a depth feature to reflect the temporal clustering and persistence of delayed behavior. By introducing a negative linear weighted model to fuse the two, an exemplary expression for collaborative relationship strength is given by: In the formula Indicates the strength of the cooperative relationship. , These represent the percentage of delayed adjacent collaborative events and the propagation duration, respectively. , Denotes the corresponding weight factor, and satisfies , , .

[0082] When applying the above expression for the strength of collaborative relationships, since the proportion of delayed adjacent collaborative events ranges from [0, 1], and the propagation persistence may be greater than 1 when delayed events are densely distributed, if it is directly involved in the weighted calculation, it will significantly amplify the abnormal influence in terms of scale, destroy the numerical balance of the model, and even cause the strength of collaborative relationships to be negative or exceed the reasonable semantic range, affecting the interpretability and stability of the evaluation results.

[0083] Therefore, the propagation duration needs to be normalized and mapped to the [0, 1] interval. Normalization can be achieved using methods such as maximum value normalization or Sigmoid compression.

[0084] The weighting factors for the proportion of delayed adjacent collaborative events and the duration of propagation mentioned above can be set using historical data. Specifically, key abnormal events that occurred in the collaborative links in history, such as delivery interruptions, process blockages, or serious deviations from service level agreements, can be collected. The frequency of events caused by isolated or widespread delayed links (i.e., a high proportion) and the frequency of events caused by continuous delayed propagation (i.e., high propagation duration) can be statistically analyzed. The proportion of each type of cause is calculated as a weighting factor, reflecting the actual contribution of different types of delay patterns to system-level failures, thus giving the collaborative relationship strength model a business empirical basis and causal interpretability.

[0085] This invention, when analyzing the strength of collaborative relationships based on the constructed collaborative event correlation links, uses collaboration delay as the core analytical dimension, viewing delay as a key abnormal signal and a manifestation of efficiency degradation in the upstream and downstream collaboration process of the industry chain. Collaboration delay is not only a direct manifestation of lagging business response, but also reflects the degree of mismatch in information transmission, resource scheduling, and process connection between enterprises, and is an important negative indicator for measuring the health of collaboration. Using this as a starting point, collaborative bottlenecks can be accurately identified, risk transmission paths can be revealed, and dynamic, interpretable, and fine-grained assessments of the strength of collaborative relationships can be achieved, significantly improving the accuracy of industry chain operational status perception and the targeted nature of early warning interventions.

[0086] Step 4: Locate the consecutively occurring delayed adjacent collaborative event segments in the collaborative event association link as the delayed propagation segment.

[0087] The above steps are as follows: Identify all node pairs marked as delayed adjacent collaborative events in the collaborative event association link, and record their position index in the link.

[0088] Based on the location index of delayed adjacent cooperative events, the delayed events of neighboring events are clustered and merged according to their temporal adjacency relationship in the link to form several connected sub-segments, each of which serves as a delayed propagation segment.

[0089] As an example of the above operation, in the collaborative event association chain shown in Table 1, when the delayed event index set is {2, 3, 4, 7, 8}, cluster analysis based on the positional continuity of event pairs can identify two temporally connected delayed segments. The resulting delayed propagation segments are b→c→d→e and g→i→j, respectively.

[0090] The above describes a method for locating delay propagation segments in the collaborative event linkage to accurately identify structural bottlenecks caused by response delays. This method aims to aggregate isolated delay events temporally and logically, identifying delay clusters with continuity and propagation, thereby revealing key blocking paths in the collaborative linkage.

[0091] Step 5: Identify the risk characteristics of the delayed propagation segment and generate supply chain early warning information based on the risk characteristics of the delayed propagation segment.

[0092] In the above-mentioned scheme, the risk characteristics of the delayed propagation segment are identified as follows: the number of adjacent cooperative event pairs covered by each delayed propagation segment is counted to obtain the delayed propagation length.

[0093] Calculate the average response delay deviation for all delayed events in each delayed propagation segment.

[0094] The average of the delay propagation length and the response delay deviation is used as the risk characteristic of each delay propagation segment.

[0095] It's important to understand that the length of delay propagation reflects the scope and breadth of the anomaly's impact within the collaborative chain. A longer length indicates a chain-like transmission of delays between upstream and downstream enterprises, a wider impact of collaborative blockages, weaker system resilience, and foreshadows potential structural bottlenecks or critical path failures. The average value of the response delay deviation reflects the severity of delays at each stage within that segment. A higher average value indicates widespread lag in response at each node and a significant decrease in collaborative efficiency. Combining these two aspects provides a two-dimensional characterization of delay propagation risk from both spatial and temporal dimensions, offering a quantitative basis for subsequent tiered early warning and precise intervention.

[0096] In a further feasible approach to the above scheme, the generation of supply chain early warning information based on the risk characteristics of the delayed propagation segment is described below: the risk characteristic identifier of the delayed propagation segment is matched with the preset early warning rules to obtain the early warning level. The specific early warning rules are as follows: when the propagation length exceeds the length threshold and the average value of the response delay deviation exceeds the deviation threshold, a high-risk early warning level is generated.

[0097] A medium-risk warning level is generated when the propagation length exceeds the length threshold or the average response delay deviation exceeds the deviation threshold.

[0098] The aforementioned early warning system uses a two-dimensional coupling approach to differentiate the scope and severity of risk impact, enabling a tiered response to anomalies in supply chain collaboration. Specifically, a high-risk warning indicates that the system is experiencing a large-scale and deeply delayed collaborative failure, posing a risk of cascading disruptions and requiring immediate intervention. A medium-risk warning is considered a potential threat, possibly a local bottleneck or an early sign of spread, requiring monitoring and assessment of its evolutionary trend.

[0099] The aforementioned medium-length threshold can be set to 3 to 5 consecutive delay events. The deviation threshold is set according to the tolerance of the delivery cycle agreed in the contract. For example, if a delay of ±24 hours is allowed, the deviation threshold is set to 24 hours.

[0100] Step 5 also includes sending risk notices to upstream and downstream enterprises of the delayed propagation segment. The specific implementation is as follows: extract the participating enterprises involved in the delayed propagation segment from the collaborative event association link.

[0101] Based on the warning level, a risk notice of the corresponding level is generated and sent to relevant enterprises through the enterprise communication interface.

[0102] The aforementioned operations, by accurately identifying upstream and downstream participating enterprises involved in the delayed propagation phase and proactively pushing risk notifications based on the warning level, have achieved a shift from passive response to proactive collaborative intervention. Its core value lies in breaking down information silos, promoting cross-enterprise visibility and shared responsibility for abnormal events, enhancing the overall emergency response speed and collaborative recovery capabilities of the industrial chain, and preventing further spread of risks due to information lag.

[0103] For example, risk notices typically include the warning level, a list of affected companies, the extent of the delayed spread, risk characteristics, and recommended countermeasures, such as expediting dispatch, activating backup suppliers, and adjusting logistics routes.

[0104] The parameters involved in the above formula are all dimensionless and calculated numerically. The formula is a formula obtained from the most recent real situation by collecting a large amount of data and simulating it with software. The preset parameters in the formula are set by those skilled in the art according to the actual situation.

[0105] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0106] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0107] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0108] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0109] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for analyzing the strength of upstream and downstream collaborative relationships in an industrial chain, characterized in that, Includes the following steps: Step 1: Obtain business interaction data in real time from the enterprise resource planning system and supply chain management system, extract the collaboration events and event timestamps, and construct a collaboration event sequence that includes participating enterprises and collaboration events based on business serial numbers; Step 2: Based on the collaborative event sequence, the sliding window method is used to extract adjacent collaborative event pairs, and event pairs with significant causal correlation are screened through temporal causality test to construct a collaborative event association link with enterprises as nodes and business stages as edges; Step 3: Perform response time deviation analysis on the collaborative event association links, identify delayed adjacent collaborative event pairs, and determine the collaborative relationship strength of the collaborative event association links based on the time-series propagation analysis of delayed adjacent collaborative event pairs. Step 4: Locate consecutively occurring delayed adjacent collaborative event segments in the collaborative event association chain as delayed propagation segments; Step 5: Identify the risk characteristics of the delayed propagation segment and generate supply chain early warning information based on the risk characteristics of the delayed propagation segment.

2. The method for analyzing the strength of upstream and downstream collaborative relationships in an industrial chain as described in claim 1, characterized in that: The collaborative event sequence is constructed through the following process: Extract order creation events, order modification events, and order completion events from the order management interface of the enterprise resource planning management system, and record the timestamps of each event. Extract shipment events, in-transit events, and receipt events from the logistics management interface of the supply chain management system, and record the timestamp of each event. Based on the business serial number, events from two systems that are on the same business chain are merged and aligned according to the timestamp to form a collaborative event sequence identified by the business serial number; Each collaborative event is labeled with its timestamp and the identifiers of the participating companies, generating a collaborative event sequence.

3. The method for analyzing the strength of upstream and downstream collaborative relationships in an industrial chain as described in claim 1, characterized in that: The specific construction process of the collaborative event association link is as follows: On the collaborative event sequence corresponding to a single business serial number, a sliding window with a step size of 1 and a window size of 2 is used to extract consecutively occurring predecessor-successor events to form an adjacent collaborative event pair. Based on the combination of collaborative events and their contextual position in the business process, a business phase is defined for each pair of adjacent collaborative events. For each pair of adjacent collaborative events, a temporal causality test is performed to obtain the causal strength, and adjacent collaborative event pairs that reach the set causal strength threshold are selected according to the set causal strength threshold. Arrange the selected adjacent collaborative event pairs in chronological order to identify continuous segments with node overlap; Based on the identified continuous segments, network nodes are mapped to the enterprises participating in the events according to the time sequence, and edges are formed with the business stages to which adjacent collaborative events belong.

4. The method for analyzing the strength of upstream and downstream collaborative relationships in an industrial chain as described in claim 3, characterized in that: The process of performing a temporal causality test on each pair of adjacent cooperative events to obtain the causal strength is as follows: Extract all structured collaborative event streams related to the target industry chain within a specific time period from the historical operation logs of the enterprise business system. Each record contains the collaborative event, the participating enterprise, the timestamp of the event, and the business serial number, forming a high-dimensional event sequence dataset. For a given pair of adjacent cooperative events, identify all ordered co-occurrence instances in a high-dimensional event sequence dataset that satisfy the following conditions: 1) Both events occur within the same event sequence corresponding to the same business serial number; 2) The timestamp of the preceding event is earlier than that of the subsequent event; 3) The subsequent event occurs within a reasonable business time window following the preceding event; The identified ordered co-occurrence instances are used to form an effective response sample set for adjacent cooperative event pairs; In the effective response sample set of the adjacent cooperative event pairs, the total number of occurrences of the predecessor event in the historical data is counted, and the number of times the successor event is triggered within a set time window after the occurrence of the predecessor event is counted. Thus, the proportion of the number of times the successor event is triggered within the set time window after the occurrence of the predecessor event to the total number of occurrences of the predecessor event in the historical data is defined as the causal strength of the adjacent cooperative event pair.

5. The method for analyzing the strength of upstream and downstream collaborative relationships in an industrial chain as described in claim 4, characterized in that: The implementation steps for Step 3 are as follows: In the collaborative event association chain, for each pair of adjacent collaborative events, the occurrence time of the successor event is calculated from the timestamp of the event, and the occurrence time of the predecessor event is subtracted from the occurrence time of the predecessor event to obtain the actual response time; For each pair of adjacent collaborative events, the historical average response time is calculated from the set of valid response samples as the baseline response time; The response delay deviation is defined as the positive difference between the actual response time and the reference response time. The response delay deviation of each pair of adjacent collaborative events is compared with a preset allowable deviation. If the response delay deviation of a pair of adjacent collaborative events is greater than the allowable deviation, the event pair is marked as a delayed adjacent collaborative event. In the collaborative event association chain, the spatial distribution characteristics of all identified delayed adjacent collaborative events are statistically analyzed, and the propagation duration is calculated accordingly. In the collaborative event association link, the collaborative relationship strength value of the collaborative event association link is obtained by combining the proportion of delayed adjacent collaborative events and the propagation duration of delayed adjacent collaborative events through negative correlation weighting.

6. The method for analyzing the strength of upstream and downstream collaborative relationships in an industrial chain as described in claim 5, characterized in that: The specific calculation process for the propagation duration is as follows: In the collaborative event association chain, all adjacent collaborative event pairs are numbered according to the chronological order of their occurrence. Identify all event pairs marked as delayed adjacent collaborative events from the associated links, extract their corresponding sequence numbers to form a delayed event index set, and sort the numbers in the set in ascending order; The delay span is defined as the length of the maximum position interval covered by the delayed event in the process sequence; The ratio of the number of delayed adjacent collaborative events to the delay span is used as the propagation duration.

7. The method for analyzing the strength of upstream and downstream collaborative relationships in an industrial chain as described in claim 1, characterized in that: The specific content of Step 4 is as follows: Identify all node pairs marked as delayed adjacent collaborative events in the collaborative event association link and record their position index in the link; Based on the location index of delayed adjacent cooperative events, the delayed events of neighboring events are clustered and merged according to their temporal adjacency relationship in the link to form several connected sub-segments, each of which serves as a delayed propagation segment.

8. The method for analyzing the strength of upstream and downstream collaborative relationships in an industrial chain as described in claim 1, characterized in that: The risk feature identification of the delayed propagation segment is performed as follows: The propagation length is obtained by counting the number of adjacent cooperative event pairs covered by each delayed propagation segment. Calculate the average response delay deviation of all delayed adjacent cooperative events in each delayed propagation segment; The average value of the delay propagation length and the response delay deviation is used as the risk characteristics of each delay propagation segment.

9. The method for analyzing the strength of upstream and downstream collaborative relationships in an industrial chain as described in claim 1, characterized in that: The generation of supply chain early warning information based on the risk characteristics of the delayed propagation segment is described below: The risk characteristics of the delayed propagation segment are matched with preset warning rules to obtain the warning level. The specific warning rules are as follows: A high-risk warning level is generated when the propagation length exceeds the length threshold and the average response delay deviation exceeds the deviation threshold. A medium-risk warning level is generated when the propagation length exceeds the length threshold or the average response delay deviation exceeds the deviation threshold.

10. The method for analyzing the strength of upstream and downstream collaborative relationships in an industrial chain as described in claim 1, characterized in that: Step 5 also includes sending risk notifications to upstream and downstream enterprises in the delayed propagation segment, and the specific implementation is as follows: Extract the participating companies involved in the delayed propagation segment from the collaborative event's correlation chain; Based on the warning level, a risk notice of the corresponding level is generated and sent to relevant enterprises through the enterprise communication interface.

Images