Digital auditing method and system for supply chain

By collecting real-time transaction and logistics information from supply chain nodes and using sliding windows and anomaly detection algorithms to dynamically adjust the audit window, the problem of the inability to monitor abnormal signals in real time in existing technologies has been solved, enabling real-time risk warning and efficient operation of the supply chain.

CN121073222AActive Publication Date: 2025-12-05STATE GRID SICHUAN ELECTRIC POWER CO
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202511586927.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-11-03
Publication Date
2025-12-05
Estimated Expiration
2045-11-03

AI Technical Summary

Technical Problem

Existing supply chain auditing methods cannot monitor transaction and logistics information in real time, resulting in the neglect of time-sensitive abnormal signals and the inability to intervene in a timely manner. Furthermore, the lack of dynamic adjustment mechanisms leads to audit delays.

Method used

By collecting real-time transaction records and logistics information from each node of the supply chain, the time series data is processed using a sliding window algorithm, combined with anomaly detection and clustering algorithms, and the audit window is dynamically adjusted to capture abnormal signals and generate real-time monitoring results.

Benefits of technology

It enables precise capture of time-sensitive anomaly signals, preventing early signs from being masked by historical data, improving the accuracy of supply chain risk warnings, reducing operational losses, and enhancing the resilience of supply chain operations.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121073222A_ABST
    Figure CN121073222A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of information. The invention provides a supply chain digital auditing method and a supply chain digital auditing system in order to solve the problem that time-sensitive abnormal signals are easy to ignore when massive dynamic data sets are processed in the prior art. The method comprises the following steps: collecting real-time transaction and logistics information of each node of a supply chain to obtain a timestamp-containing dynamic data set, and determining initial distribution of abnormal signals; processing the time sequence data by using a sliding window algorithm to obtain a latest transaction and historical data fusion result; if the delivery delay exceeds the threshold value, judging a potential risk, and obtaining a quality fluctuation correlation index; analyzing logistics anomalies by using an anomaly detection algorithm to determine an evidence chain fracture position; automatically sliding an auditing window according to the fracture position to obtain a timeliness analysis framework; if the framework shows risk deviation, obtaining an optimized early warning sequence by using a clustering algorithm; a complete evidence chain path is determined by fusing the newest transaction and logistics abnormity; and processing the legacy data according to the path to obtain a real-time audit result. According to the method, time-sensitive abnormal signal preliminary distribution is accurately captured and is prevented from being masked by historical data.
Need to check novelty before this filing date? Find Prior Art

Description

TECHNICAL FIELD

[0001] The present application relates to the field of information technology, in particular to a digital auditing method and system for a supply chain. BACKGROUND

[0002] As a key field to ensure the safety and compliance of enterprise operations, supply chain digital auditing is important in that it can monitor the entire chain process from raw material procurement to product delivery in real time, avoiding potential risks that may cause significant losses to the business. In a complex and changing business environment, this field directly affects the economic benefits and market competitiveness of enterprises. Current supply chain auditing methods often rely on fixed-cycle batch checks, which cannot adapt to the real-time influx of transaction records and logistics information at each node of the supply chain, resulting in an auditing process that lags behind the actual operational rhythm.

[0003] Traditional methods tend to overlook time-sensitive abnormal signals when dealing with massive dynamic data sets, such as early signs of supplier delivery delays being submerged in historical data accumulation, thus missing the opportunity for intervention. This limitation further magnifies the core technical difficulty, namely the lack of a dynamic adjustment mechanism for the auditing window, making it impossible to effectively integrate historical data with the latest transaction and logistics information. SUMMARY

[0004] The present application provides a digital auditing method and system for a supply chain to address the problem of overlooking time-sensitive abnormal signals when dealing with massive dynamic data sets in the prior art.

[0005] The technical solution adopted by the present application is as follows: A digital auditing method for a supply chain, comprising the following steps: Step S100, acquiring real-time transaction records and logistics information at each node of the supply chain to obtain a dynamic data set containing timestamps, and determining the preliminary distribution of abnormal signals therein; Step S200, processing time series data using a sliding window algorithm based on the obtained dynamic data set to obtain a fusion result of the latest transaction information and historical data within the window; Step S300, if the delivery delay signs in the fusion result exceed a preset threshold, determining it as a potential risk event and obtaining corresponding quality fluctuation correlation indicators; if the delivery delay signs in the fusion result do not exceed the preset threshold, staying at this step; Step S400, using the obtained quality fluctuation correlation indicators to analyze the pattern of logistics abnormality capture using an anomaly detection algorithm to determine the occurrence position of evidence chain rupture; Step S500, adjusting the mechanism to automatically slide the auditing window based on the determined occurrence position of evidence chain rupture to obtain an updated time-sensitive data dominant analysis framework; Step S600, if the delivery delay indication in the fusion result exceeds the preset threshold, a potential risk event is determined, and the corresponding quality fluctuation correlation index is obtained; if the delivery delay indication in the fusion result does not exceed the preset threshold, the step is stayed; if the update framework does not show risk prediction deviation, the step is stayed; Step S700, by the obtained optimized prediction warning sequence, the latest transaction information and the logistics abnormality capture are fused to determine the complete path of the continuity evidence chain; Step S800, according to the complete path of the continuity evidence chain determined, the remaining data of the fixed period inspection is processed to obtain the audit output result of real-time monitoring.

[0006] Further, in step S100, by collecting real-time transaction records and logistics information of each node of the supply chain, a dynamic data set containing a time stamp is obtained, and a preliminary distribution of abnormal signals is determined, including the following sub-steps: The real-time transaction records and logistics information of each node of the supply chain are collected through the supply chain node interface to generate a dynamic data set containing a time stamp; The time sequence analysis is adopted to extract the time sequence characteristics of the transaction records and logistics information from the dynamic data set to obtain abnormal signals; If the time sequence characteristics of the abnormal signals exceed the preset threshold, the abnormal signals are classified by a clustering algorithm to determine the abnormal categories; According to the abnormal categories, a decision tree algorithm is used to analyze the relevance of the transaction records and logistics information to obtain the distribution mode of the abnormal signals; Through the distribution mode, the frequency and intensity of the abnormal signals in the dynamic data set are calculated to determine the preliminary distribution; If the frequency or intensity of the preliminary distribution exceeds the preset threshold, the future change trend of the abnormal signals is obtained through a time sequence prediction algorithm; According to the change trend, the data collection frequency is adjusted, the monitoring parameters of the supply chain nodes are optimized, and finally the preliminary distribution of the abnormal signals is determined.

[0007] Further, in step S200, according to the obtained dynamic data set, a sliding window algorithm is used to process time sequence data to obtain the fusion result of the latest transaction information and historical data in the window, including the following sub-steps: For the obtained dynamic data set, a sliding window algorithm is used to traverse the local mode of the time sequence data contained therein, and the statistical indicators of the transaction information in the window are extracted to obtain the statistical characteristics of the local mode; If the statistical characteristics of the local mode deviate from the preset threshold, a principal component analysis algorithm is used to compress the feature dimension to obtain a reduced feature set; According to the feature set after dimension reduction, a k-means clustering algorithm is used to group the local patterns, and a clustering result of the transaction information is obtained; Through the clustering result, the timestamp interval distribution of each group of transaction information is calculated to determine the periodic pattern of the time series; If the timestamp interval of the periodic pattern exceeds a preset threshold, a moving average algorithm is used to smooth the time series data to obtain a smoothed sequence trend; According to the smoothed sequence trend, the size and step length of the sliding window are adjusted to obtain an optimized time series local pattern; Through the optimized time series local pattern, dynamic features of transaction information and historical data fusion are extracted, and finally the fusion result of the latest transaction information and historical data in the window is obtained.

[0008] Further, in step S300, if the delivery delay indication in the fusion result exceeds a preset threshold, a potential risk event is determined, and the corresponding quality fluctuation correlation index is obtained; if the delivery delay indication in the fusion result does not exceed the preset threshold, the step is stayed, and the following substeps are specifically included: For the obtained fusion result, if the delivery delay indication therein exceeds a preset threshold, the distribution characteristics of the abnormal points are extracted through time series analysis to determine a potential risk event; According to the distribution characteristics of the abnormal points, a principal component analysis algorithm is used to compress the quality fluctuation data to obtain a feature set after dimension reduction; Through the feature set after dimension reduction, the correlation strength between supply chain nodes is calculated to obtain a dynamic correlation matrix between nodes; If the correlation strength in the dynamic correlation matrix is lower than a preset threshold, a moving average algorithm is used to smooth the time series data to obtain a smoothed trend sequence; According to the smoothed trend sequence, the fluctuation frequency of the periodic pattern is extracted to determine the abnormal distribution pattern of the supply chain node; Through the abnormal distribution pattern, the transaction information deviation value of each node is calculated to obtain a quantitative index of potential risk; According to the quantitative index of potential risk, the window parameters of time series analysis are adjusted to obtain an optimized risk prediction model, and finally the corresponding quality fluctuation correlation index is obtained.

[0009] Further, in step S400, through the obtained index, an anomaly detection algorithm is used to analyze the mode of logistics anomaly capture to determine the occurrence position of the evidence chain break, and the following substeps are specifically included: For the obtained quality fluctuation correlation index, an anomaly detection algorithm is used to process the associated logistics data, an anomaly pattern is extracted through an isolation forest algorithm, and an abnormal distribution characteristic is obtained; According to the abnormal distribution characteristic, the transaction information deviation between supply chain nodes is calculated to obtain a node abnormal value; If the node anomaly value exceeds the preset threshold value, the time trend of the node transaction is extracted by the time series decomposition method to determine the abnormal fluctuation mode; According to the abnormal fluctuation mode, a clustering algorithm is used to group the supply chain nodes to obtain a node anomaly cluster; Through the node anomaly cluster, the statistical deviation of the transaction information in each cluster is calculated to determine the evidence chain fracture position; According to the evidence chain fracture position, the sensitivity parameter of the anomaly detection algorithm is adjusted to obtain an optimized anomaly detection model; Through the optimized anomaly detection model, the logistics data is reprocessed to finally determine the occurrence position of the evidence chain fracture.

[0010] Further, in step S500, according to the determined position, the mechanism adjusts the audit window to automatically slide to obtain an updated time-sensitive data dominant analysis framework, which specifically includes the following sub-steps: According to the determined evidence chain fracture position, abnormal node data is obtained therefrom, and a time series analysis is used to extract the fluctuation period of the node transaction to determine the periodic abnormal mode; According to the periodic abnormal mode, a clustering algorithm is used to group the supply chain nodes to obtain an abnormal node cluster; The transaction information deviation is obtained from the abnormal node cluster, and a statistical analysis is used to calculate the deviation distribution to determine the deviation concentration area; If the deviation concentration area exceeds the preset threshold value, the time sequence characteristics of the abnormal node are extracted by the time series decomposition method to determine the abnormal duration; According to the abnormal duration, the time span of the sliding window is adjusted to obtain an updated audit time period; From the updated audit time period, transaction data is obtained, and an anomaly detection algorithm is used to process the data to determine the optimized evidence chain fracture position; According to the optimized evidence chain fracture position, a time-sensitive dominant analysis logic is generated to finally obtain an updated time-sensitive data dominant analysis framework.

[0011] Further, in step S600, if the delivery delay indication in the fusion result exceeds the preset threshold value, it is determined as a potential risk event, and the corresponding quality fluctuation correlation index is obtained; if the delivery delay indication in the fusion result does not exceed the preset threshold value, it stays in this step; if the updated framework does not show risk prediction deviation, it stays in this step, which specifically includes the following sub-steps: According to the risk prediction deviation shown by the updated time-sensitive data dominant analysis framework, the abnormal signals are grouped by a clustering algorithm according to the abnormal signal grouping basis to obtain a signal cluster set; The signal intensity distribution is obtained from the signal cluster set, and the distribution characteristics are calculated using statistical analysis methods to determine the concentrated distribution area. If the concentrated distribution area exceeds the preset threshold, the dynamic change trend of the signal cluster is extracted through time series analysis to determine the change trend; Based on the changing trend, the input parameters of the prediction model are adjusted to obtain the optimized set of prediction parameters; From the optimized set of prediction parameters, regression analysis is used to generate prediction and early warning sequences, thus obtaining early warning sequence data; Based on the early warning sequence data, the risk prediction framework is adjusted using the deviation correction method to obtain the updated risk prediction model; From the updated risk prediction model, the priority ranking of abnormal signals is extracted, and finally the optimized prediction and early warning sequence is obtained.

[0012] Furthermore, in step S700, the obtained sequence is used to integrate the latest transaction information with the captured logistics anomalies to determine the complete path of the continuous evidence chain, specifically including the following sub-steps: For the obtained optimized prediction and early warning sequence, transaction information sequence and logistics anomaly signal are extracted from it, and multi-source data are integrated using data fusion rules to obtain a fused data set. From the fused dataset, transaction timestamps and logistics status updates are extracted, and time series analysis is used to determine time-series correlation features to obtain a time-series correlation set. If the time-series correlation set meets the anomaly triggering conditions, the length of the evidence chain is analyzed through path consistency verification to determine the continuous evidence path; Based on the continuous evidence path, the association rule mining algorithm is used to extract the transaction association identifier and logistics node sequence to obtain the association path set; From the set of associated paths, cluster analysis is used to group the path consistency verifications to obtain optimized path clusters; By optimizing the path cluster and using a preset threshold to determine the validity of the complete path structure, the final complete path is determined. By extracting the matching relationship between transaction association identifiers and logistics node sequences from the final complete path, the complete path of the continuous evidence chain is finally determined.

[0013] Furthermore, in step S800, based on the complete path, the legacy data from the fixed-period checks is processed to obtain the audit output results for real-time monitoring, specifically including the following sub-steps: For the complete path of the established continuous chain of evidence, the corresponding fixed-period inspection legacy data is extracted, and the effective transaction records and logistics status updates are extracted using data cleaning rules to obtain a standardized data set; If the standardized data set meets the preset integrity threshold, the transaction timestamp and the logistics node sequence are extracted through time series analysis to obtain a time series correlation set; According to the time series correlation set, a correlation rule mining algorithm is used to extract the matching relationship between the transaction and the logistics node to obtain an associated path set; If the path consistency in the associated path set meets the preset threshold, the paths are grouped through clustering analysis to obtain an optimized path cluster; From the optimized path cluster, a preset threshold is used to judge the effectiveness of the path structure to obtain an effective path set; According to the effective path set, the continuity feature of the transaction timestamp and the logistics node sequence is extracted to obtain the audit output basic data for real-time monitoring; From the audit output basic data, a data mapping method is used to generate a structured transaction and logistics matching relationship, and finally the audit output result for real-time monitoring is obtained.

[0014] Based on the same inventive concept, the application also discloses a digital audit system for a supply chain, which executes the aforementioned digital audit method for a supply chain, and includes the following modules: The data acquisition and abnormality preliminary distribution module inputs real-time transaction records and logistics information of each node of the supply chain, acquires a dynamic data set containing a timestamp through information acquisition, determines the preliminary distribution of abnormal signals, and outputs the dynamic data set and the preliminary distribution of abnormal signals. The sliding window data fusion module inputs the dynamic data set, processes time series data using a sliding window algorithm to obtain a fusion result of the latest transaction information and historical data in the window, and outputs the fusion result. The risk judgment and quality index module inputs the fusion result, determines a potential risk event if the delivery delay indication exceeds a preset threshold, acquires a quality fluctuation correlation index, and outputs the potential risk event and the quality fluctuation correlation index. The abnormality detection and fracture positioning module inputs the quality fluctuation correlation index, analyzes the logistics abnormality mode using an abnormality detection algorithm, determines the fracture position of the evidence chain, and outputs the fracture position of the evidence chain. The audit window adjustment module inputs the fracture position, automatically slides the audit window through an adjustment mechanism to obtain a time-effectiveness data dominant analysis framework, and outputs the analysis framework. The early warning sequence optimization module inputs the analysis framework, groups an abnormal signal ignored cluster using a clustering algorithm if a risk prediction deviation is displayed, obtains an optimized prediction early warning sequence, and outputs the optimized early warning sequence. The complete evidence chain determination module inputs the optimized early warning sequence, fuses the latest transaction information and logistics abnormality capture, determines a continuous evidence chain complete path, and outputs the complete path. The audit result generation module inputs a complete path, processes fixed period check residual data, obtains real-time monitoring audit output results, and outputs the audit output results.

[0015] Each of the above modules can be realized by computer software.

[0016] The beneficial effects of the present application are: the supply chain risk audit method and system based on dynamic data set disclosed in the present application can accurately capture the preliminary distribution of time-sensitive abnormal signals by collecting real-time transaction records and logistics information of each node in the supply chain to obtain dynamic data set, avoid that early signs such as supplier delivery delay are covered by historical data, and provide a data basis for timely intervention; the sliding window algorithm is used to process time series data, effectively fusing the latest transaction information and historical data, and breaking the limitation that historical and latest information cannot be effectively fused; the abnormal detection algorithm is used to locate the evidence chain breaking position, and the audit window is automatically slid based on the position to make up for the short board of the missing dynamic adjustment mechanism of the audit window; subsequently, the clustering optimization early warning sequence is combined to determine the complete evidence chain and process residual data, and finally the real-time audit results are output, which guarantees the timeliness and continuity of the audit throughout the whole process, greatly improves the accuracy of supply chain risk early warning, reduces the operation loss caused by missing intervention opportunity, and enhances the overall operation resilience of the supply chain. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creating any inventive labor.

[0018] Figure 1 The flow chart of the supply chain risk audit method based on dynamic data set is shown in the figure. Figure 2 The block diagram of the supply chain risk audit system based on dynamic data set is shown in the figure. DETAILED DESCRIPTION

[0019] In the description of the present application, it should be understood that the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application.

[0020] The following disclosure provides many different embodiments or examples for implementing different structures of the application. For simplicity of the disclosure, the following paragraphs generally describe the components and settings of specific examples. Of course, they are merely examples and are not intended to limit the application.

[0021] Embodiments of the application will be described in detail with reference to the drawings.

[0022] As shown in the accompanying Figure 1 The digital auditing method for supply chain disclosed in the embodiment specifically includes the following steps: Step S100: Collecting real-time transaction records and logistics information of each node in the supply chain, and determining the preliminary distribution of abnormal signals Through the supply chain node interface, real-time transaction records and logistics information of each node in the supply chain are collected to generate a dynamic data set containing time stamps. Time series analysis is used to extract the time sequence characteristics of transaction records and logistics information from the dynamic data set to obtain abnormal signals. If the time sequence characteristics of the abnormal signals exceed the preset threshold, the abnormal signals are classified by a clustering algorithm to determine the abnormal categories. According to the abnormal categories, a decision tree algorithm is used to analyze the relevance of transaction records and logistics information to obtain the distribution pattern of abnormal signals. Through the distribution pattern, the frequency and intensity of abnormal signals in the dynamic data set are calculated to determine the preliminary distribution. If the frequency or intensity of the preliminary distribution exceeds the preset threshold, a time series prediction algorithm is used to obtain the future trend of the abnormal signals. According to the trend, the data collection frequency is adjusted and the monitoring parameters of the supply chain nodes are optimized to finally determine the preliminary distribution of the abnormal signals.

[0023] Specifically, when the supply chain node interface collects real-time transaction records and logistics information of each node in the supply chain, the data containing time stamps can be extracted from the supply chain management system through API. For example, the warehouse node of an e-commerce platform uploads the logistics status and transaction order information of packages every minute to generate a dynamic data set, recording, for example, order ID, delivery time, logistics status, etc. Time series analysis can use a sliding window method to extract the time sequence characteristics of transaction records and logistics information. For example, the frequency of package status changes is analyzed every hour, and if 10 packages are processed per minute in a normal state, but it suddenly increases to 50 in a certain hour, it is considered as an abnormal signal. When the abnormal signal exceeds the threshold (such as 30 / minute), the subsequent processing is triggered.

[0024] In one possible implementation, the abnormal signals are classified by a K-means clustering algorithm. For example, the abnormal signals are classified into two categories, high-frequency abnormal signals and low-frequency abnormal signals, based on transaction volume and logistics delay time. The high-frequency abnormal signals can be repeated orders caused by system errors, and the low-frequency abnormal signals can be logistics blockage. After classification, a decision tree algorithm is used to analyze the relevance between transaction records and logistics information. For example, a certain abnormal category shows that the transaction volume increases sharply but the logistics state is not updated, and the decision tree infers that the repeated orders are caused by system errors, and generates a distribution pattern indicating that the abnormal signals are concentrated in a specific time period such as a promotion period.

[0025] Specifically, the distribution pattern can calculate the frequency and intensity of the abnormal signals. For example, the frequency of the abnormal signals is 20 times / hour and the intensity is 1000 orders, which is much higher than the threshold of 10 times / hour or 500 orders. A time series prediction algorithm such as ARIMA can predict the abnormal trend in the next 24 hours based on historical abnormal data. For example, the prediction shows that the frequency of the abnormal signals will continue to rise at midnight the next day. Based on this trend, the data collection frequency is adjusted, for example, from once every minute to once every 30 seconds, the monitoring parameters are optimized, and finally the preliminary distribution of the abnormal signals is determined.

[0026] It can be understood that the optimized preliminary distribution of the abnormal signals can improve the efficiency of the supply chain. For example, after adjusting the monitoring parameters in real time, the system can issue a warning within 5 minutes of the occurrence of an abnormality, instead of the original 15 minutes, reducing the order processing delay. In addition, the abnormal classification and prediction results can provide decision support for supply chain managers, such as strengthening system verification for repeated order abnormalities and reducing the error rate by 10%.

[0027] Preferably, through multi-dimensional analysis, the time sequence characteristics, classification and distribution pattern of the abnormal signals form a closed loop feedback, continuously optimizing the monitoring effect of the supply chain nodes and ensuring the stability of logistics and transactions.

[0028] Step S200: Based on the dynamic data set, a fusion result of the latest transaction information and historical data is obtained For the acquired dynamic dataset, a sliding window algorithm is used to traverse the local patterns of time series data contained therein, extract statistical indicators of transaction information within the window, and obtain statistical features of the local patterns. If the statistical features of the local patterns deviate from a preset threshold, a principal component analysis algorithm is used to compress the feature dimension, and a feature set after dimension reduction is obtained. According to the feature set after dimension reduction, a k-means clustering algorithm is used to group the local patterns, and a clustering result of the transaction information is obtained. Through the clustering result, the timestamp interval distribution of each group of transaction information is calculated to determine the periodic pattern of the time series. If the timestamp interval of the periodic pattern exceeds a preset threshold, a moving average algorithm is used to smooth the time series data to obtain a smoothed sequence trend. According to the smoothed sequence trend, the size and step length of the sliding window are adjusted to obtain an optimized time series local pattern. Through the optimized time series local pattern, dynamic features of the transaction information and historical data fusion are extracted, and finally the fusion result of the latest transaction information and historical data within the window is obtained.

[0029] For example, for a dynamic dataset, a sliding window algorithm can divide the supply chain transaction time series contained therein into windows of a fixed size, such as one window per hour, to traverse the local patterns to capture transaction fluctuations. In one possible implementation, statistical indicators are extracted within the window, including transaction volume mean and variance. If the transaction volume mean of a certain window reaches 500 and the threshold is only 300, it deviates from the normal.

[0030] Specifically, the principal component analysis algorithm compresses the dimensions of these indicators from 10 to 3, retains the main variance such as transaction peak and delay distribution, and obtains a dimension-reduced feature set for subsequent processing. For example, the k-means clustering algorithm groups the local patterns into 3 categories based on the dimension-reduced set: high-peak transaction group, stable group, and fluctuation group, and the center points of each group reflect the differences in transaction density.

[0031] It should be noted that the timestamp interval distribution is calculated through the clustering result, such as an average interval of 5 minutes in the peak group, which reveals a periodic pattern such as a daily promotion peak. In one embodiment, if the interval of the periodic pattern exceeds the threshold of 10 minutes, the moving average algorithm applies smoothing with a window size of 15 to filter out noise and obtain a trend line to show increasing transactions.

[0032] Preferably, according to the smoothed trend, the size of the sliding window is increased from 1 hour to 30 minutes, and the step length is halved to improve the sensitivity to sudden transactions. For example, the optimized local pattern fuses historical data to extract dynamic features such as order matching rate, and finally obtains the fusion result of the latest transaction information and historical data within the window.

[0033] In one possible implementation, this process forms a rigorous chain from window traversal to clustering grouping: statistical deviation triggers dimension reduction, clustering reveals periodicity, smoothing optimizes window, and finally fusion features generate fusion results. Specifically, with multi-faceted support, peak group interval distribution aligns with historical promotion data, smooth trend predicts next-day peak, and supports decision-making such as increasing logistics resources.

[0034] It can be understood that this method can accurately generate fusion results, improve supply chain monitoring accuracy, and reduce transaction delay risk.

[0035] Step S300: Based on the fusion results, obtain quality fluctuation correlation indicators For the obtained fusion results, if the delivery delay signs exceed the preset threshold, the distribution characteristics of the abnormal points are extracted through time series analysis to determine potential risk events. According to the distribution characteristics of the abnormal points, the principal component analysis algorithm is used to compress the quality fluctuation data to obtain a feature set after dimension reduction. Through the feature set after dimension reduction, the correlation strength between supply chain nodes is calculated to obtain a dynamic correlation matrix between nodes. If the correlation strength in the dynamic correlation matrix is lower than the preset threshold, the moving average algorithm is used to smooth the time series data to obtain a smoothed trend sequence. According to the smoothed trend sequence, the fluctuation frequency of the periodic pattern is extracted to determine the abnormal distribution pattern of the supply chain nodes. Through the abnormal distribution pattern, the transaction information deviation value of each node is calculated to obtain a quantitative indicator of potential risks. According to the quantitative indicator of potential risks, the window parameters of time series analysis are adjusted to obtain an optimized risk prediction model, and finally the corresponding quality fluctuation correlation indicators are obtained.

[0036] For example, in supply chain management, for the fusion results, if the delivery delay exceeds the preset threshold (such as 2 hours set to 3 hours actually delayed), the distribution characteristics of the abnormal points can be extracted through time series analysis. Time series analysis observes the timestamps of transaction data to identify abnormal points, such as a warehouse node continuously delaying for multiple days. Assuming that the daily order volume of a certain node is 1000, the proportion of delayed orders increases to 20%, which is much higher than the normal threshold of 5%, it can be determined as a potential risk event. Specifically, by analyzing the time distribution of abnormal points, it is revealed that the delay is concentrated in the logistics peak period of 17:00-19:00 every day, which indicates the scheduling bottleneck.

[0037] In one possible implementation, the principal component analysis algorithm is used to compress the quality fluctuation data. The quality fluctuation data includes 10-dimensional indicators such as order delay time, return rate, and inventory turnover rate. Through principal component analysis, the first 3-dimensional features such as the combination of delay time and return rate are extracted, retaining 80% of the original data variance. For example, the average delay time of a certain warehouse node is 2.5 hours, and the return rate is 10%. After dimension reduction, a low-dimensional feature set is generated, highlighting the correlation between delay and return, facilitating subsequent analysis of node correlation.

[0038] For example, the correlation strength between supply chain nodes is calculated by the reduced feature set, and a dynamic correlation matrix is constructed. The correlation strength is calculated based on the order flow frequency and delay correlation between nodes. Assuming that the order flow frequency between warehouse A and distribution center B is 500 times per day, but the delay correlation is as low as 0.3, which is lower than the threshold value 0.5, indicating that the cooperation efficiency between the two is low. The dynamic correlation matrix intuitively reflects the weak correlation between A and B, prompting the need to optimize the logistics path.

[0039] Specifically, if the correlation strength is lower than the threshold value, a moving average algorithm can be used to smooth the time series data. The moving average uses a 7-day window to smooth the daily delay data and filter out short-term fluctuations. For example, a node delay time series shows daily fluctuations of 1.8-3.2 hours, and the smoothed trend shows that the delay gradually increases to 2.8 hours, revealing that the potential risk is increasing. The smoothed trend sequence helps identify long-term patterns, such as delays increasing with order volume.

[0040] In one possible implementation, the fluctuation frequency of the periodic pattern is extracted by smoothing the trend sequence. Assuming that the analysis finds that the delay peak occurs 2 times a week, the frequency coincides with the promotion activity, and it can be determined that the abnormal distribution pattern is promotion-driven. Further calculate the transaction information bias value, such as the delay bias of a certain node is 1.5 hours, which is higher than the average of 0.5 hours, and the risk level is quantified as high. According to the quantification index of potential risk, the window parameter of time series analysis is adjusted (the original window is 1 hour, if the risk index shows high-frequency fluctuations, it can be shortened to 30 minutes), the sensitivity to sudden delays is improved, and finally the corresponding quality fluctuation correlation index is obtained.

[0041] This method forms a complete chain through abnormal distribution analysis, dimensionality reduction, smoothing, and window optimization, accurately locates supply chain risk points, and enhances monitoring capabilities. In one embodiment, the dynamic correlation matrix between nodes is combined with periodic patterns to reveal the impact of promotion activities on delays. For example, the order volume increases by 30% on the promotion day, causing abnormal delay distribution at a specific node, and the model accordingly recommends additional transportation resources. This multi-faceted analysis, from abnormal points to risk quantification, has a rigorous logic that collectively supports the accuracy of the quality fluctuation correlation index and improves supply chain management efficiency.

[0042] Step S400: determining the occurrence position of the evidence chain break based on the quality fluctuation correlation index For the obtained quality fluctuation correlation indicators, an anomaly detection algorithm is used to process the associated logistics data, an anomaly pattern is extracted through an isolation forest algorithm, and an abnormal distribution feature is obtained. According to the abnormal distribution feature, the transaction information deviation between the nodes of the supply chain is calculated, and the node abnormal value is obtained. If the node abnormal value exceeds the preset threshold value, the time trend of the node transaction is extracted through a time series decomposition method, and the abnormal fluctuation mode is determined. According to the abnormal fluctuation mode, a clustering algorithm is used to group the nodes of the supply chain, and the node abnormal cluster is obtained. Through the node abnormal cluster, the statistical deviation of the transaction information in each cluster is calculated, and the evidence chain breaking position is determined. According to the evidence chain breaking position, the sensitivity parameter of the anomaly detection algorithm is adjusted, and the optimized anomaly detection model is obtained. Through the optimized anomaly detection model, the logistics data is reprocessed, and finally the position of the evidence chain breaking is determined.

[0043] For example, in supply chain management, for quality fluctuation correlation indicators, an isolation forest algorithm quickly identifies potential risks by analyzing abnormal points in associated logistics data. Isolation forest is based on the principle of randomly partitioning data points, and abnormal points are more easily isolated due to their sparse distribution. Suppose a logistics node processes 5000 orders per day, and the delivery time of 20 orders is abnormally extended to 5 hours, far exceeding the normal average of 2 hours. The isolation forest algorithm marks these orders as abnormal, generates abnormal distribution features, and reflects the degree of deviation of delivery time from the normal mode. This method can efficiently process high-dimensional data and is suitable for real-time monitoring of complex supply chains.

[0044] Specifically, according to the abnormal distribution feature, the transaction information deviation between nodes is calculated. For example, warehouse C sends 1000 orders to distribution center D every day, with a normal deviation of 0.5 hours, but the deviation suddenly increases to 2 hours on a certain day, exceeding the preset threshold of 1 hour. This indicates that there may be problems in the logistics cooperation between C and D, such as insufficient transportation vehicle scheduling. Deviation calculation compares the difference between actual delivery time and historical average value to quantify the degree of abnormality, providing a basis for subsequent analysis.

[0045] In one embodiment, a time series decomposition method is used to extract the time trend of node transactions. Suppose the time series data of node D shows that delivery delays are concentrated on Friday afternoons every week, and after decomposition, it is found that the trend component shows that the delay gradually increases, and the seasonal component reveals the fluctuation of the peak period every Friday. This suggests that the delay may be related to the surge of weekend orders. Time series decomposition splits data into trend, seasonality, and residual, helping to accurately locate abnormal fluctuation patterns.

[0046] For example, using the K-means clustering algorithm to group supply chain nodes, based on abnormal fluctuation patterns to divide node clusters. Assuming the analysis of 20 nodes, the clustering result generates 3 clusters: high-risk, medium-risk and low-risk clusters. The high-risk cluster contains 5 nodes, such as warehouse C, whose average delay time is 3 hours, much higher than the 1.5 hours of other clusters. Clustering groups by the similarity of abnormal features between nodes facilitates targeted optimization of resource allocation.

[0047] Specifically, through node abnormal clusters, the statistical deviation of transaction information in each cluster is calculated to preliminarily locate the evidence chain break position. For example, in the high-risk cluster, the order flow deviation between warehouse C and distribution center D is 2.5 hours, far exceeding the normal 0.8 hours, indicating that there is a bottleneck in the logistics path from C to D, such as insufficient transportation capacity. The statistical deviation reveals the key nodes of supply chain disruption by aggregating the delay and order volume data of nodes within the cluster.

[0048] In one embodiment, the sensitivity parameter of the Isolation Forest algorithm is adjusted to optimize the anomaly detection model. The original sensitivity setting is to detect 5% abnormal points, and if the high-risk cluster shows frequent abnormalities, it can be increased to 10% to capture more potential risks. For example, the adjusted model identifies that the C node's delay on the promotion day has increased to 6 hours, which is better than the original model's missed detection. This optimization improves the model's response to sudden abnormalities.

[0049] For example, the optimized model reprocesses the logistics data to finally determine the occurrence position of the evidence chain break. Assuming the analysis finds that the break of the C to D path is concentrated in 16:00-18:00 every day, combined with order volume data, it shows that the promotion activity has caused a 30% surge in cargo volume. The accurate break position distribution supports scheduling optimization, such as adding night transportation vehicles. This multi-level analysis, from anomaly detection to model optimization, forms a complete logical chain, significantly improving the supply chain risk monitoring capability.

[0050] Step S500: Based on the evidence chain break position, obtain the time-sensitive data dominant analysis framework According to the determined position of the evidence chain breakage, abnormal node data is obtained, time series analysis is used to extract the fluctuation period of node transactions, and periodic abnormal patterns are determined. According to the periodic abnormal patterns, clustering algorithms are used to group the supply chain nodes, and abnormal node clusters are obtained. From the abnormal node cluster, transaction information deviation is obtained, and statistical analysis is used to calculate the deviation distribution to determine the deviation concentration area. If the deviation concentration area exceeds the preset threshold, the time series decomposition method is used to extract the time sequence characteristics of the abnormal node to determine the abnormal duration. According to the abnormal duration, the time span of the sliding window is adjusted to obtain the updated audit time period. From the updated audit time period, transaction data is obtained, and an anomaly detection algorithm is used to process the data to determine the optimized evidence chain breakage position. According to the optimized evidence chain breakage position, time-sensitive analysis logic is generated, and finally the updated time-sensitive data-driven analysis framework is obtained.

[0051] For example, in supply chain management, abnormal node data is obtained from the determined position of the evidence chain breakage, and time series analysis can be used to mine transaction fluctuation periods. Time series analysis identifies the rules of abnormal occurrence by decomposing the periodic patterns in the data. Suppose a distribution center E processes 6000 orders per day, and analysis finds that order processing time fluctuates greatly every Wednesday, with an average delay of 2.5 hours, far exceeding the normal 1 hour. Time series analysis reveals that the fluctuation period is every week, suggesting that the anomaly may be related to the weekly peak of goods volume. This method extracts periodic features to provide a basis for subsequent clustering.

[0052] For example, based on periodic abnormal patterns, K-means clustering algorithm is used to group supply chain nodes. Clustering algorithms divide nodes into different risk clusters based on transaction fluctuation characteristics. Suppose 30 nodes are analyzed, and clustering generates 4 clusters, of which the high-risk cluster contains 6 nodes, such as warehouse F, whose average delay is 3 hours. Clustering groups by similarity of fluctuation period and delay degree, making it easier to identify abnormal concentration areas. This grouping method helps focus on problem nodes and improves resource allocation efficiency.

[0053] Specifically, transaction information deviation is obtained from the abnormal node cluster, and statistical analysis is used to calculate the deviation distribution. Statistical analysis judges the deviation concentration area by aggregating order delay and goods volume data. Suppose the order flow deviation from warehouse F to distribution center G is 2.8 hours, concentrated in 14:00-16:00 every day, exceeding the threshold of 1.5 hours. Deviation distribution analysis shows that the anomaly is mainly caused by insufficient vehicle scheduling during the peak period of goods volume. This analysis clearly shows the spatio-temporal distribution of deviation, providing support for accurately locating the problem.

[0054] In one embodiment, if the deviation concentration area exceeds the threshold, the time series decomposition method can extract the time series features of the abnormal node and determine the abnormal duration. The decomposition method splits the data into trend, seasonal, and residual components. Assuming that the G node data shows that the delay lasts for 4 hours on Thursday afternoon, the seasonal component reveals the peak fluctuation, and the trend component shows that the delay gradually intensifies. This indicates that the anomaly is related to the surge in cargo volume on Thursday. The identification of abnormal duration helps to optimize the monitoring strategy.

[0055] For example, according to the abnormal duration, the time span of the sliding window is adjusted to obtain the updated audit time period. The sliding window focuses on the period with high incidence of anomalies by dynamically adjusting the analysis range. Assuming that the original window is 24 hours, it is adjusted to 12 hours, covering 14:00-02:00 every day, and the audit finds that the G node anomaly is concentrated on the promotion day, and the delay increases to 5 hours. This adjustment improves the timeliness of data analysis and optimizes the accuracy of anomaly capture.

[0056] Specifically, transaction data is obtained from the updated audit time period, and an anomaly detection algorithm such as DBSCAN is used to process the data to determine the optimized evidence chain break position. DBSCAN identifies abnormal points through density clustering, and assuming that the break of the F to G path is found to be concentrated on the promotion day from 16:00 to 18:00, the order volume increases by 40%. This method efficiently processes irregularly distributed data and accurately locates the break position.

[0057] In one embodiment, based on the optimized evidence chain break position, a timeliness-dominated analysis logic is generated, and an updated timeliness data-dominated analysis framework is finally obtained. This analysis framework integrates time series features, clustering results, and deviation distribution to dynamically monitor abnormal trends. Assuming that the framework shows that the abnormal distribution of the F to G path fluctuates with the promotion activities, and the delay peak is concentrated on weekends. This framework improves the response speed and accuracy of supply chain anomaly monitoring through multi-dimensional analysis.

[0058] Step S600: Based on the timeliness data-dominated analysis framework, an optimized prediction and early warning sequence is obtained For the risk prediction deviation displayed by the updated timeliness data dominant analysis framework, the abnormal signals are grouped according to the grouping basis, and the signal cluster set is obtained through the clustering algorithm. The signal intensity distribution is obtained from the signal cluster set, the distribution characteristics are calculated by using the statistical analysis method, and the concentrated distribution area is judged. If the concentrated distribution area exceeds the preset threshold, the dynamic change trend of the signal cluster is extracted through time series analysis to determine the change trend. According to the change trend, the input parameters of the prediction model are adjusted to obtain the optimized prediction parameter set. From the optimized prediction parameter set, the prediction warning sequence is generated by using the regression analysis method to obtain the warning sequence data. According to the warning sequence data, the risk prediction framework is adjusted by the deviation correction method to obtain the updated risk prediction model. From the updated risk prediction model, the priority ranking of the abnormal signals is extracted, and the optimized prediction warning sequence is finally obtained.

[0059] For example, in supply chain management, for the risk prediction deviation displayed by the updated timeliness data dominant analysis framework, abnormal signal grouping is the key to optimizing abnormal monitoring. Abnormal signals usually refer to abnormal fluctuations in order processing, such as delays, inventory shortages, etc. By grouping abnormal signals through clustering algorithms, similar characteristics of abnormalities can be classified into a signal cluster set. Assuming that a logistics network contains 50 distribution nodes, and the total number of orders processed daily is 100,000, the abnormal signals include order delays and cargo surges. Using the K-means clustering algorithm, according to the delay time and cargo fluctuation amplitude, the abnormal signals are divided into three clusters: high-frequency high-intensity cluster, medium-frequency medium-intensity cluster, and low-frequency low-intensity cluster. The high-frequency high-intensity cluster contains 8 nodes, such as distribution center H, with an average delay of 3 hours per day and a cargo fluctuation of more than 30%. This grouping method facilitates the identification of abnormal concentration areas and provides a basis for subsequent analysis.

[0060] In one possible implementation, the signal intensity distribution is obtained from the signal cluster set, and the distribution characteristics are calculated by using the statistical analysis method. The signal intensity distribution reflects the concentration degree of abnormalities, such as by calculating the standard deviation and mean of the delay time to determine the abnormal concentration area. Assuming that the standard deviation of the delay time of the high-frequency high-intensity cluster is 1.5 hours, and the mean is 3 hours, which exceeds the preset threshold of 1 hour. Statistical analysis shows that the abnormality is concentrated in 15:00-17:00 every day, and the cargo surge causes scheduling pressure. Such distribution characteristics provide data support for subsequent dynamic analysis.

[0061] For example, by time series analysis to extract the dynamic trend of signal cluster, determine the periodicity of the abnormal. Time series analysis decomposes data into trend component and seasonal component, reveals the change pattern of abnormal over time. Assuming that the order data analysis of distribution center H shows that the delay is significant on Friday afternoon, the trend component indicates that the delay is gradually increasing, and the seasonal component reveals the correlation with the high peak of goods volume on promotion day. Based on this trend, adjust the input parameters of the prediction model, such as increasing the weight of high peak of goods volume, get the optimized prediction parameter set, improve the sensitivity of the model to abnormal.

[0062] In one possible implementation, regression analysis method is used to generate prediction warning sequence from the optimized prediction parameter set. Regression analysis predicts the possibility of future abnormal occurrence by fitting historical delay and goods volume data. Assuming that the prediction model shows that distribution center H may have a 4-hour delay on the future promotion day, the warning sequence data marks the high-risk period as 16:00-18:00 every day. This prediction warning sequence provides a precise time window for risk prediction.

[0063] For example, by adjusting the risk prediction framework through bias correction method, get the updated risk prediction model. Bias correction optimizes model parameters by comparing predicted values with actual values. Assuming that the actual delay is 4.5 hours and the predicted value deviates by 0.5 hours, the corrected model more accurately identifies high-risk nodes. From the updated risk prediction model, extract the priority ranking of abnormal signals, and finally obtain the optimized prediction warning sequence. For example, distribution center H is marked as the highest priority, and the warning sequence shows that its abnormality is concentrated on the promotion day, which requires priority allocation of scheduling resources. This way, through multi-dimensional analysis, the accuracy and response efficiency of supply chain abnormal monitoring are improved.

[0064] Step S700: Based on the prediction warning sequence, determine the complete path of the continuity evidence chain For the obtained optimized prediction warning sequence, extract the transaction information sequence and logistics abnormal signal from it, integrate multi-source data by using data fusion rules to obtain a fusion data set. From the fusion data set, extract transaction timestamps and logistics state updates, determine time sequence correlation characteristics by using time series analysis to obtain a time sequence correlation set. If the time sequence correlation set meets the abnormal trigger condition, analyze the length of the evidence chain by path consistency verification to determine the continuity evidence path. According to the continuity evidence path, extract transaction correlation identifiers and logistics node sequences by using association rule mining algorithm to obtain an association path set. From the association path set, group the path consistency verification by using clustering analysis method to obtain an optimized path cluster. By the optimized path cluster, judge the effectiveness of the complete path structure by using a preset threshold to determine the final complete path. From the final complete path, extract the matching relationship between transaction correlation identifiers and logistics node sequences to finally determine the complete path of the continuity evidence chain.

[0065] For example, in supply chain management, the fusion of transaction information sequence and logistics abnormal signal is a key step for optimized monitoring from the obtained optimized prediction warning sequence. The transaction information sequence usually contains order time, amount, customer information, etc., while the logistics abnormal signal includes delivery delay, inventory anomaly, etc. The data fusion rule forms a unified data set by integrating these multi-source data. For example, the logistics network of an e-commerce platform processes 200,000 orders daily, including transaction timestamps and logistics status updates, such as order confirmation time, delivery node arrival time, etc. The fusion rule can align the transaction data with the logistics status through timestamps to generate a fusion data set that records the entire information from order placement to delivery for each order.

[0066] In one possible implementation, the transaction timestamp and logistics status update are extracted from the fusion data set, and time series analysis is used to determine the time series correlation characteristics. Time series analysis focuses on data changes in the time dimension, such as extracting delay characteristics by analyzing the difference between order confirmation time and delivery arrival time. Assuming that the order confirmation time of a certain delivery node is 10:00 and the delivery arrival time is 16:00, the delay is 6 hours, and the analysis shows that the delay is concentrated in the peak period of 12:00-14:00. The time series correlation set records these characteristics, reflecting the regularity of abnormal occurrence.

[0067] For example, if the time series correlation set shows that the delay exceeds 4 hours, an abnormal condition is triggered, and path consistency verification is required. Path consistency verification ensures the continuity of transactions and logistics nodes by analyzing the length of the evidence chain. For example, an order needs to go through warehouse A, sorting center B, and delivery station C from order placement to delivery, and the verification finds that the sorting center B is delayed by 2 hours, and the length of the evidence chain is 3 nodes. The continuity evidence path records the complete process to ensure that there is no data gap.

[0068] In one possible implementation, the association rule mining algorithm is used to extract transaction association identifiers and logistics node sequences from the continuity evidence path. For example, the association strength between order ID 123 and warehouse A and sorting center B is 0.9, indicating high correlation. The association path set records the correspondence between multiple orders and nodes, such as 1000 orders involving 10 nodes. Clustering analysis groups the paths into high-frequency paths and low-frequency paths, such as high-frequency paths concentrated in warehouse A to delivery station C.

[0069] For example, by optimizing the path cluster, the effectiveness of the path structure is judged by a preset threshold. Assuming that the threshold requires the path to cover 80% of the order volume, the high-frequency path meets the condition and is determined as the final complete path. The matching relationship between the transaction association identifier and the logistics node sequence is extracted from the final complete path, such as order ID 123 matching with delivery station C, and the complete path of the continuity evidence chain is finally determined. This way improves the accuracy and efficiency of supply chain monitoring through multi-dimensional analysis.

[0070] Step S800: Based on the complete path, the audit output result of real-time monitoring is obtained For the determined complete path of the continuity evidence chain, the corresponding fixed period check legacy data is extracted, and the effective transaction record and logistics state update are extracted by using the data cleaning rule to obtain a standardized data set. If the standardized data set meets the preset integrity threshold, the transaction timestamp and logistics node sequence are extracted by time series analysis to obtain a time series association set. According to the time series association set, the matching relationship between the transaction and the logistics node is extracted by using the association rule mining algorithm to obtain an association path set. If the path consistency in the association path set meets the preset threshold, the paths are grouped by clustering analysis to obtain an optimized path cluster. From the optimized path cluster, the effectiveness of the path structure is judged by a preset threshold to obtain an effective path set. According to the effective path set, the continuity features of the transaction timestamp and the logistics node sequence are extracted to obtain the audit output basic data of real-time monitoring; from the audit output basic data, a structured transaction and logistics matching relationship is generated by using a data mapping method, and finally the audit output result of real-time monitoring is obtained.

[0071] For example, in supply chain management, for the determined complete path of the continuity evidence chain, extracting the corresponding fixed period check legacy data is an important link in optimizing monitoring. The legacy data usually includes historical order records and logistics state logs, and may contain noise data such as repeated orders or invalid logistics states. The data cleaning rule removes invalid or redundant records by filtering valid fields such as order ID, transaction timestamp and logistics node state. For example, an e-commerce platform generates 300,000 orders per day, of which 5% are repeated records. The data cleaning rule can remove duplicate records by order ID, retain unique records, and verify the integrity of the logistics state, such as ensuring that the delivery node timestamp exists, to generate a standardized data set.

[0072] In an embodiment, the standardized data set needs to meet an integrity threshold, for example, requiring 90% of order records to contain complete transaction timestamps and logistics node information. If the condition is met, the transaction timestamp and logistics node sequence timing features can be extracted through time series analysis. For example, by analyzing an order from the order time 10:00 to the delivery node arrival time 15:00, the time difference feature is extracted, and it is found that the delivery delay is concentrated in the peak period 13:00-14:00. The time sequence association set records these features, reflecting the time sequence relationship between the order and the logistics node.

[0073] For example, the association rule mining algorithm extracts the matching relationship between transactions and logistics nodes from the time sequence association set. Assuming that the association strength of order ID 456 with warehouse X and delivery station Y is 0.85, indicating high correlation. The association path set records the corresponding relationship between multiple orders and nodes, such as 2000 orders involving 8 logistics nodes. If the path consistency meets the preset threshold, for example, 80% of the order paths cover the main nodes, further processing is performed.

[0074] In an embodiment, the clustering analysis groups the association path set into high-frequency and low-frequency paths. For example, the high-frequency path is from warehouse X to delivery station Y, covering 70% of the order volume. The optimized path cluster filters out efficient paths by analyzing path frequency. The preset threshold requires that the path covers 85% of the order volume, and the high-frequency path meets the condition and is selected as the effective path set.

[0075] For example, the effective path set is used to extract continuity features, such as the complete trajectory of order ID 456 from ordering to delivery station Y, to obtain the audit output basic data for real-time monitoring. From the audit output basic data, a structured transaction and logistics matching relationship is generated using data mapping methods, for example, generating a matching record of order ID 456 and delivery station Y, and finally obtaining the audit output result for real-time monitoring. The final audit output result records the complete path from ordering to delivery, ensuring the accuracy and efficiency of monitoring.

[0076] As shown in the accompanying Figure 2 Based on the same inventive concept, the embodiment also discloses a digital audit system for a supply chain, which executes the aforementioned digital audit method for a supply chain, and includes the following modules: The data acquisition and abnormality preliminary distribution module 1 inputs real-time transaction records and logistics information of each node of the supply chain, acquires a dynamic data set containing timestamps through information acquisition, determines the preliminary distribution of abnormal signals, and outputs the dynamic data set and the preliminary distribution of abnormal signals. The sliding window data fusion module 2 inputs the dynamic data set, processes time series data using a sliding window algorithm, obtains the fusion result of the latest transaction information and historical data in the window, and outputs the fusion result. The risk judgment and quality index module 3 inputs the fusion result, judges a potential risk event if a delivery delay sign exceeds a preset threshold, acquires a quality fluctuation correlation index, and outputs the potential risk event and the quality fluctuation correlation index; The abnormality detection and fracture positioning module 4 inputs the quality fluctuation correlation index, analyzes a logistics abnormality mode by using an abnormality detection algorithm, determines a fracture position of an evidence chain, and outputs the fracture position of the evidence chain; The audit window adjustment module 5 inputs the fracture position, automatically slides an audit window by using an adjustment mechanism, obtains a time-effectiveness data dominant analysis framework, and outputs the analysis framework; The early warning sequence optimization module 6 inputs the analysis framework, uses a clustering algorithm to group an abnormal signal ignored cluster if a risk prediction deviation is displayed, obtains an optimized prediction early warning sequence, and outputs the optimized early warning sequence; The complete evidence chain determination module 7 inputs the optimized early warning sequence, fuses the latest transaction information and a logistics abnormality capture, determines a complete path of a continuous evidence chain, and outputs the complete path; The audit result generation module 8 inputs the complete path, processes fixed-period inspection residual data, obtains a real-time monitoring audit output result, and outputs the audit output result.

Claims

1. A digital auditing method for supply chains, characterized in that, Includes the following steps: Step S100: By collecting real-time transaction records and logistics information from each node of the supply chain, a dynamic dataset containing timestamps is obtained, and the preliminary distribution of abnormal signals is determined. Step S200: Based on the acquired dynamic dataset, the time series data is processed using the sliding window algorithm to obtain the fusion result of the latest transaction information and historical data within the window; Step S300: If the delivery delay indicators in the fusion results exceed the preset threshold, it is judged as a potential risk event, and the corresponding quality fluctuation correlation index is obtained; if the delivery delay indicators in the fusion results do not exceed the preset threshold, this step is stopped. Step S400: Using the obtained quality fluctuation correlation indicators, an anomaly detection algorithm is used to analyze the pattern of logistics anomaly capture and determine the location where the evidence chain breaks. Step S500: Based on the location of the determined evidence chain break, the adjustment mechanism automatically slides the audit window to obtain the updated timeliness data-driven analysis framework. Step S600: If the delivery delay indicators in the fusion results exceed the preset threshold, it is judged as a potential risk event, and the corresponding quality fluctuation correlation index is obtained; if the delivery delay indicators in the fusion results do not exceed the preset threshold, this step is stopped. If the updated framework does not show any deviation in risk prediction, then stop at this step; Step S700: By integrating the latest transaction information and logistics anomaly detection with the obtained optimized prediction and early warning sequence, the complete path of the continuous evidence chain is determined; Step S800: Based on the complete path of the determined continuous evidence chain, process the legacy data of the fixed-period inspection to obtain the audit output results of real-time monitoring.

2. The digital auditing method for the supply chain according to claim 1, characterized in that, In step S100, by collecting real-time transaction records and logistics information from each node of the supply chain, a dynamic dataset containing timestamps is obtained, and the preliminary distribution of abnormal signals is determined. This specifically includes the following sub-steps: By collecting real-time transaction records and logistics information from each node in the supply chain through the supply chain node interface, a dynamic dataset containing timestamps is generated. Time series analysis is used to extract the temporal features of transaction records and logistics information from dynamic datasets to obtain anomaly signals. If the temporal characteristics of the abnormal signal exceed a preset threshold, the abnormal signal is classified using a clustering algorithm to determine the abnormal category. Based on the anomaly category, the decision tree algorithm is used to analyze the correlation between transaction records and logistics information to obtain the distribution pattern of anomaly signals; By analyzing the distribution patterns, the frequency and intensity of abnormal signals in the dynamic dataset are calculated to determine the preliminary distribution. If the frequency or intensity of the initial distribution exceeds a preset threshold, the future trend of the abnormal signal is obtained through a time series prediction algorithm. Based on the changing trends, the data collection frequency was adjusted, the monitoring parameters of the supply chain nodes were optimized, and the preliminary distribution of abnormal signals was finally determined.

3. The digital auditing method for the supply chain according to claim 1, characterized in that, In step S200, based on the acquired dynamic dataset, the time series data is processed using a sliding window algorithm to obtain the fusion result of the latest transaction information and historical data within the window. This specifically includes the following sub-steps: For the acquired dynamic dataset, the sliding window algorithm is used to traverse the local patterns of the time series data it contains, extract statistical indicators of transaction information within the window, and obtain the statistical features of the local patterns. If the statistical features of the local pattern deviate from the preset threshold, the feature dimension is compressed by the principal component analysis algorithm to obtain the feature set after dimensionality reduction. Based on the reduced feature set, the k-means clustering algorithm is used to group the local patterns to obtain the clustering results of the transaction information. By using the clustering results, the timestamp interval distribution of each group of transaction information is calculated to determine the periodic pattern of the time series; If the timestamp interval of the periodic pattern exceeds the preset threshold, the moving average algorithm is used to smooth the time series data and obtain the smoothed series trend. Based on the smoothed sequence trend, the size and step size of the sliding window are adjusted to obtain an optimized local pattern of the time series. By optimizing the local time series model, the dynamic features of the fusion of transaction information and historical data are extracted, and the fusion result of the latest transaction information and historical data within the window is finally obtained.

4. The digital auditing method for the supply chain according to claim 1, characterized in that, In step S300, if the delivery delay indicators in the fusion result exceed a preset threshold, it is judged as a potential risk event, and the corresponding quality fluctuation correlation index is obtained; if the delivery delay indicators in the fusion result do not exceed the preset threshold, this step is stopped, specifically including the following sub-steps: If the delivery delay indicators in the obtained fusion results exceed the preset threshold, the distribution characteristics of the anomalies are extracted through time series analysis to determine potential risk events. Based on the distribution characteristics of outliers, principal component analysis algorithm is used to compress quality fluctuation data and obtain a dimensionality-reduced feature set; By using the reduced feature set, the correlation strength between supply chain nodes is calculated to obtain the dynamic correlation matrix between nodes; If the correlation strength in the dynamic correlation matrix is ​​lower than a preset threshold, the moving average algorithm is used to smooth the time series data to obtain the smoothed trend series. Based on the smoothed trend sequence, the fluctuation frequency of the periodic pattern is extracted to determine the abnormal distribution pattern of the supply chain nodes. By analyzing abnormal distribution patterns, the deviation values ​​of transaction information at each node are calculated to obtain quantitative indicators of potential risks. Based on the quantitative indicators of potential risks, the window parameters of time series analysis are adjusted to obtain an optimized risk prediction model, and finally the corresponding quality fluctuation correlation indicators are obtained.

5. The digital auditing method for the supply chain according to claim 1, characterized in that, In step S400, the obtained indicators are used to analyze the pattern of logistics anomaly capture using an anomaly detection algorithm to determine the location of the evidence chain break. This specifically includes the following sub-steps: For the obtained quality fluctuation correlation indicators, anomaly detection algorithms are used to process the associated logistics data, and anomaly patterns are extracted using the isolated forest algorithm to obtain anomaly distribution characteristics. Based on the abnormal distribution characteristics, the deviation of transaction information between supply chain nodes is calculated to obtain node anomaly values; If the node outlier exceeds the preset threshold, the time trend of the node's transactions is extracted using the time series decomposition method to determine the abnormal fluctuation pattern. Based on the abnormal fluctuation pattern, a clustering algorithm is used to group the supply chain nodes to obtain abnormal node clusters; By analyzing abnormal node clusters, the statistical deviation of transaction information within each cluster is calculated to determine the location of the broken evidence chain. Based on the location of the break in the chain of evidence, the sensitivity parameters of the anomaly detection algorithm are adjusted to obtain an optimized anomaly detection model; By reprocessing the logistics data using the optimized anomaly detection model, the location where the chain of evidence broke was ultimately determined.

6. The digital auditing method for the supply chain according to claim 1, characterized in that, In step S500, the audit window is automatically slid according to the determined position to obtain the updated timeliness data-driven analysis framework, which specifically includes the following sub-steps: For the location where the chain of evidence breaks, abnormal node data is obtained, and time series analysis is used to extract the fluctuation cycle of node transactions to determine the periodic abnormal pattern. Based on periodic anomaly patterns, clustering algorithms are used to group supply chain nodes to obtain anomaly node clusters; Obtain transaction information deviations from abnormal node clusters, use statistical analysis to calculate the deviation distribution, and determine the areas of concentrated deviation. If the area of ​​concentrated deviation exceeds a preset threshold, the temporal features of the abnormal nodes are extracted using a time series decomposition method to determine the duration of the abnormality. Adjust the time span of the sliding window based on the duration of the anomaly to obtain the updated audit period; Transaction data is obtained from the updated audit period, and anomaly detection algorithms are used to process the data to determine the location of the optimized chain of evidence break. Based on the optimized evidence chain break points, a timeliness-driven analysis logic is generated, ultimately resulting in an updated timeliness-driven data-driven analysis framework.

7. The digital auditing method for the supply chain according to claim 1, characterized in that, In step S600, if the delivery delay indicators in the fusion results exceed a preset threshold, it is judged as a potential risk event, and the corresponding quality fluctuation correlation index is obtained; if the delivery delay indicators in the fusion results do not exceed the preset threshold, the process stops at this step; if the updated framework does not show a risk prediction deviation, the process stops at this step, which specifically includes the following sub-steps: To address the risk prediction biases displayed by the updated timeliness-driven data analysis framework, a clustering algorithm is used to group the abnormal signals according to the grouping criteria, resulting in a set of signal clusters. The signal intensity distribution is obtained from the signal cluster set, and the distribution characteristics are calculated using statistical analysis methods to determine the concentrated distribution area. If the concentrated distribution area exceeds the preset threshold, the dynamic change trend of the signal cluster is extracted through time series analysis to determine the change trend; Based on the changing trend, the input parameters of the prediction model are adjusted to obtain the optimized set of prediction parameters; From the optimized set of prediction parameters, regression analysis is used to generate prediction and early warning sequences, thus obtaining early warning sequence data; Based on the early warning sequence data, the risk prediction framework is adjusted using the deviation correction method to obtain the updated risk prediction model; From the updated risk prediction model, the priority ranking of abnormal signals is extracted, and finally the optimized prediction and early warning sequence is obtained.

8. The digital auditing method for the supply chain according to claim 1, characterized in that, In step S700, the obtained sequence is used to integrate the latest transaction information and logistics anomaly detection to determine the complete path of the continuous evidence chain, specifically including the following sub-steps: For the obtained optimized prediction and early warning sequence, transaction information sequence and logistics anomaly signal are extracted from it, and multi-source data are integrated using data fusion rules to obtain a fused data set. From the fused dataset, transaction timestamps and logistics status updates are extracted, and time series analysis is used to determine time-series correlation features to obtain a time-series correlation set. If the time-series association set meets the anomaly triggering conditions, the length of the evidence chain is analyzed through path consistency verification to determine the continuous evidence path; Based on the continuous evidence path, the association rule mining algorithm is used to extract the transaction association identifier and logistics node sequence to obtain the association path set; From the set of associated paths, cluster analysis is used to group the path consistency verifications to obtain optimized path clusters; By optimizing the path cluster and using a preset threshold to determine the validity of the complete path structure, the final complete path is determined. From the final complete path, the matching relationship between transaction association identifiers and logistics node sequences is extracted to ultimately determine the complete path of the continuous evidence chain.

9. The digital auditing method for supply chains according to any one of claims 1-8, characterized in that, In step S800, based on the complete path, the residual data from fixed-period checks is processed to obtain the audit output results for real-time monitoring, specifically including the following sub-steps: For the complete path of the established continuous chain of evidence, the corresponding fixed-period inspection legacy data is extracted, and the effective transaction records and logistics status updates are extracted using data cleaning rules to obtain a standardized data set; If the standardized dataset meets the preset integrity threshold, then transaction timestamps and logistics node sequences are extracted through time series analysis to obtain a time-series associated set; Based on the temporal association set, the matching relationship between transactions and logistics nodes is extracted using an association rule mining algorithm to obtain the association path set; If the path consistency in the associated path set meets the preset threshold, then the paths are grouped through cluster analysis to obtain an optimized path cluster. The effective path set is obtained by optimizing the path cluster and judging the effectiveness of the path structure using a preset threshold. Based on the set of effective paths, the continuity features of transaction timestamps and logistics node sequences are extracted to obtain the basic data for real-time monitoring audit output. Starting from the basic data output by the audit, a data mapping method is used to generate a structured transaction and logistics matching relationship, ultimately obtaining the audit output results under real-time monitoring.

10. A digital auditing system for the supply chain, characterized in that, To implement the digital auditing method for the supply chain as described in any one of claims 1-9, the system includes the following modules: The data acquisition and preliminary anomaly distribution module takes into account real-time transaction records and logistics information from each node of the supply chain, acquires a dynamic dataset with timestamps through the collected information, determines the preliminary distribution of anomaly signals, and outputs the dynamic dataset and the preliminary distribution of anomaly signals. The sliding window data fusion module takes a dynamic dataset as input, processes the time-series data using the sliding window algorithm, obtains the fusion result of the latest transaction information and historical data within the window, and outputs the fusion result. The Risk Assessment and Quality Indicator module takes the fusion result as input, determines potential risk events if the signs of delivery delay exceed a preset threshold, obtains quality fluctuation correlation indicators, and outputs potential risk events and quality fluctuation correlation indicators. The anomaly detection and breakage location module takes quality fluctuation correlation indicators as input, uses anomaly detection algorithms to analyze logistics anomaly patterns, determines the location of the evidence chain breakage, and outputs the location of the evidence chain breakage. The audit window adjustment module allows you to input the break location and automatically slide the audit window through an adjustment mechanism to obtain a time-sensitive data-driven analysis framework, which is then output as the analysis framework. The early warning sequence optimization module takes the input analysis framework as an example. If the risk prediction deviation is displayed, it uses a clustering algorithm to group abnormal signals and ignore clusters to obtain an optimized prediction early warning sequence, and outputs the optimized early warning sequence. The complete evidence chain determination module takes an optimized early warning sequence as input, integrates the latest transaction information and logistics anomaly detection, determines the complete path of the continuous evidence chain, and outputs the complete path. The audit result generation module takes the complete path as input, processes legacy data from fixed-period checks, obtains real-time monitoring audit output results, and outputs the audit output results.

Citation Information

Patent Citations

  • Pre-positioning algorithm for abnormal fault of data interface program based on AI

    CN119621581A

  • Oil well pressure anomaly detection method and system based on deep learning

    CN120372481A

  • Financial risk intelligent early warning method based on multi-source data fusion

    CN120509981A

  • Intelligent management system for intelligent medical treatment and management method thereof

    CN120853952A

  • Destination addressing for transactions associated with a distributed ledger

    US20230036694A1