An automated processing system and method for multi-source tracking information in supply chain logistics

Abstract

This invention discloses an automated processing system and method for multi-source tracking information in supply chain logistics, relating to the field of supply chain logistics information technology. The system includes: an initial weight assignment module, a data fusion and correction module, a credibility profile update module, and a closed-loop iteration module. The initial weight assignment module is used to acquire all multi-source logistics tracking data sources associated with the target waybill, and based on the inherent attributes and historical accuracy of each data source, generates an initial credibility benchmark value and an initial fusion weight for each data source. The automated processing system and method for multi-source tracking information in supply chain logistics proposed in this invention, through a dynamic credibility assessment and fusion correction linkage mechanism, achieves automated integration and real-time correction of multi-source logistics tracking data, significantly improving information processing efficiency. It can continuously track and evaluate the credibility changes of each data source, effectively solve the problems of conflict and inconsistency between multi-source data, and enhance data collaborative processing capabilities.

Description

Technical Field

[0001] This invention relates to the field of supply chain logistics information technology, specifically to an automated processing system and method for multi-source tracking information in supply chain logistics. Background Technology

[0002] As the supply chain logistics industry rapidly develops towards end-to-end visibility and multimodal transport integration, the dimensions of logistics tracking information collection continue to expand, forming a multi-source data system encompassing satellite positioning, IoT sensing, upstream and downstream business systems, and port node regulatory data. The automated processing capability of multi-source tracking information has become a core foundation supporting end-to-end logistics management, timeliness optimization, and risk warning. The industry has developed a series of mature technical solutions focusing on multi-source data collection, standardization conversion, and fusion correction, which are widely applied in various logistics transportation scenarios, including road, sea, rail, and air transport.

[0003] Existing technical solutions mostly adopt a pre-emptive static reliability assessment and serialized data processing approach. In complex environments where logistics and transportation scenarios dynamically switch and data source signal states fluctuate in real time, the fixed-weight fusion correction mode still has room for improvement in its adaptability to scenario changes. The two-way linkage between data reliability assessment and fusion correction is insufficient, and the fusion processing results cannot drive the dynamic iteration of the reliability assessment model. When discrepancies and conflicts occur in multi-source data, manual intervention is still required in some scenarios to complete verification and correction, making it difficult to achieve highly reliable, intervention-free automated processing throughout the entire process. To address this, we propose an automated processing system and method for multi-source tracking information in supply chain logistics. Summary of the Invention

[0004] To address the aforementioned technical issues, this paper provides an automated processing system and method for multi-source tracking information in supply chain logistics. This technical solution solves the problems of existing technologies that mostly employ static reliability assessment and serialized data processing paths, have insufficient adaptability to dynamic logistics scenarios due to fixed-weight fusion modes, lack bidirectional linkage between reliability assessment and fusion correction, fail to drive model iteration with processing results, require manual intervention when multi-source data conflicts, and are difficult to achieve highly reliable, intervention-free automated processing throughout the entire process.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: An automated processing system for multi-source tracking information in supply chain logistics includes: Initial weight assignment module, data fusion and correction module, credibility profile update module, closed-loop iteration module; The initial weight assignment module is used to obtain the full multi-source logistics tracking data sources associated with the target waybill, and generate the initial credibility benchmark value and initial fusion weight of each data source based on the inherent attributes and historical accuracy of each data source. The data fusion correction module is electrically connected to the initial weight assignment module. The data fusion correction module is used to automatically fuse and correct multi-source tracking data according to a preset time slice and based on the confidence weight corresponding to the current time slice, and output the standardized tracking data of the time slice. The credibility profile update module is electrically connected to the data fusion correction module. The credibility profile update module is used to compare the deviation between the standardized tracking data and the original data of each data source within the corresponding time slice, and update the dynamic credibility profile of each data source throughout its entire life cycle based on the comparison results. The closed-loop iteration module is electrically connected to the credibility profile update module and the data fusion correction module. The closed-loop iteration module is used to convert the updated credibility profile into the fusion weight of the next time slice, driving the data fusion correction module to cyclically execute the processing flow.

[0006] Preferably, the initial credibility benchmark value and initial fusion weight for generating each data source are specifically as follows: Obtain the unique identifier information of the target waybill, and pull the full multi-source logistics tracking data source associated with the waybill based on the unique identifier information; Extract the inherent attribute parameters and full historical runtime data of each data source; The historical accuracy quantification value for each data source is obtained by statistically analyzing historical operational data; Based on the preset benchmark assignment rules, combined with the inherent attribute parameters of the data source and the historical accuracy quantification value, an initial confidence benchmark value is generated for each data source. Based on the initial credibility baseline value, the initial fusion weight corresponding to each data source is calculated; Perform compliance verification on the generated initial fusion weights. If the verification result is lower than the preset threshold, adjust the parameters of the benchmark assignment rule and re-execute the assignment process until the initial credibility benchmark value and initial fusion weights that meet the verification standards are output.

[0007] Preferably, the preset benchmark assignment rule is constructed as follows: Collect full-cycle historical operational data and fixed attribute information from all logistics tracking data sources; Perform statistical analysis on the collected historical operational data to extract the inherent attribute dimensions of each type of data source, including the subject type of the data source, the data reporting frequency, and the range of data coverage nodes; Accuracy statistics are performed on the collected historical operational data, and the historical accuracy quantification dimensions of each data source are extracted, including historical data deviation rate, node reporting completeness rate, and frequency of abnormal data occurrence. Based on the extracted inherent attribute dimension and historical accuracy quantification dimension, the baseline assignment parameters for each type of data source are quantitatively defined, including inherent attribute weight coefficient, historical accuracy weight coefficient, and upper and lower limit range of the baseline value. For each type of data source, design corresponding assignment calculation functions and parameter constraints for the quantization benchmark assignment parameters; The data source type identifier, inherent attribute dimensions, historical accuracy quantification dimensions, benchmark assignment parameters, and associated assignment calculation functions are structurally associated and stored to build a queryable benchmark assignment rule library. Establish a rule base update interface. When a new type of data source is added, the sample data and parameters of the new data source are imported through this interface, and after analysis through the same process, they are expanded into the benchmark assignment rule base.

[0008] Preferably, the standardized tracking data for outputting the time slice specifically includes: According to the preset fixed duration, the logistics tracking time slices are divided into continuous and non-overlapping segments, and the start and end time boundaries of the current time slice to be processed are locked. Retrieve the raw logistics tracking data associated with the target waybill reported by each data source within the current time slice; Retrieve the credibility weights of each data source corresponding to the current time slice; Spatiotemporal benchmark alignment processing is performed on the acquired multi-source raw logistics tracking data; Based on the credibility weight of the current time slice, perform automatic weighted fusion on the aligned multi-source data; Perform logical consistency correction on the logistics nodes for the merged data; Perform a standardized conversion of the calibrated tracking data into a preset format; The standardized tracking data undergoes quality verification. If the verification result is lower than the preset threshold, the fusion weight parameters are adjusted and the fusion correction process is re-executed until standardized tracking data that meets the quality standards is output.

[0009] Preferably, the automated fusion correction process of the multi-source tracking data specifically includes: Define unified spatiotemporal reference rules. The time reference adopts the millisecond-level timestamp of Coordinated Universal Time, and the spatial reference adopts the latitude and longitude coordinates of the National Geodetic Coordinate System 2000. Based on spatiotemporal reference rules, a unified transformation is performed on the timestamps and spatial coordinates of multi-source raw tracking data to complete the spatiotemporal alignment of multi-source data. Based on the Grubbs criterion, outlier identification is performed on the aligned multi-source data to remove outlier data points that exceed a preset deviation threshold. Based on the credibility weights of each data source within the current time slice, a multi-source data weighted fusion calculation matrix is ​​constructed; The effective multi-source data after removing outliers is input into the weighted fusion calculation matrix, and the weighted fusion calculation is performed to generate the fused tracking data. Based on the fixed node sequence rules of logistics and transportation, the node logical consistency check is performed on the fused tracking data to identify and correct logical conflicts in the fused data, and generate fused tracking data that has been corrected.

[0010] Preferably, updating the dynamic credibility profile of each data source throughout its entire lifecycle specifically involves: Lock onto the same time slice corresponding to the standardized tracking data, and extract the raw logistics tracking data reported by each data source within that time slice; Perform a dimension-by-dimensional deviation comparison between the raw tracking data from each data source and the standardized tracking data of the corresponding time slice; Based on the deviation comparison results, the real-time deviation quantization value of each data source within the current time slice is calculated; Retrieve dynamic credibility profiles for the entire lifecycle of each data source; Based on the preset profile update rules and combined with the real-time deviation quantification value, the corresponding dimensions of the dynamic credibility profile throughout the entire life cycle are iteratively updated. The updated dynamic credibility profile for the entire lifecycle is validated. If the validation result is lower than the preset threshold, the deviation comparison and profile update process is re-executed until the updated dynamic credibility profile that meets the validation criteria is output.

[0011] Preferably, the method for comparing the standardized tracking data with the original data from each data source within the corresponding time slice is as follows: Define a multidimensional quantitative index for deviation comparison, which includes timestamp deviation coefficient, spatial coordinate deviation value, node information matching degree, and data integrity ratio; Calculate the timestamp deviation coefficient for each data source, statistically analyze the absolute value of the timestamp difference between the original data and the standardized tracking data corresponding data points, and normalize the absolute value of the difference to a preset value range of 0 to 1. Calculate the spatial coordinate deviation values ​​of each data source, statistically analyze the absolute value of the difference in latitude and longitude coordinates between the corresponding data points of the original data and the standardized tracking data, and convert the absolute value of the difference into a standardized distance value; Calculate the matching degree of node information from each data source, compare the consistency of logistics node status and node trigger time between the original data and standardized tracking data, and generate a matching degree score in the range of 0 to 1. Calculate the data integrity percentage of each data source and calculate the ratio of the effective data reporting time of the original data within the corresponding time slice to the total time slice duration; The timestamp deviation coefficient, spatial coordinate deviation value, node information matching degree, and data integrity ratio are weighted and fused according to a preset fixed weight to generate the real-time deviation quantification value of the data source in the current time slice.

[0012] Preferably, the method for converting the updated credibility profile into the fusion weight for the next time slice is as follows: Obtain the updated dynamic credibility profile of each data source throughout its entire lifecycle, and extract the real-time quantitative values ​​of each dimension in the profile; Based on the preset weight mapping rules, the real-time quantified values ​​of each dimension of the portrait are converted into the initial credibility coefficient of the corresponding data source. Normalize the initial confidence coefficients of all data sources, mapping all coefficients to a numerical range of 0 to 1, and the sum of the coefficients of all data sources is 1; Boundary constraints are applied to the normalized confidence coefficients to lock the maximum and minimum value ranges of the confidence coefficients for a single data source; The credibility coefficient after boundary constraints are completed is determined as the fusion weight of each data source for the next time slice; Perform a stability check on the generated next time slice fusion weights. Once the check passes, complete the weight conversion process.

[0013] Preferably, the method for the driving data fusion correction module to cyclically execute the processing flow is as follows: Obtain the fusion weight of the next time slice that has passed the verification, and send it to the data fusion correction module to complete the real-time update of the fusion weight parameters; Lock the start and end time boundaries of the next time slice, and trigger the data fusion correction module to start the multi-source tracking data processing flow of the corresponding time slice; Real-time monitoring of the iterative execution status of the entire process, and statistical analysis of the fluctuation range of the fusion weights of each data source within a continuous time slice; When the fluctuation of the fused weights within a continuous time slice exceeds a preset threshold, the initial weight assignment module is triggered to perform the initial credibility benchmark value review process. When the number of consecutive iterations reaches the preset upper limit, or the fluctuation range of the dynamic credibility profile of each data source throughout its entire lifecycle is lower than the preset convergence threshold, the iteration process is determined to have reached the convergence state, and the current fusion weight is maintained to execute the subsequent processing process. When a data source is added or taken offline, the entire process of re-initialization and iteration is triggered, completing the closed-loop iterative management of the entire lifecycle.

[0014] An automated processing method for multi-source tracking information in supply chain logistics includes the following steps: S1. Obtain all multi-source logistics tracking data sources associated with the target waybill. Based on the inherent attributes and historical accuracy of each data source, generate the initial credibility benchmark value and initial fusion weight of each data source. S2. Based on the preset time slice and the confidence weight corresponding to the current time slice, complete the automatic fusion and correction of multi-source tracking data, and output the standardized tracking data of the time slice. S3. Compare the standardized tracking data with the original data from each data source within the corresponding time slice, and update the dynamic credibility profile of each data source throughout its entire lifecycle based on the comparison results. S4. Convert the updated credibility profile into the fusion weight of the next time slice, and repeat steps S2 to S3 to form a fully automated closed-loop processing flow.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: The automated processing system and method for multi-source tracking information in supply chain logistics proposed in this invention achieves automated integration and real-time correction of multi-source logistics tracking data through a dynamic credibility assessment and fusion correction linkage mechanism, significantly improving information processing efficiency. Its full lifecycle dynamic credibility profiling technology can continuously track and evaluate the credibility changes of each data source. Combined with an adaptive fusion weight adjustment strategy, it effectively solves the conflict and inconsistency problems between multi-source data, enhances data collaborative processing capabilities, and outputs standardized tracking data according to preset time slices, supporting real-time dynamic tracking of the entire logistics process and ensuring the timeliness and accuracy of tracking data. Through a closed-loop iteration module, the system converts the updated credibility profile into the fusion weight of the next time slice in real time, forming a dynamic feedback mechanism to improve intelligent decision support capabilities. Through automated processing and dynamic weight adjustment, it reduces manual intervention, optimizes logistics management strategies, provides timely and reliable decision support for supply chain logistics managers, and effectively improves the overall operational efficiency and risk control capabilities of supply chain logistics. Attached Figure Description

[0016] Figure 1 This is a system module diagram of the present invention; Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0017] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0018] Reference Figure 1 As shown, an automated processing system for multi-source tracking information in supply chain logistics includes: Initial weight assignment module, data fusion and correction module, credibility profile update module, closed-loop iteration module; The initial weight assignment module is used to obtain the full multi-source logistics tracking data sources associated with the target waybill, and generate the initial credibility benchmark value and initial fusion weight of each data source based on the inherent attributes and historical accuracy of each data source. The initial credibility benchmark value and initial fusion weight for generating each data source are specifically as follows: Obtain the unique identifier information of the target waybill, and pull the full multi-source logistics tracking data source associated with the waybill based on the unique identifier information; Extract the inherent attribute parameters and full historical runtime data of each data source; The historical accuracy quantification value for each data source is obtained by statistically analyzing historical operational data; Based on the preset benchmark assignment rules, combined with the inherent attribute parameters of the data source and the historical accuracy quantification value, an initial confidence benchmark value is generated for each data source. Based on the initial credibility baseline value, the initial fusion weight corresponding to each data source is calculated; Perform compliance verification on the generated initial fusion weights. If the verification result is lower than the preset threshold, adjust the parameters of the benchmark assignment rule and re-execute the assignment process until the initial credibility benchmark value and initial fusion weights that meet the verification standards are output.

[0019] The preset benchmark assignment rule is constructed as follows: Collect full-cycle historical operational data and fixed attribute information from all logistics tracking data sources; Perform statistical analysis on the collected historical operational data to extract the inherent attribute dimensions of each type of data source, including the subject type of the data source, the data reporting frequency, and the range of data coverage nodes; The data source entity type is stratified and quantified according to the performance qualifications and cooperation attributes of logistics participants. Participant types include carriers, warehousing providers, IoT device providers, cargo owners, and third-party supervision platforms. Corresponding quantitative score ranges are set for different entity types: core self-operated carriers: 0.8 to 1.0; long-term cooperative carriers: 0.6 to 0.8; temporary cooperative carriers: 0.4 to 0.6; self-operated warehousing providers: 0.75 to 0.95; third-party cooperative warehousing providers: 0.55 to 0.75; compliant certified IoT device providers: 0.7 to 0.9; ordinary device providers: 0.4 to 0.7; and official supervision platforms: 0.85 to 1.0. This completes the stratified quantitative assignment of entity type dimensions.

[0020] The fluctuation range of data reporting frequency is mapped to a standardized numerical indicator through a frequency fluctuation coefficient. First, a standard reporting interval for logistics tracking data is preset. Then, the actual average reporting interval within the historical period of the data source is used. The formula for calculating the frequency fluctuation coefficient is: in the formula For frequency fluctuation coefficient, The historical average reporting interval of the data source. The preset standard reporting interval is used, and the coefficient of variation of the reporting interval is also included to adjust the final quantified value. The coefficient of variation is the ratio of the standard deviation of the reporting interval to the arithmetic mean. The smaller the coefficient of variation, the higher the final quantified value of the frequency dimension. Finally, the quantified value of the frequency dimension is mapped to a numerical range of 0 to 1. The matching degree between the data coverage node range and the waybill path nodes is calculated. First, a standard node sequence for the entire waybill link is defined, including seven core nodes: order placement, pickup, warehousing, outbound, transit, delivery, and receipt. The ratio of the number of effective nodes historically reported by the data source to the total number of standard nodes is calculated to obtain the node quantity coverage rate. Then, the consistency ratio between the order of nodes reported by the data source and the standard node sequence is calculated to obtain the node time sequence matching degree. The formula for calculating the node coverage matching degree is: in the formula To determine the matching degree of node coverage, For node coverage, To determine the node time-series matching degree, we perform quantitative calculations to cover the node range dimension of the data.

[0021] Accuracy statistics are performed on the collected historical operational data, and the historical accuracy quantification dimensions of each data source are extracted, including historical data deviation rate, node reporting completeness rate, and frequency of abnormal data occurrence. The statistical definition of historical data deviation rate is the error between the historically reported spatiotemporal data of the data source and the official verification standard data of the same period. It is calculated using a relative error method. The time dimension deviation is calculated based on millisecond-level timestamps, and the spatial dimension deviation is calculated based on the spherical distance in the National 2000 Geodetic Coordinate System. The formula for calculating the historical data deviation rate is as follows: in the formula Historical data deviation rate To count the total number of samples, For the first The data source reports the values ​​for each sample. For the first The official verification standard value for each sample; the lower the deviation rate value, the higher the quantified accuracy value of the corresponding data source.

[0022] The statistical definition of node reporting completeness rate is the proportion of nodes that have completed valid reporting according to the standard node time sequence within the entire historical waybill cycle. For a node to report validly, the difference between the actual reporting time and the actual triggering time must be within a preset tolerance range. The formula for calculating node reporting completeness rate is as follows: in the formula The node reports the completeness rate. This refers to the number of valid reporting nodes within a single waybill. The total number of standard nodes for a single waybill is used as the arithmetic mean of the integrity rates of all waybills within the statistical period, which is the quantified value of the node reporting integrity rate for this data source.

[0023] The statistical definition of the frequency of abnormal data occurrences distinguishes between random anomalies and systematic anomalies. Random anomalies are defined as discrete abnormal data points occurring no more than twice within a single transport cycle. Systematic anomalies are defined as consecutive abnormal data points occurring three or more times within a single transport cycle, or abnormal data occurring in three or more consecutive transport cycles. The formula for calculating the quantified value of abnormal data occurrence frequency is as follows: in the formula This is a quantified value representing the frequency of occurrence of outlier data. To count the number of systematic anomalies within a statistical period, This refers to the number of random anomalies within a statistical period. To determine the total number of data reports within the statistical period, the quantified value is ultimately mapped to a numerical range of 0 to 1.

[0024] Based on the extracted inherent attribute dimension and historical accuracy quantification dimension, the baseline assignment parameters for each type of data source are quantitatively defined, including inherent attribute weight coefficient, historical accuracy weight coefficient, and upper and lower limit range of the baseline value. The inherent attribute weight coefficient and the historical accuracy weight coefficient are determined by combining the analytic hierarchy process (AHP) with the entropy weight method. First, the hierarchical judgment matrix of the two dimensions is constructed by the AHP. Subjective weights are calculated by comparing the importance of the expert dimensions pairwise. Then, the objective weights are obtained by performing dispersion analysis on the full historical data using the entropy weight method. Finally, the subjective weights and objective weights are weighted and merged in a 6:4 ratio to obtain the final values ​​of the weight coefficients of the two dimensions. The weight coefficients of each sub-dimension are calculated and determined using the same method.

[0025] The upper and lower limits of the benchmark value are dynamically adjusted according to the data source type. The upper and lower limits of highly reliable data sources are set to 0.6 to 1.0, the range of medium reliability data sources is set to 0.3 to 0.8, and the range of temporary access non-core data sources is set to 0.1 to 0.5.

[0026] For each type of data source, design corresponding assignment calculation functions and parameter constraints for the quantization benchmark assignment parameters; The data source type identifier, inherent attribute dimensions, historical accuracy quantification dimensions, benchmark assignment parameters, and associated assignment calculation functions are structurally associated and stored to build a queryable benchmark assignment rule library. Establish a rule base update interface. When a new type of data source is added, the sample data and parameters of the new data source are imported through this interface, and after analysis through the same process, they are expanded into the benchmark assignment rule base.

[0027] The rule base update interface import process establishes clear sample collection standards, automated triggering conditions, and parameter verification mechanisms. New data sources must cover at least 30 complete order cycles of end-to-end operational data, spanning no less than 90 calendar days. The automated triggering condition for the analysis process is that the interface automatically triggers the entire analysis process upon receiving a complete sample data package. The parameter verification mechanism requires that the generated baseline parameters undergo consistency verification with existing parameters of the same type. Simultaneously, backtesting using reserved sample data is conducted, with a backtesting deviation rate of less than 10%. Only after successful verification can the data be added to the database; otherwise, the collection and analysis process must be re-executed.

[0028] The data fusion correction module is electrically connected to the initial weight assignment module. The data fusion correction module is used to automatically fuse and correct multi-source tracking data according to a preset time slice and based on the confidence weight corresponding to the current time slice, and output the standardized tracking data of the time slice. The standardized tracking data for this time slice is specifically output as follows: According to the preset fixed duration, the logistics tracking time slices are divided into continuous and non-overlapping segments, and the start and end time boundaries of the current time slice to be processed are locked. Retrieve the raw logistics tracking data associated with the target waybill reported by each data source within the current time slice; Retrieve the credibility weights of each data source corresponding to the current time slice; Spatiotemporal benchmark alignment processing is performed on the acquired multi-source raw logistics tracking data; Based on the credibility weight of the current time slice, perform automatic weighted fusion on the aligned multi-source data; Perform logical consistency correction on the logistics nodes for the merged data; Perform a standardized conversion of the calibrated tracking data into a preset format; The standardized conversion preset format clearly defines the core fields, coding rules and precision standards. The core fields include seven items: unique waybill identifier, timestamp, spatial coordinates, logistics node status, node trigger time, transportation vehicle status and data source identifier. The coding rules adopt UTF-8 unified encoding, and the enumerated fields adopt the unified industry standard encoding of the National Logistics Information Platform. The time precision is retained to the millisecond level, and the spatial coordinate latitude and longitude values ​​are retained to 8 decimal places.

[0029] The standardized tracking data undergoes quality verification. If the verification result is lower than the preset threshold, the fusion weight parameters are adjusted and the fusion correction process is re-executed until standardized tracking data that meets the quality standards is output.

[0030] The threshold setting for quality verification is refined into three core indicators: data integrity requires a core field missing rate of less than 2%, data accuracy requires a deviation rate of less than 5% between the merged data and the official verification data of the same period, and data consistency requires a 100% matching degree between the node time sequence and the standard path rules. Verification can be judged as passed only if all three indicators are met simultaneously.

[0031] The specific execution process of the automated fusion correction of the multi-source tracking data is as follows: Define unified spatiotemporal reference rules. The time reference adopts the millisecond-level timestamp of Coordinated Universal Time, and the spatial reference adopts the latitude and longitude coordinates of the National Geodetic Coordinate System 2000. Based on spatiotemporal reference rules, a unified transformation is performed on the timestamps and spatial coordinates of multi-source raw tracking data to complete the spatiotemporal alignment of multi-source data. Based on the Grubbs criterion, outlier identification is performed on the aligned multi-source data to remove outlier data points that exceed a preset deviation threshold. The Grubbs criterion's threshold selection method involves setting a significance level of 0.05, querying the Grubbs threshold table based on the number of valid data points within the current time slice, calculating the arithmetic mean and standard deviation of the data points, and then calculating the Grubbs statistic for each data point. Data points with a statistic greater than the threshold are identified as outliers and removed. The data completion and smoothing strategy after outlier removal is as follows: discrete outliers are completed using linear interpolation of adjacent valid data points; continuous outlier segments are completed using weighted valid data from high-confidence data sources within the same time slice after removal; and smoothing is then performed using a sliding window of three data points.

[0032] Based on the credibility weights of each data source within the current time slice, a multi-source data weighted fusion calculation matrix is ​​constructed; The construction of the weighted fusion calculation matrix begins by assigning corresponding fusion coefficients based on the credibility weights of each data source. The fusion coefficients are positively correlated with the credibility weights, and the sum of the fusion coefficients of all data sources is 1. An m-row, n-column weighted fusion calculation matrix is ​​then constructed, where m is the number of valid data sources within the current time slice, and n is the number of standardized data dimensions. Each row of the matrix corresponds to the fusion coefficient of one data source, and each column corresponds to the weight allocation of one data dimension. Core data dimensions are assigned higher weights, while auxiliary dimensions are assigned lower weights.

[0033] Data conflicts after spatiotemporal alignment are resolved through matrix operations. Matrix multiplication is performed on the multi-source data matrix and the weighted fusion calculation matrix to obtain the fused standardized data matrix. Data from high-confidence data sources has a higher impact on the fusion result, thus achieving conflict resolution.

[0034] The effective multi-source data after removing outliers is input into the weighted fusion calculation matrix, and the weighted fusion calculation is performed to generate the fused tracking data. Based on the fixed node sequence rules of logistics and transportation, the node logical consistency check is performed on the fused tracking data to identify and correct logical conflicts in the fused data, and generate fused tracking data that has been corrected.

[0035] The rule base for logical consistency correction of logistics nodes is built based on the preset standard route rules of waybills and the historical waybill route model of the same route. The preset standard route rules define the fixed sequence of nodes in the entire logistics chain: pickup is earlier than warehousing, warehousing is earlier than outbound, outbound is earlier than transit, transit is earlier than delivery, and delivery is earlier than signing. Transit nodes of the same type execute the sequence constraints according to the geographical route sequence.

[0036] The historical path model is trained based on historical waybill data from the past 12 months along the same route, clearly defining the reasonable time intervals and timing logic for node flow. The conflict correction algorithm employs a weighted majority voting method combined with nearest neighbor timing matching. For timing conflicts, data reported by the corresponding nodes from all valid data sources are retrieved, and weighted voting is performed according to credibility weights. Timing sequences with a vote share exceeding 60% are considered valid and used to correct the conflict. If the vote does not reach the threshold, the standard timing sequence of the historical path model is used, and nearest neighbor timing matching is performed based on the timestamps of adjacent nodes to complete the correction. The corrected timing sequence must conform to the fixed order constraint of the standard path.

[0037] The credibility profile update module is electrically connected to the data fusion correction module. The credibility profile update module is used to compare the deviation between the standardized tracking data and the original data of each data source within the corresponding time slice, and update the dynamic credibility profile of each data source throughout its entire life cycle based on the comparison results. The specific steps for updating the full lifecycle dynamic credibility profile of each data source are as follows: Lock onto the same time slice corresponding to the standardized tracking data, and extract the raw logistics tracking data reported by each data source within that time slice; Perform a dimension-by-dimensional deviation comparison between the raw tracking data from each data source and the standardized tracking data of the corresponding time slice; Based on the deviation comparison results, the real-time deviation quantization value of each data source within the current time slice is calculated; Retrieve dynamic credibility profiles for the entire lifecycle of each data source; Based on the preset profile update rules and combined with the real-time deviation quantification value, the corresponding dimensions of the dynamic credibility profile throughout the entire life cycle are iteratively updated. The specific implementation of the image update rule adopts an exponential moving average algorithm with adaptive step size, and the iterative update formula is as follows: in the formula These are the updated portrait dimension values. The values ​​for the historical portrait dimensions before the update. This refers to the real-time dimension values ​​within the current time slice. The update step size is adaptively adjusted based on the fluctuation range of the real-time deviation quantization value. When the deviation is within the stable range, the step size is set to 0.1. When the deviation exceeds the stable range, the step size increases linearly with the increase of the deviation amplitude, and the maximum step size does not exceed 0.4. The same adaptive adjustment logic is used for the update step size of each dimension.

[0038] The updated dynamic credibility profile for the entire lifecycle is validated. If the validation result is lower than the preset threshold, the deviation comparison and profile update process is re-executed until the updated dynamic credibility profile that meets the validation criteria is output.

[0039] The threshold setting for validity verification clearly defines the stability indicators of the dynamic credibility profile. The variance of a single-dimensional value must not exceed 0.02, the change rate of a single-dimensional value between two consecutive updates must not exceed 0.2, and the verification can be judged as passed only if all dimensions meet the requirements at the same time.

[0040] The method for comparing the standardized tracking data with the original data from each data source within the corresponding time slice is as follows: Define a multidimensional quantitative index for deviation comparison, which includes timestamp deviation coefficient, spatial coordinate deviation value, node information matching degree, and data integrity ratio; Calculate the timestamp deviation coefficient for each data source, statistically analyze the absolute value of the timestamp difference between the original data and the standardized tracking data corresponding data points, and normalize the absolute value of the difference to a preset value range of 0 to 1. The timestamp deviation coefficient normalization process uses a linear normalization function, and the calculation formula is as follows: in the formula This is the timestamp deviation coefficient. This is the original data timestamp. To standardize data timestamps, The coefficient is 0 when the absolute value of the difference is greater than or equal to the total duration of the time slice, and 1 when the difference is 0.

[0041] Calculate the spatial coordinate deviation values ​​of each data source, statistically analyze the absolute value of the difference in latitude and longitude coordinates between the corresponding data points of the original data and the standardized tracking data, and convert the absolute value of the difference into a standardized distance value; When converting spatial coordinate deviation values ​​into standardized distance values, the geodetic distance calculation formula under the National 2000 Geodetic Coordinate System is used to calculate the actual spherical distance between two points. During the conversion process, ellipsoidal parameters are used to correct map projection distortion. Finally, the actual distance value is divided by the preset maximum tolerance distance to obtain the standardized distance value in the range of 0 to 1. The maximum tolerance distance is dynamically adjusted according to the transportation mode.

[0042] Calculate the matching degree of node information from each data source, compare the consistency of logistics node status and node trigger time between the original data and standardized tracking data, and generate a matching degree score in the range of 0 to 1. In calculating the node information matching degree, the logistics node status is compared using industry-standard enumerated values. A complete match is considered a status match when the enumerated values ​​are identical; otherwise, the matching degree is 0. The node trigger time comparison employs a dynamic fault-tolerance mechanism. A time match is considered when the absolute value of the difference is within a preset fault-tolerance range. This range is dynamically adjusted based on the node type. The time matching score decreases linearly as the absolute value of the difference increases, and the score is 0 when the difference exceeds the fault-tolerance range. The final node information matching degree is calculated as the status matching score multiplied by 0.6 plus the time matching score multiplied by 0.4.

[0043] Calculate the data integrity percentage of each data source and calculate the ratio of the effective data reporting time of the original data within the corresponding time slice to the total time slice duration; In the calculation of data integrity ratio, the effective data reporting duration is defined as the total length of the time interval corresponding to the effective data points reported by the data source within the current time slice. Effective data points refer to data points with complete spatiotemporal information and compliant node status after removing outliers. The calculation process excludes the system's preset maintenance window period and verified network abnormal interruption periods, which are not included in the missing duration statistics.

[0044] The timestamp deviation coefficient, spatial coordinate deviation value, node information matching degree, and data integrity ratio are weighted and fused according to a preset fixed weight to generate the real-time deviation quantification value of the data source in the current time slice.

[0045] When generating real-time deviation quantification values ​​through weighted fusion, the determination of preset fixed weights is based on a combination of historical data statistical analysis and expert experience. By analyzing the correlation between deviation data and data accuracy over the past 12 months, the contribution of each dimension to accuracy is calculated. Combined with expert evaluation results, the weights of the timestamp deviation coefficient, spatial coordinate deviation value, node information matching degree, and data integrity percentage are determined to be 0.3, 0.3, 0.25, and 0.15, with a total weight of 1.

[0046] The closed-loop iteration module is electrically connected to the credibility profile update module and the data fusion correction module. The closed-loop iteration module is used to convert the updated credibility profile into the fusion weight of the next time slice, driving the data fusion correction module to cyclically execute the processing flow.

[0047] The method for converting the updated credibility profile into the fusion weights for the next time slice is as follows: Obtain the updated dynamic credibility profile of each data source throughout its entire lifecycle, and extract the real-time quantitative values ​​of each dimension in the profile; Based on the preset weight mapping rules, the real-time quantified values ​​of each dimension of the portrait are converted into the initial credibility coefficient of the corresponding data source. The weight mapping rule uses a piecewise linear mapping function to convert the real-time quantized values ​​of the credibility profile into initial credibility coefficients. The mapping function expression is as follows: in the formula This is the initial credibility coefficient. To quantify the credibility profile in real time. For mapping coefficients, This is the offset. The mapping coefficient and offset are dynamically adjusted according to the value range of the portrait value. In the range of 0 to 0.3, the mapping coefficient is 0.5 and the offset is 0; in the range of 0.3 to 0.7, the mapping coefficient is 1.2 and the offset is -0.21; in the range of 0.7 to 1.0, the mapping coefficient is 0.8 and the offset is 0.06. After mapping, the initial confidence coefficient value range is locked in the range of 0 to 1.

[0048] Normalize the initial confidence coefficients of all data sources, mapping all coefficients to a numerical range of 0 to 1, and the sum of the coefficients of all data sources is 1; Boundary constraints are applied to the normalized confidence coefficients to lock the maximum and minimum value ranges of the confidence coefficients for a single data source; In the boundary constraints of the normalization process, the maximum and minimum value ranges of the credibility coefficient of a single data source are set based on the data source type and the historical weight fluctuation range. The minimum value of the core data source is set to 0.1 and the maximum value is set to 0.8. The minimum value of the non-core stable data source is set to 0.05 and the maximum value is set to 0.6. The minimum value of the temporary access data source is set to 0 and the maximum value is set to 0.3. The coefficients after the boundary constraints are re-normalized to ensure that the sum of all coefficients remains 1.

[0049] The credibility coefficient after boundary constraints are completed is determined as the fusion weight of each data source for the next time slice; Perform a stability check on the generated next time slice fusion weights. Once the check passes, complete the weight conversion process.

[0050] The stability verification metrics are the absolute and relative change rates of the fused weights of corresponding data sources in adjacent time slices. The absolute change rate of a single data source weight must not exceed 0.15, and the average relative change rate of all data source weights must not exceed 10%. Both metrics must be met simultaneously for the verification to pass. If the verification fails, it will automatically roll back to the previously verified fused weights and trigger the profile update process for review. The cause of abnormal fluctuations will be investigated, and the profile update and weight conversion process will be re-executed until the verification passes.

[0051] The method for the driving data fusion correction module to cyclically execute the processing flow is as follows: Obtain the fusion weight of the next time slice that has passed the verification, and send it to the data fusion correction module to complete the real-time update of the fusion weight parameters; Lock the start and end time boundaries of the next time slice, and trigger the data fusion correction module to start the multi-source tracking data processing flow of the corresponding time slice; Real-time monitoring of the iterative execution status of the entire process, and statistical analysis of the fluctuation range of the fusion weights of each data source within a continuous time slice; When the fluctuation of the fused weights within a continuous time slice exceeds a preset threshold, the initial weight assignment module is triggered to perform the initial credibility benchmark value review process. When the number of consecutive iterations reaches the preset upper limit, or the fluctuation range of the dynamic credibility profile of each data source throughout its entire lifecycle is lower than the preset convergence threshold, the iteration process is determined to have reached the convergence state, and the current fusion weight is maintained to execute the subsequent processing process. In determining the convergence state, the statistical method for calculating the fluctuation amplitude of the fused weights within consecutive time slices is the moving window averaging method. The moving window width is set to 5 consecutive time slices, and the coefficient of variation of the fused weights of each data source within the window is calculated. The convergence threshold is set to 0.02. When the coefficient of variation of all data source weights within the moving window is lower than 0.02, and this condition is met for 3 consecutive moving windows, the iteration is considered to have reached the convergence state. When the number of consecutive iterations reaches the preset upper limit of 100, the convergence state is forcibly determined. After reaching the convergence state, a weight re-verification is performed every 20 time slices. If the re-verification finds that the fluctuation exceeds the threshold, the closed-loop iteration process is restarted.

[0052] When a data source is added or taken offline, the entire process of re-initialization and iteration is triggered, completing the closed-loop iterative management of the entire lifecycle.

[0053] During status change processing, the trigger condition for adding a new data source is that the new data source completes the rule base entry process and undergoes a 7-day observation period. During the observation period, at least 10 complete waybill cycles of data reporting and verification must be completed, and the data reporting completeness rate must reach 80% or higher before it can be formally included in the closed-loop iteration process. During the observation period, the fusion weight is fixed at the initial value set by the rule base and does not participate in dynamic iteration updates. The trigger condition for taking a data source offline is that the data source has no data reporting for 3 consecutive waybill cycles, or verification reveals systemic data fraud or a continuous abnormal data ratio exceeding 50%. The historical profile of the offline data source is encrypted and archived and is not included in subsequent weight calculations. At the same time, the weight re-normalization and iteration process of the remaining valid data sources is automatically triggered.

[0054] refer to Figure 2 As shown, an automated processing method for multi-source tracking information in supply chain logistics includes the following steps: S1. Obtain all multi-source logistics tracking data sources associated with the target waybill. Based on the inherent attributes and historical accuracy of each data source, generate the initial credibility benchmark value and initial fusion weight of each data source. S2. Based on the preset time slice and the confidence weight corresponding to the current time slice, complete the automatic fusion and correction of multi-source tracking data, and output the standardized tracking data of the time slice. S3. Compare the standardized tracking data with the original data from each data source within the corresponding time slice, and update the dynamic credibility profile of each data source throughout its entire lifecycle based on the comparison results. S4. Convert the updated credibility profile into the fusion weight of the next time slice, and repeat steps S2 to S3 to form a fully automated closed-loop processing flow.

[0055] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. An automated processing system for multi-source tracking information in supply chain logistics, characterized in that, include: Initial weight assignment module, data fusion and correction module, credibility profile update module, closed-loop iteration module; The initial weight assignment module is used to obtain the full multi-source logistics tracking data sources associated with the target waybill, and generate the initial credibility benchmark value and initial fusion weight of each data source based on the inherent attributes and historical accuracy of each data source. The data fusion correction module is electrically connected to the initial weight assignment module. The data fusion correction module is used to automatically fuse and correct multi-source tracking data according to a preset time slice and based on the confidence weight corresponding to the current time slice, and output the standardized tracking data of the time slice. The credibility profile update module is electrically connected to the data fusion correction module. The credibility profile update module is used to compare the deviation between the standardized tracking data and the original data of each data source within the corresponding time slice, and update the dynamic credibility profile of each data source throughout its entire life cycle based on the comparison results. The closed-loop iteration module is electrically connected to the credibility profile update module and the data fusion correction module. The closed-loop iteration module is used to convert the updated credibility profile into the fusion weight of the next time slice, driving the data fusion correction module to cyclically execute the processing flow.

2. The automated processing system for multi-source tracking information in supply chain logistics according to claim 1, characterized in that, The initial credibility benchmark value and initial fusion weight for generating each data source are specifically as follows: Obtain the unique identifier information of the target waybill, and pull the full multi-source logistics tracking data source associated with the waybill based on the unique identifier information; Extract the inherent attribute parameters and full historical runtime data of each data source; The historical accuracy quantification value for each data source is obtained by statistically analyzing historical operational data; Based on the preset benchmark assignment rules, combined with the inherent attribute parameters of the data source and the historical accuracy quantification value, an initial confidence benchmark value is generated for each data source. Based on the initial credibility baseline value, the initial fusion weight corresponding to each data source is calculated; Perform compliance verification on the generated initial fusion weights. If the verification result is lower than the preset threshold, adjust the parameters of the benchmark assignment rule and re-execute the assignment process until the initial credibility benchmark value and initial fusion weights that meet the verification standards are output.

3. The automated processing system for multi-source tracking information in supply chain logistics according to claim 2, characterized in that, The preset benchmark assignment rule is constructed as follows: Collect full-cycle historical operational data and fixed attribute information from all logistics tracking data sources; Perform statistical analysis on the collected historical operational data to extract the inherent attribute dimensions of each type of data source, including the subject type of the data source, the data reporting frequency, and the range of data coverage nodes; Accuracy statistics are performed on the collected historical operational data, and the historical accuracy quantification dimensions of each data source are extracted, including historical data deviation rate, node reporting completeness rate, and frequency of abnormal data occurrence. Based on the extracted inherent attribute dimension and historical accuracy quantification dimension, the baseline assignment parameters for each type of data source are quantitatively defined, including inherent attribute weight coefficient, historical accuracy weight coefficient, and upper and lower limit range of the baseline value. For each type of data source, design corresponding assignment calculation functions and parameter constraints for the quantization benchmark assignment parameters; The data source type identifier, inherent attribute dimensions, historical accuracy quantification dimensions, benchmark assignment parameters, and associated assignment calculation functions are structurally associated and stored to build a queryable benchmark assignment rule library. Establish a rule base update interface. When a new type of data source is added, the sample data and parameters of the new data source are imported through this interface, and after analysis through the same process, they are expanded into the benchmark assignment rule base.

4. The automated processing system for multi-source tracking information in supply chain logistics according to claim 3, characterized in that, The standardized tracking data for this time slice is specifically output as follows: According to the preset fixed duration, the logistics tracking time slices are divided into continuous and non-overlapping segments, and the start and end time boundaries of the current time slice to be processed are locked. Retrieve the raw logistics tracking data associated with the target waybill reported by each data source within the current time slice; Retrieve the credibility weights of each data source corresponding to the current time slice; Spatiotemporal benchmark alignment processing is performed on the acquired multi-source raw logistics tracking data; Based on the credibility weight of the current time slice, perform automatic weighted fusion on the aligned multi-source data; Perform logical consistency correction on the logistics nodes for the merged data; Perform a standardized conversion of the calibrated tracking data into a preset format; The standardized tracking data undergoes quality verification. If the verification result is lower than the preset threshold, the fusion weight parameters are adjusted and the fusion correction process is re-executed until standardized tracking data that meets the quality standards is output.

5. The automated processing system for multi-source tracking information in supply chain logistics according to claim 4, characterized in that, The specific execution process of the automated fusion correction of the multi-source tracking data is as follows: Define unified spatiotemporal reference rules. The time reference adopts the millisecond-level timestamp of Coordinated Universal Time, and the spatial reference adopts the latitude and longitude coordinates of the National Geodetic Coordinate System 2000. Based on spatiotemporal reference rules, a unified transformation is performed on the timestamps and spatial coordinates of multi-source raw tracking data to complete the spatiotemporal alignment of multi-source data. Based on the Grubbs criterion, outlier identification is performed on the aligned multi-source data to remove outlier data points that exceed a preset deviation threshold. Based on the credibility weights of each data source within the current time slice, a multi-source data weighted fusion calculation matrix is ​​constructed; The effective multi-source data after removing outliers is input into the weighted fusion calculation matrix, and the weighted fusion calculation is performed to generate the fused tracking data. Based on the fixed node sequence rules of logistics and transportation, the node logical consistency check is performed on the fused tracking data to identify and correct logical conflicts in the fused data, and generate fused tracking data that has been corrected.

6. The automated processing system for multi-source tracking information in supply chain logistics according to claim 5, characterized in that, The specific steps for updating the full lifecycle dynamic credibility profile of each data source are as follows: Lock onto the same time slice corresponding to the standardized tracking data, and extract the raw logistics tracking data reported by each data source within that time slice; Perform a dimension-by-dimensional deviation comparison between the raw tracking data from each data source and the standardized tracking data of the corresponding time slice; Based on the deviation comparison results, the real-time deviation quantization value of each data source within the current time slice is calculated; Retrieve dynamic credibility profiles for the entire lifecycle of each data source; Based on the preset profile update rules and combined with the real-time deviation quantification value, the corresponding dimensions of the dynamic credibility profile throughout the entire life cycle are iteratively updated. The updated dynamic credibility profile for the entire lifecycle is validated. If the validation result is lower than the preset threshold, the deviation comparison and profile update process is re-executed until the updated dynamic credibility profile that meets the validation criteria is output.

7. The automated processing system for multi-source tracking information in supply chain logistics according to claim 6, characterized in that, The method for comparing the standardized tracking data with the original data from each data source within the corresponding time slice is as follows: Define a multidimensional quantitative index for deviation comparison, which includes timestamp deviation coefficient, spatial coordinate deviation value, node information matching degree, and data integrity ratio; Calculate the timestamp deviation coefficient for each data source, statistically analyze the absolute value of the timestamp difference between the original data and the standardized tracking data corresponding data points, and normalize the absolute value of the difference to a preset value range of 0 to 1. Calculate the spatial coordinate deviation values ​​of each data source, statistically analyze the absolute value of the difference in latitude and longitude coordinates between the corresponding data points of the original data and the standardized tracking data, and convert the absolute value of the difference into a standardized distance value; Calculate the matching degree of node information from each data source, compare the consistency of logistics node status and node trigger time between the original data and standardized tracking data, and generate a matching degree score in the range of 0 to 1. Calculate the data integrity percentage of each data source and calculate the ratio of the effective data reporting time of the original data within the corresponding time slice to the total time slice duration; The timestamp deviation coefficient, spatial coordinate deviation value, node information matching degree, and data integrity ratio are weighted and fused according to a preset fixed weight to generate the real-time deviation quantification value of the data source in the current time slice.

8. The automated processing system for multi-source tracking information in supply chain logistics according to claim 7, characterized in that, The method for converting the updated credibility profile into the fusion weights for the next time slice is as follows: Obtain the updated dynamic credibility profile of each data source throughout its entire lifecycle, and extract the real-time quantitative values ​​of each dimension in the profile; Based on the preset weight mapping rules, the real-time quantified values ​​of each dimension of the portrait are converted into the initial credibility coefficient of the corresponding data source. Normalize the initial confidence coefficients of all data sources, mapping all coefficients to a numerical range of 0 to 1, and the sum of the coefficients of all data sources is 1; Boundary constraints are applied to the normalized confidence coefficients to lock the maximum and minimum value ranges of the confidence coefficients for a single data source; The credibility coefficient after boundary constraints are completed is determined as the fusion weight of each data source for the next time slice; Perform a stability check on the generated next time slice fusion weights. Once the check passes, complete the weight conversion process.

9. The automated processing system for multi-source tracking information in supply chain logistics according to claim 8, characterized in that, The method for the driving data fusion correction module to cyclically execute the processing flow is as follows: Obtain the fusion weight of the next time slice that has passed the verification, and send it to the data fusion correction module to complete the real-time update of the fusion weight parameters; Lock the start and end time boundaries of the next time slice, and trigger the data fusion correction module to start the multi-source tracking data processing flow of the corresponding time slice; Real-time monitoring of the iterative execution status of the entire process, and statistical analysis of the fluctuation range of the fusion weights of each data source within a continuous time slice; When the fluctuation of the fused weights within a continuous time slice exceeds a preset threshold, the initial weight assignment module is triggered to perform the initial credibility benchmark value review process. When the number of consecutive iterations reaches the preset upper limit, or the fluctuation range of the dynamic credibility profile of each data source throughout its entire lifecycle is lower than the preset convergence threshold, the iteration process is determined to have reached the convergence state, and the current fusion weight is maintained to execute the subsequent processing process. When a data source is added or taken offline, the entire process of re-initialization and iteration is triggered, completing the closed-loop iterative management of the entire lifecycle.

10. A method for automated processing of multi-source tracking information in supply chain logistics, characterized in that, Includes the following steps: S1. Obtain all multi-source logistics tracking data sources associated with the target waybill. Based on the inherent attributes and historical accuracy of each data source, generate the initial credibility benchmark value and initial fusion weight of each data source. S2. Based on the preset time slice and the confidence weight corresponding to the current time slice, complete the automatic fusion and correction of multi-source tracking data, and output the standardized tracking data of the time slice. S3. Compare the standardized tracking data with the original data from each data source within the corresponding time slice, and update the dynamic credibility profile of each data source throughout its entire lifecycle based on the comparison results. S4. Convert the updated credibility profile into the fusion weight of the next time slice, and repeat steps S2 to S3 to form a fully automated closed-loop processing flow.

Images