A logistics task execution state consistency maintenance method and system

Abstract

The application is a logistics task execution state consistency maintenance method and system, and relates to the technical field of logistics information management, comprising: fusing the multi-source state evidence vector based on dynamic credibility weight according to D-S evidence theory, outputting the fusion state of each task node and its credibility probability; when the credibility probability is lower than a high confidence threshold, marking the task node as a to-be-detected state. In the application, dynamic credibility weight is assigned to multiple data sources and D-S evidence theory is used for fusion, so that the credibility probability of the fusion state can comprehensively reflect the historical accuracy, signal quality, data timeliness and independence of each data source, thereby reducing the probability of state misjudgment caused by signal loss or data delay of a single data source.

Description

Technical Field

[0001] This invention relates to the field of logistics information management technology, and in particular to a method and system for maintaining the consistency of logistics task execution status. Background Technology

[0002] Logistics tasks involve multiple stages during execution, including warehousing, transportation, delivery, and receipt confirmation. The status data for each stage is distributed across different business systems. Existing methods for maintaining state consistency typically rely on a single data source and simple timeout checks to determine the current state of task nodes.

[0003] However, in real-world logistics scenarios, conflicting status descriptions of the same task node from multiple business systems can arise due to data delays, equipment signal loss, or lag in manual operations. Existing methods lack effective multi-source data fusion techniques, cannot quantify the reliability of each data source, and lack the capability for tiered detection and differentiated compensation for deviations of varying severity. This leads to business process interruptions due to inconsistent statuses or erroneous decisions based on incorrect status assessments. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for maintaining the consistency of logistics task execution status in order to solve the above-mentioned problems.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A method for maintaining consistency in the execution state of logistics tasks, comprising: Collect multi-source state data during the execution of logistics tasks, and generate multi-source state evidence vectors for each task node after spatiotemporal alignment processing. The multi-source state evidence vectors are fused using DS evidence theory based on dynamic confidence weights, and the fused state and confidence probability of each task node are output. When the confidence probability is lower than the high confidence threshold, the task node is marked as a state to be detected. The task nodes under test perform spatiotemporal consistency testing, multi-source cross-validation testing, and historical pattern anomaly detection, and output the deviation level based on the comprehensive test results from the three dimensions. Perform differentiated state compensation operation based on the deviation level, and output the compensated task node state; Evaluate the compensation effect and adjust the dynamic credibility weights and deviation grading thresholds.

[0006] Preferably, the determination of the dynamic credibility weight includes the historical accuracy of the data source, current data quality indicators, data timeliness factors, and data source independence factors, wherein the data source... The dynamic credibility weight is: ; in, The total number of data sources participating in the integration; For data source Historical accuracy; For data source Current data quality indicators after normalization; For data timeliness factors; This is the data source independence factor.

[0007] Preferably, the data timeliness factor is defined as follows: ,in The time difference between the data timestamp and the current time. The time-related decay coefficient is defined as follows: The data source independence factor is defined as: ; in For data source With data source The correlation coefficient between historical states, when season .

[0008] Preferably, the spatiotemporal alignment process includes: using the standard timeline after NTP synchronization as a reference, converting the original timestamps of each data source to a unified time coordinate; grouping multiple data records under the same task node ID according to time windows, and combining the source data within the same time window into a multi-source state evidence vector of the node within the corresponding time window; the length of the time window is determined according to the task stage type.

[0009] Preferably, the output deviation level of the comprehensive three-dimensional detection results includes: when there is a single data source delay in the multi-source cross-validation detection and the confidence probability of the fusion state is higher than the medium confidence threshold, it is judged as a slight deviation; when multiple data sources are inconsistent in the multi-source cross-validation detection but the spatiotemporal logic constraints are not violated, it is judged as a moderate deviation; when the spatiotemporal logic constraints are violated, or when the multi-source cross-validation detection and the historical pattern anomaly detection are both judged as anomalies, it is judged as a severe deviation; wherein the medium confidence threshold is lower than the high confidence threshold.

[0010] Preferably, the differentiated status compensation operation based on the deviation level includes: when the deviation level is a slight deviation, sending a fast status confirmation request to the data source terminal with the abnormal status; if a reply is received within a preset time window, updating the node status with the reply content; otherwise, using the data source status with the highest dynamic credibility weight as the compensated status; when the deviation level is a moderate deviation, obtaining the historical data sequence of each data source within the time window before and after the task node, and correcting the current node status based on a linear interpolation algorithm to generate the compensated status; when the deviation level is a severe deviation, sending a status confirmation request to the task executor, the scheduling center, and the customer, determining the arbitration status as the compensated status using the majority consensus principle, and freezing the downstream business processes associated with the task node before the arbitration is completed.

[0011] Preferably, the majority consensus principle means that when at least two of the three parties' responses describe the state in the same way, the consistent state is taken as the arbitration result; if the responses from all three parties are inconsistent, the response from the scheduling center shall prevail and the node shall be marked as manually reviewed.

[0012] Preferably, the evaluation of the compensation effect and adjustment of the dynamic credibility weight and deviation classification threshold includes: tracking the subsequent actual execution results of the task nodes after compensation and calculating the compensation accuracy rate; if the compensation accuracy rate of a certain type of deviation is consistently lower than the preset accuracy rate threshold, then the historical accuracy rate of the data source that caused the compensation failure is reduced to adjust its dynamic credibility weight; if the misjudgment rate of a certain deviation level is too high, then the corresponding deviation classification threshold is adjusted.

[0013] Preferably, the method further includes: storing the verified deviation patterns and their corresponding effective compensation strategies in a knowledge base, wherein the knowledge base is indexed by deviation level, abnormal data source type, and task stage type; when performing compensation operations, matching and searching are performed in the knowledge base based on the characteristics of the current deviation, and if a historical record with a similarity higher than a preset matching threshold is matched, the compensation strategy corresponding to that historical record is invoked and executed.

[0014] A logistics task execution status consistency maintenance system includes: The system includes a multi-source evidence acquisition module, which collects multi-source state data during the execution of logistics tasks and generates multi-source state evidence vectors after spatiotemporal alignment processing; an evidence fusion module, which performs DS evidence theory fusion on the multi-source state evidence vectors based on dynamic credibility weights and outputs the fusion state and credibility probability of each task node; a deviation detection module, which performs spatiotemporal consistency detection, multi-source cross-validation detection, and historical pattern anomaly detection on the task nodes to be detected and outputs the deviation level; a compensation execution module, which performs differentiated state compensation operations according to the deviation level and outputs the compensated task node state; and a self-optimization module, which evaluates the compensation effect and adjusts the dynamic credibility weights and deviation grading thresholds.

[0015] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. This invention assigns dynamic credibility weights to multiple data sources and uses DS evidence theory to fuse them, so that the credibility probability of the fused state can comprehensively reflect the historical accuracy, signal quality, data timeliness and independence of each data source, thereby reducing the probability of misjudgment of the state due to signal loss or data delay of a single data source.

[0016] 2. This invention classifies state deviations from three dimensions: spatiotemporal consistency, multi-source cross-validation, and historical pattern anomalies. This allows for matching compensation operations for minor, moderate, and severe deviations, thus avoiding over-intervention of minor deviations or insufficient compensation for severe deviations under a unified processing strategy. Attached Figure Description

[0017] Further details, features, and advantages of this application are disclosed in the following description of exemplary embodiments in conjunction with the accompanying drawings, in which: Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0018] Several embodiments of this application will now be described in more detail with reference to the accompanying drawings to enable those skilled in the art to implement this application. This application may be embodied in many different forms and for various purposes and should not be limited to the embodiments set forth herein. These embodiments are provided to make this application thorough and complete, and to fully convey the scope of this application to those skilled in the art. The embodiments described do not limit this application.

[0019] Unless otherwise defined, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It will be further understood that terms such as those defined in commonly used dictionaries shall be interpreted as having a meaning consistent with their meaning in the relevant field and / or the context of this specification, and shall not be interpreted in an idealized or overly formal sense unless expressly defined herein.

[0020] Example 1

[0021] Its specific implementation method is combined with the appendix Figure 1 Please provide a detailed explanation.

[0022] In this embodiment, it includes: Logistics tasks involve multiple stages, including warehousing, transportation, delivery, and receipt confirmation, with status data for each stage distributed across different business systems. Existing methods for maintaining state consistency rely on a single data source and simple timeout checks, making it difficult to handle complex scenarios such as data delays between multiple systems, equipment signal loss, and lag in manual operations. When multiple systems provide conflicting state descriptions of the same task node, the lack of effective deviation detection and tiered compensation mechanisms leads to business process interruptions or erroneous decisions due to inconsistent states.

[0023] A logistics task status consistency maintenance method according to an embodiment of the present invention includes the following steps: Step 1: Collect multi-source state data during the execution of logistics tasks, and generate multi-source state evidence vectors after spatiotemporal alignment processing; Throughout the entire logistics task execution process, status data at each stage is acquired through multi-source status acquisition channels. These channels include GPS / BDS positioning data from vehicle-mounted and handheld terminals, operation logs from the warehouse management system, vehicle status data from the transportation management system, check-in and attendance records from delivery personnel's mobile devices, environmental and payload data from IoT sensors, and confirmation feedback data from the client.

[0024] The aforementioned heterogeneous data is spatiotemporally aligned using a unified timestamp and task node ID. The unified timestamp synchronizes the clocks of each data source based on the NTP protocol, and the task node ID associates data from different sources with the same logistics task node. After spatiotemporal alignment, each logistics task node forms a multi-source state evidence vector, which contains the state information reported by each data source at that node and its corresponding timestamp.

[0025] It should be noted that the above spatiotemporal alignment process specifically involves: using the standard timeline synchronized with NTP as a reference, converting the original timestamps of each data source to a unified time coordinate; for multiple data records under the same task node ID, grouping them according to time windows, and combining the source data within the same time window into a multi-source state evidence vector for that node at that moment. The length of the time window is determined based on the task stage type, with shorter time windows used for warehousing stages and longer time windows used for in-transit transportation stages.

[0026] Step 2: Based on the dynamic credibility weight, fuse the multi-source state evidence vectors and output the fused state and credibility probability of each task node; Dynamic credibility weights are assigned to each data source in the multi-source state evidence vector of each task node. The dynamic credibility weights are determined based on the following four factors: the historical accuracy of the data source, the current signal strength or data quality index, the timeliness of the data, and the independence of the data source.

[0027] Data timeliness is quantified by the time deviation between the data timestamp and the current moment; the greater the deviation, the lower the reliability. Data source independence is quantified by the correlation between the data source and other data sources; the higher the correlation, the lower the independence, and the smaller the corresponding weight contribution.

[0028] It should be noted that the above dynamic credibility weight is calculated as follows: Assume the data source... The dynamic credibility weight is ,but ; in, For data source Dynamic credibility weights; The total number of data sources participating in the integration; For data source Historical accuracy, values ​​are taken from the data source. The proportion of reported statuses that match the final confirmed status within the statistical period, with a value range of [value missing]. ; For data source The current signal strength or data quality indicators are reported in real time by the corresponding multi-source status acquisition channels and are evaluated before being used in the product operation. Perform mean normalization based on the range to map it to... scope; The data timeliness factor is defined as follows: ,in This represents the time difference (in seconds) between the data timestamp and the current time. This is the aging decay coefficient, used to control the rate at which aging properties decay with time deviation. The range of values ​​is ; The data source independence factor is defined as follows: ; in For data source With data source The historical correlation coefficient between them, with a value range of 100%. , The total number of data sources participating in the integration. The range of values ​​is All four factors mentioned above are dimensionless proportional quantities, and their multiplication operations are consistent in dimensions. The denominator is the sum of the product terms corresponding to all data sources, used for normalization to make the sum of all dynamic confidence weights equal to 1.

[0029] It should be noted that when hour, If the summation term is empty, then let In other words, the data source independence factor does not participate in the penalty in a single data source scenario.

[0030] The state evidence and its dynamic credibility weights from each data source are input into the DS evidence theory fusion algorithm. The input to the DS evidence theory fusion algorithm is the basic probability assignment function of each data source for the task node state and the corresponding dynamic credibility weights. The output is the fused state of the task node and the credibility probability of the fused state.

[0031] When the confidence probability of the fused state is higher than the preset high confidence threshold, the fused state is directly adopted as the current true state. When the confidence probability of the fused state is lower than the high confidence threshold, the node state is marked as a state to be detected and passed to step 3 for deep detection.

[0032] It should be noted that the above historical accuracy rates The status is updated on a rolling basis with each final confirmation result, reflecting the data source. The percentage of reported statuses that are consistent with the final confirmed status within the past statistical period.

[0033] Step 3: Perform multi-dimensional deviation detection on the task nodes and key state transition moments to be detected, and output the deviation level; For task nodes marked as being in a state to be detected, and for key state transition moments in all nodes, perform deviation detection in the following three dimensions.

[0034] Step 3.1: Perform a spatiotemporal consistency check to determine whether the state change violates the normal spatiotemporal logic of task execution. Extract the time sequence of state changes and the corresponding spatial location sequence of task nodes. Check whether the occurrence time and location of state transition events meet the preset spatiotemporal logical constraints. If the time or location of a state transition event does not conform to the spatiotemporal logical constraints, the spatiotemporal consistency detection dimension is judged as abnormal.

[0035] It should be noted that the aforementioned spatiotemporal logic constraints refer to the state switching conditions preset based on task type and route. For example, the triggering of the "signed for delivery" status requires that the vehicle's location be within the preset range of the destination and that the arrival time be no earlier than the reasonable transportation time.

[0036] Step 3.2: Perform multi-source cross-validation to compare whether the descriptions of the same state from different independent data sources are consistent: The descriptions of the node's current state from each independent data source are extracted from the multi-source state evidence vector. The state descriptions from each data source are compared pairwise, and the number of inconsistent data source pairs is counted. When the number of inconsistent pairs exceeds a preset cross-validation threshold, the multi-source cross-validation detection dimension is judged as an anomaly.

[0037] Step 3.3: Perform historical pattern anomaly detection to identify anomalous transitions in the current state sequence relative to historical normal patterns: Obtain a historical normal execution state sequence of the same type as the current task as a reference pattern. Compare the state sequence of the current task node with the reference pattern to identify abnormal state transitions that deviate from the normal state transition path.

[0038] Based on the combined detection results from steps 3.1 to 3.3, the deviation level is output according to the following rules: If only a single data source has a delay in step 3.2, but the probability of the fusion state is still higher than the medium confidence threshold, it is judged as a slight deviation. If multiple data sources are inconsistent in step 3.2, but the spatiotemporal logic constraints of step 3.1 are not violated, it is judged as a moderate deviation. If the spatiotemporal logic constraints of step 3.1 are violated, or if steps 3.2 and 3.3 are both judged as abnormal, it is judged as a severe deviation.

[0039] It should be noted that the above-mentioned medium confidence threshold is lower than the high confidence threshold in step 2. Together, they constitute a three-level interval division of the confidence probability: when it is higher than the high confidence threshold, the fusion state is directly adopted; when it is between the medium confidence threshold and the high confidence threshold, it enters the deviation detection and is judged at most as a moderate deviation; when it is lower than the medium confidence threshold, the slight deviation judgment condition is not applied.

[0040] Step 4: Perform differentiated state compensation operation according to the deviation level, and output the compensated task node status; Based on the deviation level, the corresponding compensation strategy is triggered.

[0041] When the deviation level is minor, a quick status confirmation request is sent to the data source terminal with the abnormal status. If a response is received within a preset time window, the node status is updated with the response content; if no response is received, the status of the data source with the highest dynamic credibility weight is used as the compensated status.

[0042] When the deviation level is moderate, historical data sequences from each data source within the time window before and after the task node are obtained. After time alignment of each historical data sequence, the current node state is corrected based on a linear interpolation algorithm to generate a compensated state. At the same time, inconsistency records are generated and pushed to the manual review queue.

[0043] When the deviation level is classified as severe, a status confirmation request is simultaneously sent to the task executor, the dispatch center, and the customer. After collecting responses from all three parties, the arbitration status is determined using the majority consensus principle as the post-compensation status. Before arbitration is completed, downstream business processes associated with this task node are frozen; these processes are unfrozen upon arbitration confirmation.

[0044] All compensation operations generate audit logs, recording the deviation type, the compensation strategy executed, the execution time, and the result status.

[0045] It should be noted that the aforementioned majority consensus principle means that if at least two of the three parties' responses describe the status in agreement, that consistent status will be used as the arbitration result. If all three parties' responses are inconsistent, the response from the scheduling center will prevail, and the node will be marked as requiring manual review.

[0046] Step 5: Evaluate the compensation effect and adjust the confidence weight parameters and deviation grading thresholds; After the compensation operation is completed, track the subsequent actual execution results of the task node. Verify whether the compensated state is consistent with the actual execution by using the status data of subsequent nodes.

[0047] Calculate the compensation accuracy rate, which is the proportion of task nodes whose post-compensation state matches the subsequent actual execution result out of all compensated nodes. Also, calculate the time consumed and manual intervention rate for each compensation operation.

[0048] Regularly perform statistical analysis on the compensation accuracy of various types of deviations. If the compensation accuracy of a certain type of deviation consistently falls below a preset accuracy threshold, backtrack and analyze the data sources that caused the compensation failure, and reduce the historical accuracy of those data sources. This allows for the adjustment of the dynamic credibility weight of the data source. If the misjudgment rate for a certain deviation level is too high, then adjust the corresponding deviation classification threshold in step 3.

[0049] In this embodiment, to improve the compensation efficiency for subsequent similar scenarios, verified typical deviation patterns and their corresponding effective compensation strategies are stored in a knowledge base. The knowledge base uses the features of the deviation patterns as indexes, including deviation level, abnormal data source type, and task stage type. When performing compensation in step 4, a matching search is first performed in the knowledge base based on the features of the current deviation. If a historical record with a similarity higher than a preset matching threshold is found, the compensation strategy corresponding to that historical record is directly invoked to reduce compensation response time.

[0050] Example 2

[0051] This implementation overcomes the problem of misjudgment due to signal loss or data delay when judging from a single data source by collecting multi-source state data and fusing DS evidence theory based on dynamic credibility weights. State deviations are graded through three dimensions: spatiotemporal consistency detection, multi-source cross-validation detection, and historical pattern anomaly detection. This ensures that deviations of different severity receive appropriate compensation, avoiding over-intervention for minor deviations or under-compensation for severe deviations under a uniform processing strategy. The dynamic credibility weights and deviation grading thresholds are adjusted through compensation effect evaluation feedback, continuously improving the accuracy of consistency maintenance as operational data accumulates.

[0052] On the other hand, the present invention also proposes a logistics task status consistency maintenance system, including: The multi-source evidence acquisition module is used to collect multi-source state data during the execution of logistics tasks and generate multi-source state evidence vectors after spatiotemporal alignment processing.

[0053] The evidence fusion module is used to fuse multi-source state evidence vectors based on dynamic credibility weights and output the fusion state and credibility probability of each task node.

[0054] The deviation detection module is used to perform multi-dimensional deviation detection and output the deviation level at task nodes and critical state transition times.

[0055] The compensation execution module is used to perform differentiated state compensation operations based on the deviation level and output the compensated task node status.

[0056] The self-optimization module is used to evaluate the compensation effect and adjust the dynamic confidence weight parameters and deviation grading thresholds.

[0057] By way of example, the system proposed in this invention may also include various features and combinations thereof in the method embodiments, which will not be elaborated here.

[0058] A cold chain logistics company undertakes pharmaceutical cold chain distribution services, responsible for delivering vaccine products from its warehouse to various medical institutions. On a certain day in B month of 20XX, a vaccine delivery task with task number CL-20XX-0831 was being executed. This task involves three main stages: warehouse outbound, cold chain transportation, and arrival handover. Data sources include five channels: vehicle-mounted GPS terminals, cold chain IoT sensors, warehouse management system, delivery personnel's handheld terminals, and medical institution client terminals. Due to unstable signal coverage in some sections of the transportation route and human delays in the medical institution's receiving process, conflicting status descriptions of the delivery arrival nodes across multiple systems triggered the full-process processing described in this method.

[0059] For the "arrival and handover" node of task CL-20XX-0831, five acquisition channels report status data respectively. Due to clock drift, the original timestamps of each channel have some deviation; after synchronization via the NTP protocol, they are uniformly converted to standard time coordinates. The task node ID corresponding to this node is NODE-0831-03, and the time window length is set to the longer window corresponding to the transportation stage (120 seconds). After the data from each source within the window are grouped, a multi-source status evidence vector is formed for this node.

[0060] Table 1 Results of Multi-Source State Data Acquisition and Spatiotemporal Alignment

[0061] After spatiotemporal alignment, the multi-source state evidence vector of node NODE-0831-03 contains state information from source 1 to source 5 and a unified timestamp. Source 5 has no data reported within the current time window and is marked as missing.

[0062] Calculate dynamic credibility weights for each of the five data sources. The weights are determined by historical accuracy. Normalized signal quality indicators Timeliness factor and independence factor The product of the four terms is obtained by normalizing. , The time deviation of each source data (unit: seconds). Retrieve the current configuration value; , For source With source The historical correlation coefficient. Source 5 had no data reported, therefore its... After normalization, it becomes 0 and does not actually participate in the fusion.

[0063] Table 2 Calculation parameters for dynamic credibility weight of each data source

[0064] The state evidence from sources 1 to 4, along with their corresponding weights, are input into the DS evidence theory fusion algorithm. Sources 1 and 4 both point to the "arrived" state, source 2 points to the "en route, not yet handed over" state, and source 3 points to the "outbound" state. After fusion, the confidence probability of the "arrived" state is 0.61, which does not exceed the high confidence threshold (set to 0.80). NODE-0831-03 is marked as a state to be detected, and the process proceeds to step 3.

[0065] For node NODE-0831-03, deviation detection is performed sequentially across three dimensions.

[0066] In the spatiotemporal consistency check, the state change time of the node and the vehicle's location coordinates are extracted. Source 1 reports that the vehicle has entered the preset destination range (radius 500 meters), and the arrival time is 14:32:00, which meets the reasonable transportation time constraint (the shortest transportation time is 3 hours and 40 minutes, and the actual duration is 3 hours and 52 minutes). The spatiotemporal logic constraint has not been violated, and the spatiotemporal consistency check is judged to be normal.

[0067] In the multi-source cross-validation detection, the status descriptions of sources 1 to 4 are compared pairwise, resulting in a total of 6 data source pairs. Among them, source 1 and source 4 are consistent (both "arrived at the station"), source 2 and source 3 are consistent (both "handover not completed"), and the following four pairs are inconsistent: source 1 and source 2, source 1 and source 3, source 2 and source 4, and source 3 and source 4. The number of inconsistent pairs is 4, which exceeds the preset cross-validation threshold (threshold is 2), and the multi-source cross-validation detection is judged as abnormal.

[0068] In the historical pattern anomaly detection, the historical normal state sequence of similar pharmaceutical cold chain delivery tasks is obtained. In the historical normal pattern, the typical state transition path of the "arrival handover" node is: outbound confirmation → in-transit transportation → arrival check-in → customer signature. The current sequence lacks the "customer signature" status after "arrival check-in," but there are similar patterns in the historical data where medical institution clients report delayed data. The current deviation does not exceed the historical anomaly jump threshold, and the historical pattern anomaly detection is judged as normal.

[0069] Table 3 Summary of Multidimensional Deviation Detection Results

[0070] Based on the combined results of the three tests: the spatiotemporal logic constraints were not violated, the multi-source cross-validation test showed anomalies, and the historical pattern anomaly detection was normal. According to the deviation grading rules, the output deviation level is classified as moderate.

[0071] The deviation level is moderate, triggering a historical data sequence interpolation compensation strategy. First, a matching search is performed in the knowledge base based on the current deviation characteristics (moderate deviation, multi-source inconsistency, transportation link). A historical record with a similarity of 0.87 exists in the knowledge base (exceeding the matching threshold of 0.75). The effective compensation strategy corresponding to this historical record is: prioritize linear interpolation after time alignment of the vehicle GPS and handheld terminal data sequences to correct the node status.

[0072] Obtain historical data sequences from source 1 (vehicle GPS) and source 4 (handheld terminal) within the time window before and after NODE-0831-03, perform time alignment, then linear interpolation to correct the result to the "arrived at the station, awaiting signature" status. Simultaneously, generate inconsistency records and push them to the manual review queue.

[0073] Table 4 Compensation Operation Execution Record

[0074] The compensation operation synchronously generates an audit log, recording the deviation type as "multi-source inconsistency (moderate)", the compensation strategy as "linear interpolation correction", the execution time as 14:32:38, and the result status as "arrived at the station, awaiting signature".

[0075] During the follow-up execution of mission CL-20XX-0831, at 14:47:22, the medical institution's client reported a confirmation of receipt. The Source 2 cold chain IoT sensor recorded a door opening event, verifying that the post-compensation status "Arrived at the station, awaiting confirmation" was consistent with the actual execution. This compensation was deemed accurate.

[0076] Comparing recent accuracy rates for similar tasks, the delayed reporting from Source 5 (medical institution client) resulted in a moderate bias that led to a consistently higher accuracy rate than the accuracy threshold within the most recent statistical period. Historical accuracy rates for each data source were also analyzed. Maintain the current value. Source 3 (Warehouse Management System) exhibited a lagging status description in this task, and its historical accuracy... The verification result was entered and then updated on a rolling basis, decreasing from 0.79 to 0.77, corresponding to a dynamic credibility weight. The reduction will be correspondingly lowered in subsequent tasks.

[0077] Table 5 Compensation Effect Evaluation and Parameter Update Record

[0078] The typical deviation patterns of this task (moderate deviation, no data on the medical institution's client, and transportation process) and the corresponding effective compensation strategies have been updated and stored in the knowledge base for direct use in subsequent similar scenarios.

[0079] Throughout the processing, data originates from the raw, heterogeneous reports from five acquisition channels. After spatiotemporal alignment and grouping into multi-source state evidence vectors, dynamic credibility weight calculation and DS fusion output insufficiently credible probabilities for the detected states. Subsequently, a three-dimensional deviation detection determines the moderate deviation level, triggering an interpolation compensation strategy to accelerate knowledge base matching. Finally, the verified compensated state is output. The accuracy statistics in step 5 correct the historical accuracy parameters of source 3 in step 2, forming a complete closed-loop data flow. The inputs to each step are derived from the outputs of upstream steps, and parameter updates directly affect the weight calculations of subsequent tasks, ensuring the logical coherence of the data flow.

[0080] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0081] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0082] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for maintaining consistency in the execution status of logistics tasks, characterized in that, include: Collect multi-source state data during the execution of logistics tasks, and generate multi-source state evidence vectors for each task node after spatiotemporal alignment processing. The DS evidence theory is used to fuse the multi-source state evidence vectors based on dynamic credibility weights, and the fusion state and credibility probability of each task node are output. When the confidence probability is lower than the high confidence threshold, the task node is marked as a state to be detected; The task nodes under test perform spatiotemporal consistency testing, multi-source cross-validation testing, and historical pattern anomaly detection, and output the deviation level based on the comprehensive test results from the three dimensions. Perform differentiated state compensation operation based on the deviation level, and output the compensated task node state; Evaluate the compensation effect and adjust the dynamic credibility weights and deviation grading thresholds.

2. The method for maintaining consistency in the execution state of logistics tasks according to claim 1, characterized in that, The determination of dynamic credibility weights is based on factors including the historical accuracy of the data source, current data quality indicators, data timeliness factors, and data source independence factors. The dynamic credibility weight is: ； in, The total number of data sources participating in the integration; For data source Historical accuracy; For data source Current data quality indicators after normalization; For data timeliness factors; This is the data source independence factor.

3. The method for maintaining consistency in the execution state of logistics tasks according to claim 1, characterized in that, Data timeliness factor is defined as ,in The time difference between the data timestamp and the current time. This is the time-related decay coefficient; The data source independence factor is defined as follows: ； in For data source With data source The correlation coefficient between historical states, when season .

4. The method for maintaining consistency in the execution state of logistics tasks according to claim 1, characterized in that, The spatiotemporal alignment process includes: using the standard timeline after NTP synchronization as a reference, converting the original timestamps of each data source to a unified time coordinate; grouping multiple data records under the same task node ID according to time windows, and combining the source data within the same time window into a multi-source state evidence vector of that node within the corresponding time window; the length of the time window is determined according to the task stage type.

5. The method for maintaining consistency in the execution status of logistics tasks according to claim 1, characterized in that, The output deviation level of the detection results across three dimensions includes: a slight deviation when there is a single data source delay in multi-source cross-validation detection and the confidence probability of the fusion state is higher than the medium confidence threshold; a moderate deviation when multiple data sources are inconsistent in multi-source cross-validation detection but the spatiotemporal logic constraints are not violated; and a severe deviation when the spatiotemporal logic constraints are violated, or when both multi-source cross-validation detection and historical pattern anomaly detection are identified as anomalies. The medium confidence threshold is lower than the high confidence threshold.

6. The method for maintaining consistency in the execution state of a logistics task according to claim 1, characterized in that, The differentiated status compensation operation based on the deviation level includes: when the deviation level is minor, a rapid status confirmation request is sent to the data source terminal with the abnormal status. If a reply is received within a preset time window, the node status is updated with the reply content; otherwise, the status of the data source with the highest dynamic credibility weight is used as the compensated status. When the deviation level is moderate, historical data sequences of each data source within the time window before and after the task node are obtained, and the current node status is corrected based on a linear interpolation algorithm to generate the compensated status. When the deviation level is severe, a status confirmation request is sent to the task executor, the scheduling center, and the customer. The arbitration status is determined using the majority consensus principle as the compensated status, and the downstream business processes associated with the task node are frozen before the arbitration is completed.

7. The method for maintaining consistency in the execution state of a logistics task according to claim 6, characterized in that, The majority consensus principle means that if at least two of the three parties' responses describe the status in the same way, the consistent status shall be used as the arbitration result; if the responses from all three parties are inconsistent, the response from the scheduling center shall prevail and the node shall be marked as manually reviewed.

8. The method for maintaining consistency in the execution state of a logistics task according to claim 1, characterized in that, Evaluating the compensation effect and adjusting the dynamic credibility weight and deviation classification threshold includes: tracking the subsequent actual execution results of the task nodes after compensation and calculating the compensation accuracy rate; if the compensation accuracy rate of a certain type of deviation is consistently lower than the preset accuracy rate threshold, the historical accuracy rate of the data source that caused the compensation failure is reduced to adjust its dynamic credibility weight; if the misjudgment rate of a certain deviation level is too high, the corresponding deviation classification threshold is adjusted.

9. The method for maintaining consistency in the execution state of a logistics task according to claim 8, characterized in that, Also includes: The verified deviation patterns and their corresponding effective compensation strategies are stored in a knowledge base, which is indexed by deviation level, abnormal data source type and task stage type. When performing the compensation operation, the knowledge base is matched and searched based on the features of the current deviation. If a historical record with a similarity higher than the preset matching threshold is matched, the compensation strategy corresponding to that historical record is invoked and executed.

10. A system for maintaining consistency in the execution state of a logistics task, and a method for maintaining consistency in the execution state of a logistics task according to any one of claims 1-9, characterized in that, include: The multi-source evidence acquisition module is used to collect multi-source state data during the execution of logistics tasks and generate multi-source state evidence vectors after spatiotemporal alignment processing. The evidence fusion module is used to perform DS evidence theory fusion on the multi-source state evidence vector based on dynamic credibility weights and output the fusion state and credibility probability of each task node. The deviation detection module is used to perform spatiotemporal consistency detection, multi-source cross-validation detection, and historical pattern anomaly detection on the task nodes to be detected, and output the deviation level. The compensation execution module is used to perform differentiated state compensation operations according to the deviation level and output the compensated task node status; the self-optimization module is used to evaluate the compensation effect and adjust the dynamic credibility weight and deviation classification threshold.

Images