A trade chain intelligent risk control processing method fusing multi-dimensional business data

Abstract

The present application belongs to the technical field of trade chain intelligent risk control, and particularly relates to a trade chain intelligent risk control processing method fusing multi-dimensional business data. The method acquires data and establishes a cargo flow transfer node; extracts a stable weighing section from a weight sequence, determines node quality, and generates mass conservation residual error, space-time reachable residual error and bearing residual error; constructs a batch quality mapping matrix with an unallocated quality column according to the measured flow transfer quality in the cargo splitting, merging or reloading process, so as to allocate upstream and downstream node risk quantity; respectively performs weighing rechecking and positioning data rechecking on the quality parameters and space-time parameters, distinguishes equipment collection abnormalities, batch correlation abnormalities and cargo flow transfer abnormalities, and updates node data, batch correlation and downstream risk quantity according to the rechecking or verification results. The present application makes the trade risk judgment correspond to the mass conservation of the physical cargo, the transportation space-time constraint and the equipment measurement state.

Description

Technical Field

[0001] This invention belongs to the technical field of intelligent risk control in the trade chain, specifically relating to an intelligent risk control processing method for the trade chain that integrates multi-dimensional business data. Background Technology

[0002] As global trade becomes increasingly digitalized, platform-based, and network-based, trade activities have evolved from simple commodity transactions between single entities into complex business chains encompassing procurement, production, warehousing, logistics, customs clearance, settlement, financing, and after-sales service. During the operation of these trade chains, multi-dimensional business data is continuously generated, including enterprise information, order information, contract information, invoice information, logistics information, cash flow, customs data, and credit data. By conducting correlation analysis and risk identification on this data, data support can be provided for trade business review, transaction supervision, credit assessment, and risk management.

[0003] Existing trade risk control technologies typically employ methods such as rule review, indicator scoring, or monitoring of single business segments. Rule review primarily relies on pre-defined criteria such as entity qualifications, transaction amount, transaction frequency, geographical scope, and blacklists to match and assess business data. Indicator scoring usually selects indicators such as enterprise credit, performance records, and financial status to build risk assessment models. Monitoring of single business segments identifies anomalies in order, logistics, invoices, settlement, or customs declaration data. Because various data in the trade chain typically originate from different business systems, with variations in data format, update frequency, and entity identification, existing technologies struggle to establish complete and unified data relationships, easily leading to fragmentation of the same trade transaction across different stages. Furthermore, risk assessment methods based on fixed rules or limited indicators have limited ability to identify complex risks such as implicit connections between trading entities, inconsistencies between goods and funds flows, mismatches between contract and invoice content, circular transactions, fictitious transactions, and abnormal performance. In addition, existing risk assessment results often focus on individual enterprises or single transactions, making it difficult to trace the source of risk along the trade chain, identify risk propagation paths, and assess the impact on related entities, resulting in insufficient accuracy, timeliness, and interpretability of risk identification.

[0004] Therefore, it is necessary to establish an intelligent risk control and processing method for the entire business process of the trade chain, so as to improve the comprehensive utilization of multi-dimensional business data and the level of trade risk identification. Summary of the Invention

[0005] To address the above problems, the present invention aims to propose an intelligent risk control method for trade chains that integrates multi-dimensional business data, comprising the following steps: S1. Acquire order data, invoice data, warehousing data, transportation task data, and cargo status data collected by weighing and positioning equipment. Based on the unified cargo batch identifier, transportation task identifier, and collection time, establish cargo flow nodes connected according to the actual flow sequence of the cargo. S2. Extract stable weighing segments from the weight sequence of cargo flow nodes, determine the median weight in the stable weighing segments as the node mass, generate mass conservation residuals based on the node masses of adjacent cargo flow nodes, generate spatiotemporal reachability residuals based on road distance, transportation time and vehicle speed range, generate load-bearing residuals based on measured cargo mass and rated load capacity, and combine the three residuals into a cargo flow residual vector. S3. Construct a batch quality mapping matrix based on the measured flow quality of each batch of goods when the goods are split, merged or repackaged. Set an unallocated quality column in the batch quality mapping matrix and allocate the risk of the upstream goods flow node to the downstream goods flow node according to the matrix elements. S4. Perform weighing verification on cargo flow nodes with mass conservation residuals or load-bearing residuals greater than zero, and perform positioning data verification on cargo flow nodes with spatiotemporal reachability residuals greater than zero. Based on the verification results, determine whether the equipment acquisition is abnormal, the batch association is abnormal, or the cargo flow is abnormal. S5. Generate equipment calibration instructions, batch association correction instructions, or cargo verification instructions based on the anomaly type, and recalculate the risk amount of the corresponding cargo flow node and its downstream cargo flow nodes based on the returned data.

[0006] As a preferred technical solution, step S1 includes: S11. Extract the goods name, goods specifications, batch number, packaging identification, trading entity identification, warehousing node identification, transportation task identification, vehicle identification, and business occurrence time from the order data, invoice data, warehousing data, and transportation task data. S12. Generate a unified cargo batch identifier based on cargo name, cargo specifications, batch number and packaging identification, and establish the inbound data, outbound data, loading data, departure data, arrival data and unloading data with the same unified cargo batch identifier and transportation task identifier as cargo flow nodes respectively; S13. Establish directed connections between adjacent cargo flow nodes according to the time of business occurrence and the order of warehousing, outbound, loading, departure, arrival and unloading.

[0007] As a preferred technical solution, each cargo flow node includes a unified cargo batch identifier, a transportation task identifier, a node type, a node location, a node time, a node quality, a trading entity identifier, and a means of transport identifier; the node type is one of an inbound node, an outbound node, a loading node, a departure node, an arrival node, or an unloading node.

[0008] As a preferred technical solution, the stable weighing section is extracted in the following manner: Arrange the weight values ​​continuously output by the weighing equipment according to the sampling time, and calculate the absolute difference between adjacent weight values; A continuous weight segment with a preset number of consecutive samples and an absolute difference between adjacent weight values ​​not greater than the sum of the weighing device resolution and the no-load fluctuation is identified as a candidate weighing segment. The no-load fluctuation is the difference between the maximum and minimum weight values ​​continuously collected by the weighing device in no-load condition. The candidate weighing segment with the smallest weight range is selected as the stable weighing segment, and the median weight value within the stable weighing segment is determined as the node mass.

[0009] As a preferred technical solution, the mass conservation residual is generated in the following manner: For goods that have not been split or merged, calculate the absolute value of the difference between the upstream outflow quality and the downstream inflow quality; for goods that have been split, calculate the absolute value of the difference between the upstream outflow quality and the sum of the individual downstream inflow qualities; for goods that have been merged, calculate the absolute value of the difference between the sum of the individual upstream outflow qualities and the downstream inflow quality.

[0010] As a preferred technical solution, the allowable natural loss mass of the goods during the actual transportation time and the packaging mass corresponding to the change in packaging quantity are deducted from the obtained difference. When the deduction result is less than zero, it is set to zero, and the deduction result is divided by the corresponding upstream outflow mass to obtain the mass conservation residual.

[0011] As a preferred technical solution, the spatiotemporal reachability residual and bearing capacity residual are generated in the following manner: The shortest reachable time is obtained by dividing the road distance between adjacent cargo transfer nodes by the maximum speed limit of the vehicle; the longest reachable time is obtained by dividing the road distance by the minimum speed limit of the vehicle and adding the allowed dwell time. When the actual transport time is within the range between the shortest and longest reachable time, the spatiotemporal reachability residual is set to zero; when the actual transport time is outside the range, the absolute value of the difference between the actual transport time and the corresponding time boundary is divided by the corresponding time boundary to obtain the spatiotemporal reachability residual.

[0012] As a preferred technical solution, when the measured load mass is not greater than the rated load mass, the load residual is set to zero; when the measured load mass is greater than the rated load mass, the difference between the two is divided by the rated load mass to obtain the load residual. The square root of the sum of the squares of the quality conservation residual, the spatiotemporal reachability residual, and the carrying capacity residual is determined as the local risk quantity of the cargo flow node.

[0013] As a preferred technical solution, step S3 includes: S31. Use the output batches in the upstream cargo flow nodes as matrix rows, and the input batches and unassigned quality columns in the downstream cargo flow nodes as matrix columns; S32. Divide the measured flow quality of any output batch into any input batch by the outflow quality of that output batch to obtain the matrix elements of the corresponding input batch. S33. Subtract the sum of the measured flow quality of each input batch from the outflow quality of the output batch to obtain the unallocated quality, and write the ratio of the unallocated quality to the outflow quality of the output batch into the unallocated quality column. S34. Multiply the risk amount of the upstream cargo transfer node by the corresponding matrix element to obtain the allocated risk amount to each downstream cargo transfer node, and add the allocated risk amount to the local risk amount of the downstream cargo transfer node.

[0014] As a preferred technical solution, when the sum of the measured flow quality of each input batch is greater than the outflow quality of the output batch, the measured flow quality of each input batch is divided by the sum of the measured flow quality of each input batch to redetermine the corresponding matrix elements, and the matrix elements of the unassigned quality column are set to zero.

[0015] As a preferred technical solution, when the unified batch identifier of goods changes after the goods are repackaged, and the goods flow nodes before and after the repackage are located at the same storage node, the time interval between nodes does not exceed the preset repackage duration, and the difference in node quality does not exceed the sum of the allowable measurement errors of the corresponding weighing equipment, a batch association relationship is established between the unified batch identifiers of goods before and after the repackage. Beneficial effects

[0016] This invention extracts the stable weighing segment with the smallest fluctuation range from a continuous weight sequence and determines the node mass using its median. This ensures that the data entering the mass conservation analysis comes from multiple sampling results when the goods are in a relatively stable state. This process reduces the impact of equipment resolution, no-load fluctuations, and instantaneous vibrations on the node mass, allowing the mass differences between adjacent goods flow nodes to more accurately correspond to actual changes in goods flow.

[0017] This invention sets an unallocated quality column in the batch quality mapping matrix, separately retaining the proportion of quality in the output batch that cannot be mapped to downstream input batches, along with its risk amount, instead of allocating this portion of quality to existing input batches through matrix normalization. Therefore, when goods are split, merged, or repackaged, it can retain the abnormal information resulting from unclosed batch quality closures and distribute the risk amount along the actual quality flow direction of each batch.

[0018] This invention processes weighing parameters and positioning parameters using corresponding verification methods: it verifies quality-related residuals using repeated weighing results from the original weighing equipment and the verification weighing equipment, and re-verifies spatiotemporal reachability residuals using the location points and timestamps stored in the positioning equipment. This split verification method avoids using a single data source to explain different types of anomalies and can distinguish between equipment acquisition deviations, batch association errors, and anomalies in the actual flow of goods. Attached Figure Description

[0019] Figure 1 A flowchart illustrating the overall process of intelligent risk control in the trade chain; Figure 2 This is a schematic diagram illustrating the establishment of cargo flow nodes and the generation of residuals; Figure 3 This is a schematic diagram of the batch quality mapping matrix and risk allocation; Figure 4 This is a schematic diagram for weighing and positioning branch verification and anomaly handling. Detailed Implementation

[0020] To enhance understanding of the present invention, the present invention will be further described in detail below with reference to embodiments. These embodiments are only used to explain the present invention and do not constitute a limitation on the scope of protection of the present invention. Example

[0021] This embodiment provides a smart risk control processing method for trade chains that integrates multi-dimensional business data.

[0022] This method is executed by a server, which communicates with the trade business database, the warehouse management system, the weighing equipment, and the positioning equipment.

[0023] The trade business database provides order and invoice data, the warehouse management system provides inbound, outbound and packaging information, the weighing equipment collects the weight sequence of goods during outbound, loading, unloading and inbound processes, and the positioning equipment collects the location points and corresponding timestamps of the vehicles during transportation.

[0024] It should be noted that the specific values ​​below are only used to illustrate the execution process of this method and do not constitute a limitation on the type of goods, transportation distance or equipment parameters.

[0025] This embodiment uses the transportation process of a batch of bagged granular goods as an example. The batch of goods is shipped from the supplier's warehouse and split into batch B1 and batch B2 during transportation. Batch B2 is then repackaged at a transit warehouse to form batch C1; (Refer to...) Figure 1 As shown, the specific steps include: S1. Establish cargo transfer nodes.

[0026] First, extract the following information from order data, invoice data, warehousing data, and transportation task data: goods name, goods specifications, batch number, packaging label, trading entity label, warehousing node label, transportation task label, means of transport label, and transaction time. Then, map the fields representing the same goods from different data sources to a unified goods batch label, using the combination of goods name, goods specifications, batch number, and packaging label.

[0027] For the same unified batch identifier of goods, data generated during its entry into the warehouse, exit from the warehouse, loading, departure, arrival, and unloading are established as goods flow nodes. Each goods flow node records the node type, node location, node time, node quality, trading entity identifier, transportation task identifier, and means of transport identifier. Directed connections are established between adjacent goods flow nodes according to the time of business occurrence and the actual flow sequence of goods.

[0028] For example, the outbound process of the original batch A1 from the supplier's warehouse is established as the first cargo flow node, and the loading process is established as the second cargo flow node; the split batches B1 and B2 are established as the third and fourth cargo flow nodes, respectively; the unloading and warehousing process of batch B1 in the buyer's warehouse is established as the fifth cargo flow node; the unloading and transshipment process of batch B2 in the transit warehouse is established as the sixth cargo flow node, and the transshipped batch C1 is established as the seventh cargo flow node.

[0029] Order and invoice data are used in this step to determine the business correspondence between goods batches and trading entities, but are not directly used to calculate risk levels. Node quality, node location, and node time are determined by the cargo status data collected by weighing and positioning equipment, respectively.

[0030] S2. Generate cargo flow residual vector.

[0031] For each cargo flow node where the node quality needs to be determined, the weight values ​​continuously collected by the weighing equipment are first arranged according to the sampling time, and then the absolute difference between adjacent weight values ​​is calculated. Continuous weight segments with a preset number of consecutive samples, where the absolute difference between any two adjacent weight values ​​is no greater than the sum of the weighing equipment resolution and the no-load fluctuation, are identified as candidate weighing segments.

[0032] The no-load fluctuation is determined by multiple weight values ​​continuously collected by the weighing equipment under no-load conditions, specifically the difference between the maximum and minimum values ​​among these weight values. The weight range of each candidate weighing segment is calculated, and the candidate weighing segment with the smallest weight range is selected as the stable weighing segment. The median of the weight values ​​within this stable weighing segment is determined as the node mass, with reference to... Figure 2 As shown.

[0033] For example, the weighing equipment in the supplier's warehouse continuously collected 20018 kg, 20015 kg, 20014 kg, 20013 kg, 20013 kg, 20012 kg, 20013 kg, 20012 kg, 20012 kg, 20013 kg, 20015 kg, and 20014 kg. The weighing equipment resolution was 1 kg, the no-load fluctuation was 1 kg, and the number of consecutive samples was set to 5. After screening, the continuous weight range consisting of the fourth to tenth weight values ​​had the smallest weight range. Therefore, this weight range was determined as the stable weighing range, and its median, 20013 kg, was determined as the outflow mass of batch A1.

[0034] After obtaining the node mass, the mass conservation residual, spatiotemporal reachability residual, and bearing capacity residual are calculated respectively.

[0035] For goods that have not been split or merged, calculate the absolute value of the difference between the upstream outflow mass and the downstream inflow mass; for goods that have been split, calculate the absolute value of the difference between the upstream outflow mass and the sum of the masses of each downstream inflow; for goods that have been merged, calculate the absolute value of the difference between the sum of the masses of each upstream outflow and the mass of each downstream inflow. Then, subtract the allowable natural loss mass and the packaging mass corresponding to the changes in packaging quantity from the obtained difference. If the subtraction result is less than zero, set it to zero. Then divide the subtraction result by the corresponding upstream outflow mass to obtain the dimensionless mass conservation residual.

[0036] For example, the outflow mass of batch A1 is 20013 kg. The inflow masses of the split batches B1 and B2 are 11982 kg and 7998 kg respectively, totaling 19980 kg, with a mass difference of 33 kg. The actual transportation time is 6 hours, the allowable natural loss per unit time is 0.5 kg / h, and the allowable natural loss mass is 3 kg. The number of packages decreased by 2 before and after splitting, the mass of each package is 1.5 kg, and the change in package mass is 3 kg. Therefore, subtracting 6 kg from 33 kg yields a mass of 27 kg for calculating the mass conservation residual.

[0037] For the spatiotemporal reachability residual, the allowable transportation time is determined based on the road distance between adjacent cargo flow nodes and the speed range of the transport vehicles. The shortest reachable time is obtained by dividing the road distance by the upper speed limit of the transport vehicles; the longest reachable time is obtained by dividing the road distance by the lower speed limit of the transport vehicles and adding the allowable dwell time recorded in the transportation task data. When the actual transportation time falls within this range, the spatiotemporal reachability residual is set to zero; when the actual transportation time falls outside this range, the difference between it and the corresponding time boundary is divided by the corresponding time boundary to obtain the spatiotemporal reachability residual.

[0038] For example, batch B1 has a road distance of 480 km, a maximum vehicle speed of 100 km / h, a minimum speed of 40 km / h, a permitted stop time of 2 hours, and an actual transport time of 8 hours. This actual transport time falls between the shortest and longest reachable times, therefore the spatiotemporal reachability residual is zero.

[0039] For the load-bearing residual, the actual load weight is first determined based on the difference between the loaded vehicle weight and the unloaded vehicle weight. When the actual load weight is not greater than the rated load weight, the load-bearing residual is set to zero; when the actual load weight exceeds the rated load weight, the excess portion is divided by the rated load weight to obtain the load-bearing residual.

[0040] The mass conservation residual, spatiotemporal reachability residual, and carrying capacity residual are combined in a fixed order to form a cargo flow residual vector, and the square root of the sum of the squares of the three residual components is determined as the local risk quantity of the corresponding cargo flow node. Since the three residuals are dimensionless values, they can be combined under a unified dimension, such as... Figure 4 As shown.

[0041] S3. Construct the batch quality mapping matrix and allocate risk quantities.

[0042] When goods are split, merged, or repackaged between adjacent goods transfer nodes, the output batches in the upstream goods transfer nodes are used as the matrix rows, and the input batches in the downstream goods transfer nodes are used as the matrix columns. An unassigned quality column is added after the input batches.

[0043] For any output batch, divide the measured flow quality of any input batch by the output batch's outflow quality, and use the resulting ratio as the corresponding matrix element. Subtract the sum of the measured flow qualities of all input batches from the output batch's outflow quality to obtain the unallocated quality; when the unallocated quality is greater than zero, divide the unallocated quality by the output batch's outflow quality, and write the resulting ratio into the unallocated quality column.

[0044] For example, the outflow mass of batch A1 is 20013 kg, and the measured flow masses entering batches B1 and B2 are 11982 kg and 7998 kg, respectively. Dividing these two measured flow masses by 20013 kg yields the matrix elements corresponding to batches B1 and B2. The remaining 33 kg cannot be mapped to either batch B1 or B2; dividing 33 kg by 20013 kg and using the resulting ratio as a matrix element in the unassigned mass column.

[0045] Multiply the risk amount of each upstream cargo flow node by each matrix element to obtain the allocated risk amount distributed to each downstream cargo flow node and the retained risk amount corresponding to the unallocated quality. For any downstream cargo flow node, add the allocated risk amount it receives to its local risk amount to obtain the risk amount of that node.

[0046] When the sum of the measured flow quality of each input batch exceeds the outflow quality of the output batch, the measured flow quality of each input batch is divided by the sum of the measured flow quality of each input batch to redetermine the matrix elements, and the matrix elements of the unallocated quality column are set to zero. This process only corrects cases where the sum of proportions is greater than 1 due to duplicate associations or duplicate inclusion of multiple warehouse records; it does not forcibly allocate the actual unallocated quality to downstream batches.

[0047] When the batch identifier of goods changes after repackaging, the system determines whether the flow nodes of goods before and after repackaging are located at the same storage node, whether the time interval between nodes does not exceed the preset repackaging time, and whether the difference in node quality does not exceed the sum of the allowable measurement errors of the corresponding weighing equipment. If all the above conditions are met, a batch association relationship is established between the batch identifiers of goods before and after repackaging, and corresponding matrix elements are constructed accordingly.

[0048] S4. Perform verification of the repetition check and the location data.

[0049] Quality parameters and spatiotemporal parameters employ different verification paths. For cargo flow nodes with quality conservation residuals or load-bearing residuals greater than zero, the original weighing equipment and the verification weighing equipment are controlled to perform at least two weighings on the same cargo. For cargo flow nodes with spatiotemporal reachability residuals greater than zero, the location points and corresponding timestamps stored in the positioning equipment are read, and the location sequences are filtered.

[0050] In the weighing verification process, the maximum difference between the weighing results of the original weighing equipment and the maximum difference between the weighing results of the verification weighing equipment are calculated respectively. If the maximum difference of the original weighing equipment is greater than its allowable measurement error, while the maximum difference of the verification weighing equipment is not greater than its allowable measurement error, the original weighing equipment is determined to have an abnormal acquisition.

[0051] When the difference between the node masses determined by two weighing devices is no greater than the sum of their allowable measurement errors, and the matrix elements in the unassigned mass column are greater than zero, it indicates that the mass acquisition results can be mutually verified, but there are still unassigned masses between batches, thus identifying a batch correlation anomaly. When the node masses determined by the two weighing devices are consistent, and the mass conservation residuals calculated separately are both greater than zero, it is identified as a cargo flow anomaly.

[0052] For example, the original weighing equipment yielded two weighing results of 20011 kg and 20012 kg, while the verification weighing equipment yielded two weighing results of 20012 kg. The difference in node mass obtained by the two equipment is within the allowable measurement error range, but the matrix elements of the unassigned mass column are still greater than zero. Therefore, this node is identified as a batch correlation anomaly.

[0053] In the location data verification process, the ratio of road distance to time interval between adjacent location points is calculated sequentially. If the resulting speed exceeds the vehicle's speed limit, the corresponding next location point is identified as an abnormal location point and deleted. The remaining location points are used to re-determine the road distance and actual transportation time, and the spatiotemporal reachability residual is recalculated. If the recalculated spatiotemporal reachability residual is zero, it is determined that the location equipment data collection is abnormal; if it is still greater than zero after recalculation, it is determined that the cargo flow is abnormal.

[0054] For example, in the original location sequence of batch B2, there is a location jump point where the calculated speed between this jump point and the previous location point exceeds the vehicle speed limit. After deleting this location point, the recalculated actual transportation time falls within the allowable transportation time range; therefore, this node is identified as an anomaly in the positioning device's data collection.

[0055] S5. Perform exception handling and update downstream risk levels.

[0056] When it is determined that the weighing equipment is abnormal, a calibration command is sent to the original weighing equipment, and the original node quality is replaced with the node quality determined by the verification weighing equipment; when it is determined that the positioning equipment is abnormal, the abnormal location point is deleted, and a positioning data supplementation command including the transportation task identifier, sampling interval and sampling duration is sent to the positioning equipment.

[0057] When a batch association anomaly is identified, the association between the output batch and the input batch is re-verified based on the unified batch identifier, node location, node time, node quality, and packaging identifier of adjacent cargo flow nodes. For example, if it is found that an inbound record originally assigned to batch B2 actually belongs to another transportation task, the record is removed from batch B2, and the batch quality mapping matrix is ​​reconstructed.

[0058] When an anomaly in cargo flow is identified, a cargo verification instruction is generated. This instruction includes a unified cargo batch identifier, the target warehousing node, the weight range of the cargo to be verified, packaging identification, and the cargo storage location. Warehouse verification equipment or warehouse operators re-weigh the target batch according to the instruction and verify the packaging quantity and batch identifier.

[0059] Based on the verified weighing results, the selected location points, the corrected batch association, or the actual quality obtained from cargo verification, the quality conservation residual, spatiotemporal reachability residual, carrying capacity residual, and local risk quantity of the corresponding cargo flow node are regenerated. Then, according to the reconstructed batch quality mapping matrix, the allocated risk quantity and risk quantity of downstream cargo flow nodes are updated sequentially along the actual cargo flow direction until no downstream cargo flow node is associated with the corresponding unified cargo batch identifier. Figure 3 As shown.

[0060] In this embodiment, each step first processes the physical cargo status based on weighing stability, mass conservation principles, spatiotemporal constraints during transportation, and the carrying capacity of the transport vehicle, and then uses specific data to illustrate the processing results. Trade business data is used to establish the correspondence between cargo batches and transportation tasks, while risk identification and anomaly verification are completed based on the physical data collected by weighing and positioning equipment.

[0061] 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 illustrative of the 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 present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A smart risk control processing method for trade chains that integrates multi-dimensional business data, characterized in that, Includes the following steps: S1. Acquire order data, invoice data, warehousing data, transportation task data, and cargo status data collected by weighing and positioning equipment. Based on the unified cargo batch identifier, transportation task identifier, and collection time, establish cargo flow nodes connected according to the actual flow sequence of the cargo. S2. Extract stable weighing segments from the weight sequence of cargo flow nodes, determine the median weight in the stable weighing segments as the node mass, generate mass conservation residuals based on the node masses of adjacent cargo flow nodes, generate spatiotemporal reachability residuals based on road distance, transportation time and vehicle speed range, generate load-bearing residuals based on measured cargo mass and rated load capacity, and combine the three residuals into a cargo flow residual vector. The stable weighing section is extracted in the following manner: Arrange the weight values ​​continuously output by the weighing equipment according to the sampling time, and calculate the absolute difference between adjacent weight values; A continuous weight segment with a preset number of consecutive samples and an absolute difference between adjacent weight values ​​not greater than the sum of the weighing device resolution and the no-load fluctuation is identified as a candidate weighing segment. The no-load fluctuation is the difference between the maximum and minimum weight values ​​continuously collected by the weighing device in no-load condition. The candidate weighing segment with the smallest weight range is selected as the stable weighing segment, and the median weight value within the stable weighing segment is determined as the node mass. S3. Construct a batch quality mapping matrix based on the measured flow quality of each batch of goods when the goods are split, merged or repackaged. Set an unallocated quality column in the batch quality mapping matrix and allocate the risk of the upstream goods flow node to the downstream goods flow node according to the matrix elements. S4. Perform weighing verification on cargo flow nodes with mass conservation residuals or load-bearing residuals greater than zero, and perform positioning data verification on cargo flow nodes with spatiotemporal reachability residuals greater than zero. Based on the verification results, determine whether the equipment acquisition is abnormal, the batch association is abnormal, or the cargo flow is abnormal. S5. Generate equipment calibration instructions, batch association correction instructions, or cargo verification instructions based on the anomaly type, and recalculate the risk amount of the corresponding cargo flow node and its downstream cargo flow nodes based on the returned data.

2. The intelligent risk control processing method for trade chains that integrates multi-dimensional business data according to claim 1, characterized in that, Step S1 includes: S11. Extract the goods name, goods specifications, batch number, packaging identification, trading entity identification, warehousing node identification, transportation task identification, vehicle identification, and business occurrence time from the order data, invoice data, warehousing data, and transportation task data. S12. Generate a unified cargo batch identifier based on cargo name, cargo specifications, batch number and packaging identification, and establish the inbound data, outbound data, loading data, departure data, arrival data and unloading data with the same unified cargo batch identifier and transportation task identifier as cargo flow nodes respectively; S13. Establish directed connections between adjacent cargo flow nodes according to the time of business occurrence and the order of warehousing, outbound, loading, departure, arrival and unloading.

3. The intelligent risk control processing method for trade chains that integrates multi-dimensional business data according to claim 2, characterized in that, Each cargo flow node includes a unified cargo batch identifier, a transportation task identifier, a node type, a node location, a node time, a node quality, a trading entity identifier, and a means of transport identifier; the node type is one of the following: inbound node, outbound node, loading node, departure node, arrival node, or unloading node.

4. The intelligent risk control processing method for trade chains that integrates multi-dimensional business data according to claim 3, characterized in that, The mass conservation residual is generated in the following manner: For goods that have not been split or merged, calculate the absolute value of the difference between the upstream outflow quality and the downstream inflow quality; for goods that have been split, calculate the absolute value of the difference between the upstream outflow quality and the sum of the individual downstream inflow qualities; for goods that have been merged, calculate the absolute value of the difference between the sum of the individual upstream outflow qualities and the downstream inflow quality. Subtract the allowable natural loss mass of the goods during the actual transportation time and the packaging mass corresponding to the change in packaging quantity from the obtained difference. If the subtraction result is less than zero, set it to zero and divide the subtraction result by the corresponding upstream outflow mass to obtain the mass conservation residual.

5. The intelligent risk control processing method for trade chains that integrates multi-dimensional business data according to claim 1, characterized in that, The spatiotemporal reachability residual and bearing capacity residual are generated in the following manner: The shortest reachable time is obtained by dividing the road distance between adjacent cargo transfer nodes by the maximum speed limit of the vehicle; the longest reachable time is obtained by dividing the road distance by the minimum speed limit of the vehicle and adding the allowed dwell time. When the actual transport time is within the range between the shortest and longest reachable time, the spatiotemporal reachability residual is set to zero; when the actual transport time is outside the range, the absolute value of the difference between the actual transport time and the corresponding time boundary is divided by the corresponding time boundary to obtain the spatiotemporal reachability residual.

6. The intelligent risk control processing method for trade chains that integrates multi-dimensional business data according to claim 5, characterized in that, When the measured load mass is not greater than the rated load mass, the load residual is set to zero; when the measured load mass is greater than the rated load mass, the difference between the two is divided by the rated load mass to obtain the load residual. The square root of the sum of the squares of the quality conservation residual, the spatiotemporal reachability residual, and the carrying capacity residual is determined as the local risk quantity of the cargo flow node.

7. The intelligent risk control processing method for trade chains that integrates multi-dimensional business data according to claim 1, characterized in that, Step S3 includes: S31. Use the output batches in the upstream cargo flow nodes as matrix rows, and the input batches and unassigned quality columns in the downstream cargo flow nodes as matrix columns; S32. Divide the measured flow quality of any output batch into any input batch by the outflow quality of that output batch to obtain the matrix elements of the corresponding input batch. S33. Subtract the sum of the measured flow quality of each input batch from the outflow quality of the output batch to obtain the unallocated quality, and write the ratio of the unallocated quality to the outflow quality of the output batch into the unallocated quality column. S34. Multiply the risk amount of the upstream cargo transfer node by the corresponding matrix element to obtain the allocated risk amount to each downstream cargo transfer node, and add the allocated risk amount to the local risk amount of the downstream cargo transfer node.

8. The intelligent risk control processing method for trade chains that integrates multi-dimensional business data according to claim 7, characterized in that, When the sum of the measured flow quality of each input batch is greater than the outflow quality of the output batch, the measured flow quality of each input batch is divided by the sum of the measured flow quality of each input batch to redetermine the corresponding matrix elements, and the matrix elements of the unassigned quality column are set to zero.

9. The intelligent risk control processing method for trade chains that integrates multi-dimensional business data according to claim 8, characterized in that, When the unified batch identifier of goods changes after the goods are repackaged, and the goods flow nodes before and after the repackage are located at the same storage node, the time interval between nodes does not exceed the preset repackage duration, and the difference in node quality does not exceed the sum of the allowable measurement errors of the corresponding weighing equipment, a batch association relationship is established between the unified batch identifiers of goods before and after the repackage.

Images