Raw material tracing management system for mixed fermentation of tremella polysaccharide and tea extract

Abstract

The invention discloses a raw material traceability management system for mixed fermentation of tremella polysaccharide and tea extract, and particularly relates to the technical field of raw material traceability management, comprising: acquiring raw material unit data, process unit data and result unit data, according to unit numbers and occurrence moments in the data, performing corresponding association and time sorting on the tremella polysaccharide raw material batch, the tea extract raw material batch, the mixed fermentation process node and the finished product batch, and outputting a traceability unit set; according to the method, raw material batches, process nodes and finished product batches are constructed into knowledge graph nodes, and a continuous traceability chain is formed along source numbers, undertaking numbers and occurrence moments, so that abnormal results can correspond to actual source combinations and propagation positions; therefore, the problem that the responsibility boundary is difficult to converge under the conditions of multi-batch cross flow and abnormal delayed display is relatively improved.

Description

Technical Field

[0001] This invention relates to the field of raw material traceability management technology, and more specifically, to a raw material traceability management system for the mixed fermentation of Tremella polysaccharide and tea extract. Background Technology

[0002] In the raw material traceability management of the mixed fermentation product of tremella polysaccharide and tea extract, the existing technology mainly focuses on how to quickly locate the source of the raw materials involved and delineate the scope of disposal when the finished product batch shows abnormal stability, flavor deviation or functional index fluctuation. In actual application, it mostly relies on the correspondence between purchase documents, arrival inspection records, warehousing batch records, material requisition and feeding records and production batch numbers, and checks back step by step according to the finished product batch, or directly takes measures such as expanding freezing, supplementary re-inspection and manual investigation to deal with the related raw materials and downstream batches after suspicious situations occur. Taking the production of fermented beverages or compound fermented foods as an example, tremella polysaccharides and tea extracts often come from different suppliers and enter the same production chain multiple times during pretreatment, temporary storage, replenishment, continuous feeding, tank replacement or line switching. Enterprises usually also face realities such as continuous supply without long-term line stoppage, limited re-inspection capabilities, parallel flow of incoming batches, and high requirements for disposal timeliness. In this situation, traditional methods often lead to the following phenomenon: raw materials from the same source show no obvious abnormalities in several previous batches of finished products, but problems are concentrated in subsequent batches. During the tracing process, several seemingly related batches can often be identified, but it is difficult to further distinguish which combination of sources, at which stage of feeding or storage, caused the actual impact. As a result, the scope of investigation is repeatedly expanded, the number of frozen objects continues to increase, and even after review, it is still impossible to converge to a clear and minimal boundary of responsibility. The crux of the problem is that the existing methods are more based on static batch correspondence for registration and back-checking, and lack management means to comprehensively characterize and identify the combination relationship, transmission relationship, and abnormal formation conditions of raw materials from multiple sources in the mixed fermentation process. The technical problem this application aims to solve is: how to provide a traceability management system for the mixed fermentation raw materials of Tremella polysaccharide and tea extract that incorporates knowledge graphs, so that it can accurately identify the actual source combination corresponding to the abnormal results under the circumstances of multiple batch cross-transfers, multiple replenishment and continuous feeding, and delayed appearance of anomalies, and determine the true scope of impact accordingly. Summary of the Invention

[0003] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a raw material traceability management system for the mixed fermentation of Tremella polysaccharide and tea extract. This system constructs a knowledge graph of raw material batches, process nodes, and finished product batches, and combines it with source number, receiving number, occurrence time, container receiving relationship, and test results to perform reverse tracing, continuous verification, and anomaly interception of the source combination path, thereby solving the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a raw material traceability management system for the mixed fermentation of Tremella polysaccharide and tea extract, comprising: The data consolidation module acquires raw material unit data, process unit data, and result unit data. It performs corresponding association and time sorting on the batches of Tremella polysaccharide raw materials, tea extract raw materials, mixed fermentation process nodes, and finished product batches according to the unit number and occurrence time in each data, and outputs a traceability unit set. The graph construction module, based on the traceability unit set, establishes raw material batches, process nodes, and finished product batches as knowledge graph nodes, and establishes knowledge graph edges according to the correspondence between raw materials entering the process, process transfer process, and process forming finished product, and outputs the traceability knowledge graph; The path tracing module, based on the source tracing knowledge graph, searches backward level by level along the edge of the knowledge graph, starting from the finished product batch. The raw material batches and process nodes connected to the finished product batch are arranged in the order of occurrence time to obtain the source combination path corresponding to the finished product batch, and output the source combination path set. The boundary determination module, based on the source combination path set, performs item-by-item comparison of the occurrence time and acceptance number of the adjacent nodes in each source combination path, retains the source combination paths with the occurrence time progressively and the corresponding acceptance numbers, and extracts the retained source combination paths before the first abnormal result of the corresponding finished product batch, and outputs the abnormal source combination and its impact boundary. The results control module extracts the corresponding raw material batches, process nodes, and finished product batches based on the combination of anomaly sources and impact boundaries, generates frozen objects, review objects, and recall objects, and outputs the raw material traceability management results.

[0005] In a preferred embodiment, the data consolidation module includes: Acquire raw material unit data, process unit data, and result unit data. Extract unit number, source number, receiving number, and occurrence time from each data. Generate Tremella polysaccharide raw material unit, tea extract raw material unit, process unit, and result unit according to the unit number, and output the unit detail set. Based on the unit detail set, the source number and occurrence time of each process unit are read sequentially. The source number is compared with the successor number of each Tremella polysaccharide raw material unit, each tea extract raw material unit, and the preceding process unit. The Tremella polysaccharide raw material unit, tea extract raw material unit, and preceding process unit that match the comparison and whose occurrence time is earlier than the occurrence time of the process unit are connected to the process unit, and the unit connection set is output. Based on the unit connection set, the source number and occurrence time of each result unit are read sequentially. The source number is compared with the successor number of each process unit item by item. Process units that match the source number and whose occurrence time is earlier than the occurrence time of the result unit are connected to the result unit. Then, the connected units are sorted from earliest to latest according to their occurrence time, and the source unit set is output.

[0006] In a preferred embodiment, the map construction module includes: Obtain the traceability unit set, read the unit number, source number, acceptance number and occurrence time of each raw material batch, each process node and each finished product batch, and establish raw material node, process node and finished product node using the unit number of each raw material batch, process node and finished product batch as node identifier, and output the graph node set; Based on the graph node set, the source number and acceptance number of each process node are read sequentially. The source number of each process node is compared with the acceptance number of each raw material node, and the raw material nodes with the same comparison are connected to the corresponding process node. Then, the source number of each process node is compared with the acceptance number of the preceding process node, and the preceding process nodes with the same comparison are connected to the corresponding process node. The process association edge set is output. Based on the process association edge set, the source number of each finished product node is read sequentially. The source number of each finished product node is compared with the acceptance number of each process node one by one. Process nodes with the same comparison are connected to the corresponding finished product node. Then, the graph node set, process association edge set and connection relationship from process node to finished product node are merged to output the traceability knowledge graph.

[0007] In a preferred embodiment, the path tracing module includes: Obtain the traceability knowledge graph, read the node number of each finished product node, and generate a reverse retrieval sequence one by one with the node number of each finished product node as the retrieval starting point, and output the starting point sequence set; Based on the starting sequence set, read the process nodes and raw material nodes directly connected to the current node level by level along the incoming edges of each finished product node. The read process nodes and raw material nodes are appended to the end of the corresponding reverse search sequence in sequence. The same reading is performed along the incoming edges for the appended process nodes until no new nodes appear in the appended nodes. The connected node sequence set is then output.

[0008] In a preferred embodiment, the path tracing module further includes: Based on the set of connected node sequences, the occurrence time of each process node and each raw material node in each connected node sequence is extracted, and the process nodes and each raw material node are rearranged from earliest to latest according to the occurrence time, and the sorted node sequence set is output. Based on the sorted node sequence set, the raw material nodes and process nodes corresponding to the same finished product node are connected end to end in the sorted order to generate the source combination path corresponding to the finished product node. The source combination paths corresponding to each finished product node are summarized and the source combination path set is output.

[0009] In a preferred embodiment, the boundary determination module includes: Obtain the source combination path set and detection records, extract the node number, source number, successor number, occurrence time and container number of each node in each source combination path in sequence, and generate path node pairs in the order of the previous node and the next node in the same path, and output the path node pair set. Based on the path node pair set, the receiving number and occurrence time of the previous node and the source number and occurrence time of the next node in each path node pair are read sequentially. The source number of the next node is compared with the receiving number of the previous node item by item, and the occurrence time of the next node is compared with the occurrence time of the previous node item by item. Path node pairs with the same number and the occurrence time of the next node is later than the occurrence time of the previous node are written into the first pass flag, and the remaining path node pairs are written into the first block flag. The first comparison result set is output.

[0010] In a preferred embodiment, the boundary determination module further includes: Based on the first comparison result set, the path node pairs marked with the first pass are collected and written according to each source combination path. The node number of the same node in the middle of two adjacent path node pairs, the successor number of the previous node, the source number and successor number of the same node, and the source number of the next node are read in sequence. The same comparison is performed on the node number of the same node in the two path node pairs, the same comparison is performed on the successor number of the previous node and the source number of the same node, and the same comparison is performed on the successor number of the same node and the source number of the next node. The source combination path with the same three comparisons is written with the through mark, and the source combination path with any one of the comparisons is written with the back mark. The path through result set is output. Based on the path connectivity result set, the container numbers of each node in each source combination path written with the connectivity mark are read sequentially. The container number of the previous node is compared with the container number of the next node item by item. For path node pairs with the same container number, the container connectivity mark is written. For path node pairs with different container numbers, the outgoing container number and the incoming container number in the transfer record are read. The container number of the previous node is compared with the outgoing container number item by item, and the container number of the next node is compared with the incoming container number item by item. For path node pairs with the same two comparisons, the container connectivity mark is written. For the remaining path node pairs, the container blocking mark is written. The container comparison result set is output.

[0011] In a preferred embodiment, the boundary determination module further includes: Based on the container comparison result set, the source combination path written to the container connection mark is read for each finished product batch, and the detection records corresponding to the finished product batch are arranged from earliest to latest according to the occurrence time. The detection results in each detection record are read one by one, and the occurrence time of the first detection record with an abnormal detection result is determined as the abnormal cutoff point. Then, the nodes in each source combination path whose occurrence time is later than the abnormal cutoff point and their subsequent path nodes are deleted, and the candidate abnormal path set is output. Based on the candidate abnormal path set, the source combination paths that are retained after deletion are collected separately for each finished product batch. The batches of Tremella polysaccharide raw materials, tea extract raw materials, and process nodes in each source combination path are extracted to form an abnormal source combination. The process node with the latest occurrence time in each abnormal source combination is combined with the acceptance number, container number, and finished product batch number as the influence boundary, and the abnormal source combination and influence boundary are output.

[0012] In a preferred embodiment, the result control module includes: Obtain the combination of abnormal sources and their impact boundaries. Extract the raw material batches of Tremella polysaccharide, raw material batches of tea extract, process nodes, and finished product batches from each abnormal source combination. Write the raw material batches, process nodes, and finished product batches from each abnormal source combination into the object list according to the combination attribution relationship, and output the control object set. Based on the control object set, the occurrence count of each raw material batch in each abnormal source combination, the number of connections between each process node and each finished product batch, and the corresponding number between each finished product batch and each influence boundary are read sequentially. Raw material batches with more than one occurrence count and process nodes with more than one connection count are written into the freeze object, and raw material batches and process nodes with a corresponding count of one are written into the review object. The process control set is then output.

[0013] In a preferred embodiment, the result control module further includes: Based on the process control set, the batch number of each finished product batch and the acceptance number in the corresponding impact boundary are read sequentially. Finished product batches with the same batch number and the acceptance number located in the same impact boundary are merged into recall units. Each recall unit connected to the frozen object or the review object is written into the recall object, and the result control set is output. Based on the process control set and the result control set, the frozen objects, the review objects and the recalled objects are summarized. The raw material batch number and process node number corresponding to each frozen object, the raw material batch number and process node number corresponding to each review object, and the finished product batch number and impact boundary corresponding to each recalled object are written into the raw material traceability management result, and the raw material traceability management result is output.

[0014] The technical effects and advantages of this invention are as follows: 1. By constructing raw material batches, process nodes, and finished product batches as knowledge graph nodes, and forming a continuous traceability chain along the source number, receiving number, and occurrence time, abnormal results can be correlated with the actual source combination and propagation location, thereby relatively improving the problem of difficulty in converging the responsibility boundary under the conditions of multiple batch cross-flow and delayed manifestation of abnormalities; 2. By sequentially performing number connection comparison, time sequence progression comparison, and container acceptance comparison on adjacent nodes in the source combination path, path segments that only have superficial associations but no actual transmission relationship can be gradually eliminated, thereby relatively reducing the expansion of the scope of false associations caused by static batch back-checking. 3. By aligning the detection records with the source combination path according to the finished product batch and the time of occurrence, and using the time corresponding to the first abnormal detection record as the abnormal cutoff point, subsequent paths that are later than the time of abnormal formation can be deleted from the traceability scope, thereby relatively improving the consistency between the abnormal source location and the actual formation process. 4. By counting the number of times the raw material batches in the abnormal source combination are executed, and counting the number of times the process node is executed to the finished product batch, it is possible to distinguish between objects whose influence has not yet converged and objects whose influence has initially converged, thus providing an executable basis for object division for freezing and review. 5. By merging finished product batches into recall units according to batch number, acceptance number, and container number, and writing recall units that are continuously connected to frozen or reviewed objects into the recall objects, the recall scope can be kept in line with the boundary of abnormal propagation, thereby relatively improving the problems of the expansion of recall objects and the repeated increase of disposal objects. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the system modules of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Refer to the instruction manual appendix Figure 1 The present invention provides a raw material traceability management system for the mixed fermentation of Tremella polysaccharide and tea extract, comprising: The data consolidation module acquires raw material unit data, process unit data, and result unit data. It performs corresponding association and time sorting on the batches of Tremella polysaccharide raw materials, tea extract raw materials, mixed fermentation process nodes, and finished product batches according to the unit number and occurrence time in each data, and outputs a traceability unit set. In this embodiment, the data consolidation module organizes raw material records, process records, and result records into traceable units that can be linked sequentially, so that the correspondence between raw materials, processes, and results can be established according to the same set of numbers and time sequence. Specifically, the unit number, source number, successor number, and occurrence time are first extracted from the original records, and then the raw material units for Tremella polysaccharide, raw material units for tea extract, process units, and result units are generated respectively. Subsequently, the raw material units and preceding process units are first connected to the corresponding process units, then the process units are connected to the corresponding result units, and finally all connected units corresponding to the same result unit are linked according to... After sorting by occurrence time, the data is written into the traceability unit set. Here, the source number represents the upstream unit number referenced when the current unit receives materials, the receiving number represents the unit number referenced by the current unit when outputting to downstream units, and the occurrence time represents the actual completion time of the corresponding business action. When the same record simultaneously has business completion time, equipment completion time, system accounting time, and manual entry time, the business completion time is read first. If the business completion time is empty, the equipment completion time is read next. If the equipment completion time is empty, the system accounting time is read next. If the system accounting time is empty, the manual entry time is read next. This implementation process includes the following steps: First, after acquiring raw material unit data, process unit data, and result unit data, the fields of the three types of data are processed. Raw material unit data includes records of Tremella polysaccharide raw materials and tea extract raw materials; process unit data includes feeding records, replenishment records, temporary storage entry records, temporary storage transfer records, transfer records, and fermentation treatment records; result unit data includes finished product formation records and result records corresponding to finished product batches. Subsequently, the unit number, source number, receiving number, and occurrence time are extracted from each record. When a record already has a unique unit number, it is directly used as the unit number; when a record does not have a unit number but has a unique business serial number, it is used as the unit number; when a record has an upstream reference field, it is used as the source number; when raw materials enter the system for the first time and there is no upstream reference field, the source number is recorded as a null value. If a receiving number already exists in the record, it is directly retained. If no receiving number exists in the record, the current unit number is used as the receiving number. After the field extraction is completed, each record of Tremella polysaccharide raw material is generated into a Tremella polysaccharide raw material unit according to the unit number. Each record of tea extract raw material is generated into a tea extract raw material unit according to the unit number. Each feeding record, replenishment record, temporary storage entry record, temporary storage transfer out record, transfer record, and fermentation treatment record is generated into a process unit according to the unit number. Each finished product formation record and each result record that corresponds one-to-one with the finished product batch are generated into a result unit according to the unit number. When the same unit number corresponds to multiple records, the four fields of unit number, source number, receiving number, and occurrence time of each record are compared first. Records with all four fields not empty are retained. If multiple records still exist, the record with the earliest occurrence time is retained. The remaining records are written into the exception registration information and do not participate in the generation of the current unit. This is how the unit detail set is obtained. Secondly, based on the unit detail set, the source number and occurrence time of each process unit are read sequentially, and the currently read process unit is taken as the target process unit. Subsequently, the successor number and occurrence time of all Tremella polysaccharide raw material units, all tea extract raw material units, and all preceding process units are read item by item, where the preceding process unit is the process unit whose occurrence time is earlier than the occurrence time of the target process unit. For each candidate unit, the successor number of the candidate unit is first compared with the source number of the target process unit item by item, and then the occurrence time of the candidate unit is compared with the occurrence time of the target process unit item by item. When the successor number of the candidate unit is the same as the source number of the target process unit, and the occurrence time of the candidate unit is earlier than the occurrence time of the target process unit, the candidate unit is connected to the target process unit, and a connection record is written. The connection record contains the starting unit number, the target unit number, the connection type, and the connection order. When the starting unit is a Tremella polysaccharide raw material unit or a tea extract raw material unit, the connection type is written as raw material to process connection. When the starting unit is a preceding process unit, the connection type is written as process to process connection. The connection order is written sequentially from the beginning to the end according to the occurrence time of the starting unit. If the source number of the target process unit is the same as the acceptance number of multiple candidate units, then all candidate units that meet the conditions are connected to the target process unit respectively; if the acceptance number of the same candidate unit is the same as the source number of multiple subsequent process units, then the candidate unit is connected to each of the subsequent process units respectively; for candidate units with the same acceptance number but whose occurrence time is later than that of the target process unit, no connection is established; for candidate units with the same acceptance number and the same occurrence time, the system entry time in the corresponding original record is read, and the candidate unit corresponding to the record with the earlier system entry time is connected to the target process unit first. If the system entry times are still the same, connections are established in ascending order of unit number; after completing the above processing, the unit connection set from raw material unit to process unit and between process units is obtained; Based on the unit connection set, the source number and occurrence time of each result unit are read sequentially, and the currently read result unit is taken as the target result unit. Then, the successor number and occurrence time of all process units are read item by item. The successor number of each process unit is compared with the source number of the target result unit, and the occurrence time of each process unit is compared with the occurrence time of the target result unit. When the successor number of a process unit is the same as the source number of the target result unit, and the occurrence time of the process unit is earlier than the occurrence time of the target result unit, the process unit is connected to the target result unit, and a process-to-result connection record is written. If the target result unit corresponds to multiple process units that meet the conditions, all process units that meet the conditions are connected to the target result unit respectively. This completes the connection between result units and process units. Next, all connected units corresponding to each target result unit are sorted. The sorting objects include the Tremella polysaccharide raw material unit, tea extract raw material unit, process unit, and the target result unit itself, which are connected to the target result unit. During sorting, they are first arranged from earliest to latest according to the time of occurrence. If the time of occurrence is different, the unit with the earlier time of occurrence is placed first. If the time of occurrence is the same, the Tremella polysaccharide raw material unit and the tea extract raw material unit are arranged first, then the process unit is arranged, and finally the result unit is arranged. If the time of occurrence of the same type of unit is still the same, they are arranged from smallest to largest according to the unit number. This sorting only changes the writing order of each connected unit in the traceability unit set and does not change the connection records that have been established. Finally, all connected units corresponding to each target result unit and their sorting results are written as a traceability unit entry into the traceability unit set. Through the above processing, the data consolidation module first organizes the original records into raw material units, process units, and result units with consistent fields. Then, it establishes the connection relationships from raw material to process, process to process, and process to result based on the source number, receiving number, and occurrence time. Finally, it writes all connected units corresponding to the same result unit into the traceability unit set in a unified order, so that the subsequent graph construction module can directly read the unit number, source number, receiving number, and occurrence time to establish graph nodes and graph edges. In practical applications: For example, a record of weighing Tremella polysaccharide generates a Tremella polysaccharide raw material unit, a record of weighing tea extract generates a tea extract raw material unit, a feeding record generates a process unit, a replenishment record generates a subsequent process unit, and a finished product formation record generates a result unit. The system first extracts the unit number, source number, receiving number, and occurrence time from the above records. Then, it compares the receiving numbers of the two raw material units with the source numbers of the process units corresponding to the feeding, compares the receiving numbers of the process units corresponding to the feeding with the source numbers of the subsequent process units corresponding to the replenishment, and compares the receiving numbers of the subsequent process units with the source numbers of the result units. In this way, it establishes connection records from raw material to process, from process to process, and from process to result in sequence. Finally, it writes each unit into the same traceability unit entry according to the occurrence time of each unit, thereby forming a set of traceability units that can be directly used for subsequent map construction.

[0018] The graph construction module, based on the traceability unit set, establishes raw material batches, process nodes, and finished product batches as knowledge graph nodes, and establishes knowledge graph edges according to the correspondence between raw materials entering the process, process transfer process, and process forming finished product, and outputs the traceability knowledge graph; In this embodiment, the graph construction module is used to convert raw material batches, process nodes, and finished product batches in the traceability unit set into directly searchable nodes and edges, so that upstream raw material batches and process nodes can be read backwards from the finished product batches along the edges. Specifically, the module first reads the unit number, source number, receiving number, and occurrence time of each raw material batch, process node, and finished product batch from the traceability unit set and writes them into node records. Then, with the process node as the center, the receiving number of the raw material node is compared item by item with the source number of the process node, and the receiving number of the preceding process node is compared item by item with the source number of the following process node, forming edge records from raw material to process and from process to process. Then, with the finished product node as the endpoint, the receiving number of the process node is compared item by item with the source number of the finished product node, forming edge records from process to finished product. Finally, the node records and edge records are merged into a traceability knowledge graph. Here, the node identifier uniformly uses the unit number, and the edge establishment uniformly adopts the writing method where the source number and receiving number are the same and the occurrence time of the upstream node is earlier than or equal to the occurrence time of the downstream node. This implementation process includes the following steps: First, after obtaining the traceability unit set, the unit number, source number, receiving number, and occurrence time of each raw material batch, process node, and finished product batch are read sequentially, and raw material nodes, process nodes, and finished product nodes are established respectively. For the Tremella polysaccharide raw material batch and the tea extract raw material batch, a raw material node is established with its respective unit number as the node identifier, and the source number, receiving number, occurrence time, and raw material category are written into the raw material node. For each process node, a process node is established with the unit number of each process node as the node identifier, and the source number, receiving number, occurrence time, and process category are written into the process node. For each finished product batch, a finished product node is established with the unit number of each finished product batch as the node identifier, and the source number, receiving number, occurrence time, and finished product category are written into the finished product node. If there are two or more records with the same unit number in the traceability unit set, first compare the source number, successor number, and occurrence time of these records, and retain the record where the source number, successor number, and occurrence time are all non-empty; if there are still two or more records that meet this condition, retain the record with the earliest occurrence time; if the occurrence times are still the same, retain the record whose original writing order corresponds to the unit number first, and do not create nodes for the remaining records; after completing the above processing, the graph node set is obtained. Secondly, based on the graph node set, the source number and occurrence time of each process node are read sequentially, and the currently read process node is taken as the target process node. Subsequently, the acceptance number and occurrence time of all raw material nodes are read item by item, and the acceptance number of each raw material node is compared with the source number of the target process node, and the occurrence time of each raw material node is compared with the occurrence time of the target process node. When the acceptance number of the raw material node is the same as the source number of the target process node, and the occurrence time of the raw material node is earlier than or equal to the occurrence time of the target process node, the raw material node is connected to the target process node, and an edge record is written. The edge record contains the start node identifier, the end node identifier, the edge type, and the writing time, where the edge type is written as "raw material to process". Then, the connection number and occurrence time of all process nodes are read item by item. Process nodes whose occurrence time is earlier than or equal to the occurrence time of the target process node and whose node identifier is different from the target process node are designated as preceding process nodes. The connection number of each preceding process node is then compared item by item with the source number of the target process node. When the connection number of a preceding process node is the same as the source number of the target process node, and the occurrence time of the preceding process node is earlier than or equal to the occurrence time of the target process node, the preceding process node is connected to the target process node, and an edge record is written. The edge record includes the start node identifier, end node identifier, edge type, and writing time, with the edge type set to process-to-process. If the same target process node corresponds to multiple raw material nodes with the same connection number or multiple preceding process nodes with the same connection number, multiple edge records are written for each. If the same preceding process node corresponds to multiple subsequent process nodes with the same source number, multiple edge records are written for each. For nodes with the same connection number but whose upstream node occurrence time is later than the occurrence time of the target process node, no edge record is written. After completing the above processing, a process-related edge set is obtained. Next, based on the process-related edge set, the source number and occurrence time of each finished product node are read sequentially, and the currently read finished product node is taken as the target finished product node. Subsequently, the acceptance number and occurrence time of all process nodes are read item by item, and the acceptance number of each process node is compared with the source number of the target finished product node, and the occurrence time of each process node is compared with the occurrence time of the target finished product node. When the acceptance number of a process node is the same as the source number of the target finished product node, and the occurrence time of the process node is earlier than or equal to the occurrence time of the target finished product node, the process node is connected to the target finished product node, and an edge record is written. The edge record contains the starting node identifier and the ending node identifier. The graph records the node identifier, edge type, and writing time, where the edge type is written as process to finished product. If the same target finished product node corresponds to multiple process nodes with the same acceptance number, then all process nodes that meet the conditions are connected to the target finished product node and written into edge records respectively. After completing the writing of edge records from process nodes to finished product nodes, all node records in the graph node set are used as a node data table, and the process association edge set and all edge records from process to finished product are used as an edge data table. Then, the node data table and the edge data table are merged to obtain the traceability knowledge graph. In this traceability knowledge graph, each node is uniquely identified by a node identifier, and each edge is jointly identified by the start node identifier and the end node identifier. Through the above processing, the graph construction module first writes the raw material batch, process node and finished product batch into the node record respectively, and then writes the edge records from raw material to process, process to process and process to finished product in the manner that the source number and the receiving number are the same and the upstream node occurs earlier than or equal to the downstream node. Finally, the traceability knowledge graph is generated, so that the subsequent path tracing module can directly retrieve the upstream node level by level according to the node identifier and edge record. In practical applications: For example, the batch of Tremella polysaccharide raw material corresponds to raw material node A, the batch of tea extract raw material corresponds to raw material node B, feeding corresponds to process node C, replenishing corresponds to process node D, fermentation completion corresponds to process node E, and finished product formation corresponds to finished product node F. The system first establishes node records for A, B, C, D, E, and F respectively. Then, it compares the receiving number of A with the source number of C, the receiving number of B with the source number of C, the receiving number of C with the source number of D, the receiving number of D with the source number of E, and the receiving number of E with the source number of F. For each group of nodes with the same number and the occurrence time of the previous node being earlier than or equal to the occurrence time of the next node, edge records from A to C, B to C, C to D, D to E, and E to F are written respectively. Finally, all node records and edge records are merged into a traceability knowledge graph.

[0019] The path tracing module, based on the source tracing knowledge graph, searches backward level by level along the edge of the knowledge graph, starting from the finished product batch. The raw material batches and process nodes connected to the finished product batch are arranged in the order of occurrence time to obtain the source combination path corresponding to the finished product batch, and output the source combination path set. In this embodiment, the path tracing module is used to start from the finished product node in the source knowledge graph, read the process nodes and raw material nodes actually connected to the finished product node level by level, and organize them to obtain the source combination path corresponding to the finished product node without changing the original connection relationship. In specific processing, the node identifier of each finished product node is used as the starting point of the search. The starting node is read level by level along the edge record with the finished product node as the ending node. Then, the read process node is used as the new search node and the same reading is repeated until no new upstream node can be read. Then, the read raw material nodes and process nodes are reordered according to the time of occurrence. Finally, the sorted node order is used as the reading order, and the actual edge record in the source knowledge graph is used as the connection basis to generate source combination paths one by one. Here, whether a node continues to be read upward is based on whether the node type is a process node; whether a node is written repeatedly is based on whether the node identifier has already appeared in the same search branch; sorting only changes the reading order of nodes when organizing the path, and does not change the original connection relationship between nodes. This implementation process includes the following steps: First, after obtaining the source knowledge graph, the node identifiers of all finished nodes are read, and the node identifiers are used as the starting points for retrieval in ascending order. For each finished node read, a reverse retrieval sequence corresponding to that finished node is generated, and the node identifier, node type, source number, successor number, and occurrence time of the finished node are written into the first record of the reverse retrieval sequence. If there are multiple finished nodes in the source knowledge graph, multiple reverse retrieval sequences are generated for each. If there are duplicate finished nodes with the same node identifier, only the record with the same node identifier, source number, successor number, and occurrence time is retained as the starting point for retrieval, and the other duplicate records are not used to generate new reverse retrieval sequences. After completing the above processing, a set of starting point sequences is obtained, where each starting point sequence includes at least a sequence number, a finished node identifier, and a list of current nodes. Subsequently, based on the starting sequence set, the current search node in each reverse search sequence is read sequentially, and all edge records in the source knowledge graph with the termination node identifier equal to the current search node identifier are retrieved; for each retrieved edge record, its starting node identifier is read, and then the node type, source number, successor number, and occurrence time corresponding to the starting node identifier are read from the node record; if the starting node is a process node, the process node is written to the end of the current reverse search sequence, and the same edge record retrieval is continued with the process node as the new current search node; if the starting node is a raw material node, the raw material node is written to the end of the current reverse search sequence, but the edge record retrieval is no longer continued with the raw material node; If the node identifier of a node to be written has already appeared in the same search branch of the current reverse search sequence, then the node to be written will not be written to the same search branch again; if the same finished product node can be traced back to the upstream node with the same node identifier through different edge records, then the upstream node and the corresponding edge record are retained in different search branches respectively; the above search process continues to be executed until all process nodes in the current reverse search sequence have completed the upward search and no new process nodes or raw material nodes are read; after the above processing is completed, a set of connected node sequences is obtained, where each connected node sequence corresponds to a finished product node and all process nodes and raw material nodes that the finished product node is connected to upward; Secondly, based on the connected node sequence set, all node records in each connected node sequence are read sequentially. Finished product node records, which serve as the starting point for retrieval, are deleted, leaving only raw material node records and process node records as sorting objects. Then, the occurrence time of each sorting object is extracted and sorted from earliest to latest. If two nodes have different occurrence times, the node with the earlier occurrence time is ranked first. If two nodes have the same occurrence time, the raw material node is ranked before the process node. If two nodes are both raw material nodes or both process nodes and have the same occurrence time, they are sorted by node identifier from smallest to largest. After completing the above processing, a sorted node sequence set is obtained. Here, the sorted node order is only used to determine the temporal order of upstream nodes corresponding to the same finished product node and is not used to replace the connection relationships in the edge records. When generating subsequent paths, the correspondence between the starting node identifier and the ending node identifier in the edge records is still used as the basis for whether nodes are connected. Next, based on the sorted node sequence set, the first node in each sorted node sequence is read sequentially and written as the first node of the current source combination path. Then, the next node in the sorted node sequence is read, and the edge record in the source knowledge graph is searched for an edge record whose starting node identifier is equal to the identifier of the last node of the current source combination path and whose ending node identifier is equal to the identifier of the next node. If the corresponding edge record is found, the next node is written into the current source combination path. If the corresponding edge record is not found, the subsequent nodes in the sorted node sequence are read, and the same edge record search is performed one by one until a node directly connected to the last node of the current source combination path is found, and then the node is written into the current source combination path. If the last node of the current source combination path does not find a directly connected node in any subsequent node of the sorted node sequence, the writing of the current source combination path ends, and another source combination path is regenerated with the next node in the sorted node sequence that has not yet been written into a path as the new first node. This process continues until all nodes in the sorted node sequence have been written into one or more source combination paths. Subsequently, all source combination paths corresponding to the same finished node are summarized and written into the source combination path set, and the finished node identifier, path number, path node list, and occurrence time of each node are written into each source combination path. After completing the organization of all paths corresponding to the finished nodes, the source combination path set is obtained. Through the above processing, the path tracing module first reads the process nodes and raw material nodes level by level from the finished product node along the edge record, then sorts the read raw material nodes and process nodes according to the time of occurrence, and finally generates source combination paths one by one according to the sorted reading order and combined with the actual connection relationship in the edge record, so that the upstream raw material batches and process nodes corresponding to the same finished product batch can be extracted in the form of continuous and verifiable paths. In practical applications: For example, the upstream edge record corresponding to a certain finished product node sequentially connects the fermentation completion process node, the feeding process node, the feeding process node, the Tremella polysaccharide raw material node, and the tea extract raw material node. The system first uses the finished product node as the starting point for retrieval, reads the edge record whose termination node identifier is equal to the finished product node identifier, and obtains the fermentation completion process node. Then, it continues to read the edge record whose termination node identifier is equal to the fermentation completion process node identifier, and obtains the feeding process node and the feeding process node. Then, it continues to read the edge record whose termination node identifier is equal to the feeding process node identifier, and obtains the Tremella polysaccharide raw material node and the tea extract raw material node. Subsequently, the occurrence time of the above nodes is extracted and arranged from the earliest to the latest occurrence time. Finally, according to the actual connection relationship in the edge record, the Tremella polysaccharide raw material node to the feeding process node, the tea extract raw material node to the feeding process node, the feeding process node to the feeding process node, and the feeding process node to the fermentation completion process node are written into the source combination path set, thereby forming a source combination path that can be directly called by the subsequent boundary determination module.

[0020] The boundary determination module, based on the source combination path set, performs item-by-item comparison of the occurrence time and acceptance number of the adjacent nodes in each source combination path, retains the source combination paths with the occurrence time progressively and the corresponding acceptance numbers, and extracts the retained source combination paths before the first abnormal result of the corresponding finished product batch, and outputs the abnormal source combination and its impact boundary. In this embodiment, the boundary determination module is used to further determine which paths can be retained as valid anomaly tracing paths based on the established source combination paths, and thereby determine the anomaly source combination and its corresponding impact boundary. Specifically, each source combination path is first divided into pairs of adjacent path nodes, and then the numbering connection relationship, temporal sequence relationship, internal path connectivity relationship, and container acceptance relationship between the path node pairs are checked sequentially. Path node pairs and source combination paths that cannot be consecutively established are eliminated. Based on this, the retained source combination paths are aligned with the detection records corresponding to the same batch of finished products to determine the time position of the first occurrence of the anomaly, and nodes after that time position are deleted from the corresponding source combination path. Finally, the combination of abnormal sources and the boundary of influence are obtained. To ensure that the judgment process is clear and executable, in this embodiment, the test record includes at least the test object number, test time, test item, test value, and test result. The test result preferentially adopts the qualified or abnormal identification already given by the quality inspection system. When the original test record only saves the test value and not the test result, the test result is generated according to the upper and lower limit ranges of the quality control corresponding to the test item. The test value falling within the upper and lower limit ranges of the quality control is recorded as qualified, and the value exceeding the upper and lower limit ranges of the quality control is recorded as abnormal. When there are multiple test records for the same batch of finished products at the same test time, they are first arranged according to the preset order of the test items, and then the first abnormal test record is determined according to the order after the arrangement. This implementation process includes the following steps: First, after obtaining the source combination path set and detection records, read all nodes in each path one by one according to the source combination path, and extract the node number, source number, receiving number, occurrence time and container number from each node. The node number is used to uniquely identify the current node, the source number is used to identify the upstream output identifier referenced by the current node when receiving materials, the receiving number is used to identify the output identifier that the downstream node can continue to reference after the current node is completed, the occurrence time is used to identify the actual time when the business action corresponding to the current node is completed, and the container number is used to identify the container where the material corresponding to the current node is located. Then, according to the order of nodes in the same path, the preceding node and the following node are combined into a path node pair in turn, and a corresponding record is generated for each path node pair. The corresponding record should at least include the path number, the preceding node number, the following node number, the preceding node source number, the preceding node successor number, the following node source number, the following node successor number, the occurrence time of the preceding node, the occurrence time of the following node, the container number of the preceding node, and the container number of the following node. If a source combination path contains only one node, no path node pair is generated for that path, and it is directly recorded as a path to be excluded. If a source combination path has duplicate node numbers, the first node to appear is retained, and subsequent nodes with the same number are deleted to avoid forming self-circulating node pairs in subsequent comparisons. After completing the above processing, a set of path node pairs classified and stored by path number is obtained. Subsequently, based on the path node pair set, the succession number and occurrence time of the previous node, as well as the source number and occurrence time of the next node in each path node pair are read sequentially. The source number of the next node is compared with the succession number of the previous node item by item, and the occurrence time of the next node is compared with the occurrence time of the previous node item by item. When the source number of the next node is the same as the succession number of the previous node, and the occurrence time of the next node is later than the occurrence time of the previous node, the path node pair is written into the first pass marker. When the source number of the subsequent node is different from the successor number of the preceding node, or when the occurrence time of the subsequent node is earlier than that of the preceding node, the path node pair is written to the first blocking flag. If the occurrence time of the subsequent node is the same as that of the preceding node, the completion order field in the original business records corresponding to the preceding and subsequent nodes is further read. If a completion order field exists and the completion order of the preceding node is earlier than that of the subsequent node, the path node pair is written to the first passing flag. If no completion order field exists or the completion order field cannot distinguish the order, the path node pair is written to the first blocking flag. Both the first passing flag and the first blocking flag are written to the path node pair record, and each written item includes at least the flag type and the flag writing time. After completing the above processing, a first comparison result set is obtained, where each path node pair has a clear first passing flag or first blocking flag. Subsequently, based on the first comparison result set, the path node pairs marked with the first pass are collected and written according to each source combination path. Two adjacent path node pairs within the same path are taken as a group of continuous verification objects. The node number of the same node in the middle of the group of continuous verification objects, the successor number of the previous node, the source number and successor number of the same node, and the source number of the next node are read in sequence. The same comparison is performed on the node number of the same node in the two path node pairs before and after it. The same comparison is performed on the successor number of the previous node and the source number of the same node. The same comparison is performed on the successor number of the same node and the source number of the next node. When all three comparisons are the same, it indicates that the current source combination path is continuously connected between adjacent path node pairs, and the source combination path is written to the connection mark. When any of the three comparisons is different, it indicates that there is a breakpoint between adjacent path node pairs in the current source combination path, and the source combination path is written to the backtrack mark. If a source combination path contains only one path node pair, the continuous verification between two adjacent path node pairs is not performed. Instead, if the path node pair has been written to the first pass mark, the source combination path is directly written to the connection mark. Both the connection mark and the backtrack mark are written to the source combination path record. The corresponding written items include at least the path number, mark type, and mark writing time. After the above processing is completed, the path connection result set is obtained. The source combination paths with only the connection mark are entered into the subsequent container comparison step, and the source combination paths with the backtrack mark are no longer involved in the subsequent boundary determination. Secondly, based on the path connection result set, the container numbers of each node in each source combination path written to the connection mark are read sequentially, and the container comparison is performed item by item according to the order of the preceding node and the following node in the same path. Specifically, the container number of the preceding node is compared with the container number of the following node item by item. When the two are the same, it means that the path node pair is completed continuously in the same container, and the path node pair is written to the container connection mark. When the two are different, the transfer record in the time interval corresponding to the path node pair is read, and the transfer-out container number, transfer-in container number and transfer completion time are extracted from the transfer record. Then, the container number of the preceding node is compared with the transfer-out container number item by item, and the container number of the following node is compared with the transfer-in container number item by item. When both of the above comparisons are the same, and the transfer completion time is no earlier than the occurrence time of the previous node and no later than the occurrence time of the next node, the path node pair is written into the container connection marker; when no corresponding transfer record is read, or although a transfer record is read, the outgoing container number is different from the container number of the previous node, or the incoming container number is different from the container number of the next node, or the transfer completion time is not between the occurrence time of the previous node and the occurrence time of the next node, the path node pair is written into the container blocking marker; both the container connection marker and the container blocking marker are written into the path node pair record, and the corresponding written item includes at least the marker type, the corresponding transfer record number, and the marker writing time; after completing the above processing, a container comparison result set is obtained, where only when all path node pairs in a source combination path are written into the container connection marker, the source combination path is considered a valid candidate path and enters the subsequent anomaly interception step; Subsequently, based on the container comparison result set, the source combination path written to the container through mark is read for each finished product batch, and all test records corresponding to that finished product batch are read. Then, the test records are arranged from earliest to latest according to the test time. If multiple test records have the same test time, they are arranged according to the preset test item order. The preset test item order can be set to stability items first, activity items first, conventional physicochemical items first, and conventional physicochemical items first, sensory items first. After the sorting is completed, the test results in each test record are read one by one, and the test time of the first test record with an abnormal test result is determined as the abnormal cutoff point. If all test records corresponding to the finished product batch are qualified, no abnormal cutoff point is generated for the finished product batch, and its corresponding source combination path will not enter the abnormal source combination extraction step. After the abnormal cutoff point is determined, all nodes in each source combination path corresponding to the finished product batch are read one by one. Nodes whose occurrence time is later than the abnormal cutoff point and their subsequent path nodes are deleted, and only nodes whose occurrence time is earlier than or equal to the abnormal cutoff point are retained. If a source combination path no longer contains a continuous connection relationship between raw material nodes and process nodes after deletion, the source combination path is deleted from the candidate path. If at least one raw material node and at least one process node are still retained after deletion, the source combination path is retained and written into the candidate abnormal path set. After completing the above processing, a candidate abnormal path set is obtained by classifying and storing by finished product batch, where each candidate abnormal path corresponds to a clear abnormal cutoff point. Next, based on the candidate abnormal path set, the source combination paths retained after deletion are collected separately for each finished product batch, and the tremella polysaccharide raw material batch, tea extract raw material batch, and process node in each source combination path are extracted to form an abnormal source combination; specifically, firstly, the node type of each candidate abnormal path corresponding to the same finished product batch is read one by one, the nodes belonging to the tremella polysaccharide raw material batch are extracted into the tremella polysaccharide raw material batch set, the nodes belonging to the tea extract raw material batch are extracted into the tea extract raw material batch set, and the nodes belonging to the process node are extracted into the process node set. Then, the tremella polysaccharide raw material batch set, the tea extract raw material batch set, and the process node set are used as a set of abnormal source combinations corresponding to the finished product batch. Subsequently, within all process nodes of the anomaly source combination, the nodes are sorted from earliest to latest occurrence time. The process node with the latest occurrence time that is still retained in the candidate anomaly path is selected as the last-level process node. The receiving number and container number of this last-level process node are read, and combined with the current finished product batch number, the receiving number, container number, and finished product batch number are combined as the influence boundary. Here, the influence boundary indicates that the anomaly source combination can still continuously propagate to the last verifiable propagation position of the current finished product batch before the anomaly cutoff point. If the same finished product batch corresponds to multiple anomaly source combinations, corresponding influence boundaries are generated separately. If multiple anomaly source combinations correspond to the same last-level process node receiving number, the same container number, and the same finished product batch number, they are merged into the same influence boundary. Finally, the anomaly source combination and its influence boundary corresponding to each finished product batch are written into the boundary determination result, and the anomaly source combination and influence boundary are output. Through the above processing, the boundary determination module first performs number connection and time sequence verification on adjacent nodes in the source combination path, then reviews the continuous connection relationship within the same path, and then judges the continuity of materials in spatial connection by combining container number and transfer record. Finally, it determines the time position of the first occurrence of the anomaly by combining the detection record, and extracts the path range that still maintains continuous transmission before that time position. Thus, the determination of the anomaly source combination and its impact boundary is based on the simultaneous existence of four types of relationships: number continuity, time continuity, container continuity, and result continuity. Since this implementation method clearly defines the composition method of detection record, the value method of detection result, the sorting method of multiple detection records at the same detection time, the writing position of various markers, the verification method of transfer record, the determination method of anomaly cut-off point, and the value method of impact boundary, the problems of anomaly cut-off point not being able to be determined, path cut-off range not being clear, container connection relationship not being able to be judged, or the meaning of impact boundary not being clear will not occur during subsequent result control. In practical applications: For example, a batch of finished products corresponds to two source combination paths. The first source combination path includes, in sequence, the nodes for the raw material of Tremella polysaccharide, the feeding process, the replenishment process, and the fermentation completion process. The second source combination path includes, in sequence, the nodes for the raw material of tea extract, the feeding process, the replenishment process, and the fermentation completion process. The system first generates path node pairs composed of adjacent nodes and compares the receiving number of the previous node with the source number of the next node, and the occurrence time of the previous node with the occurrence time of the next node, retaining path node pairs with consistent numbers and later occurrence times. Then, it checks whether the middle node of adjacent path node pairs within each source combination path is consistent and whether the preceding and following numbers are connected end to end to determine whether the entire path remains continuous. Afterward, it reads the container number of each node. If the container numbers corresponding to the feeding process node and the replenishment process node are different, then further... Read the transfer records to confirm that the transfer-out container number and transfer-in container number correspond to the container numbers of the preceding and following nodes, and that the transfer completion time is between the occurrence times of the preceding and following nodes. After completing the above verification, read the test records of the finished product batch, such as stability test records, activity test records, and physicochemical test records. Determine the test time corresponding to the first abnormal test record as the abnormal cutoff point according to the test time and test item order, and delete the nodes in each source combination path whose occurrence time is later than the abnormal cutoff point. Finally, extract the tremella polysaccharide raw material batch, tea extract raw material batch, and process nodes in the deleted and retained paths to form an abnormal source combination. Use the receiving number, container number, and current finished product batch number of the fermentation completion process node with the latest occurrence time in the abnormal source combination as the influence boundary, thereby providing a direct, continuous, and verifiable judgment basis for the generation of subsequent frozen objects, review objects, and recall objects.

[0021] The results control module extracts the corresponding raw material batches, process nodes, and finished product batches based on the combination of anomaly sources and impact boundaries, generates frozen objects, review objects, and recall objects, and outputs the raw material traceability management results. In this embodiment, the result control module is used to further convert the abnormal source combination and impact boundary into directly executable control results, so that the raw material batches, process nodes, and finished product batches in the abnormal source combination can fall into the frozen object, the review object, and the recall object respectively, and form a unified raw material traceability management result. In specific processing, firstly, the raw material batches of Tremella polysaccharide, raw material batches of tea extract, process nodes, and finished product batches are extracted from the abnormal source combination and impact boundary, and written into the object list according to their respective abnormal source combination and impact boundary; then, the occurrence frequency of each raw material batch in different abnormal source combinations, the number of connections between each process node and each finished product batch, and the corresponding number between each finished product batch and each impact boundary are counted, and the classification and writing of frozen objects and review objects are performed accordingly; then, according to the finished product batch number and the receiving number and container number in the impact boundary, the finished product batch is divided into recall units, and the recall units with continuous connection paths with the frozen object or the review object are written into the recall object; finally, the frozen object, the review object, and the recall object are summarized and written into the raw material traceability management result; this implementation process includes the following steps: First, after obtaining the combination of anomalies and their impact boundaries, the batch numbers of Tremella polysaccharide raw materials, tea extract raw materials, process nodes, and finished product batches in each anomaly source combination are read sequentially. The acceptance number, container number, and finished product batch number in the impact boundary corresponding to the anomaly source combination are also read. Then, an object list record is established with each anomaly source combination as a combination unit, and the Tremella polysaccharide raw material batch, tea extract raw material batch, process node, and finished product batch in the combination unit are expanded and written into the object list respectively. Specifically, an object item is written for each batch of Tremella polysaccharide raw material, an object item is written for each batch of tea extract raw material, an object item is written for each process node, and an object item is written for each finished product batch. In each object item, the combination number, object type, object number, corresponding finished product batch number, corresponding receiving number, and corresponding container number are written simultaneously. If the same object appears repeatedly in the same abnormal source combination, only one object item is retained. If the same object appears in different abnormal source combinations, different object items are written according to different combination numbers. After completing the above processing, the control object set is obtained. Subsequently, based on the control object set, all raw material batch object items are first grouped by object number, and the number of combination numbers corresponding to each raw material batch object item is counted. This number of combination numbers is used as the occurrence count of the raw material batch. When a raw material batch occurs twice or more, it is written into the frozen object. When a raw material batch occurs once, it is written into the review object. Next, all process node object items are grouped by process node number and finished product batch number. The process node number and finished product batch number corresponding to each group are read, and then the traceability knowledge graph is searched for continuous connection paths starting from the process node and ending at the finished product batch. A continuous connection path is counted as one connection count. When two or more continuous connection paths are found, the process node is written into the frozen object. When a continuous connection path is found, the process node is written into the review object. For process nodes for which no continuous connection path is found, they are not written into the frozen object or the review object. After completing the above processing, a process control set is obtained, in which each frozen object or review object is written with object type, object number, combination number, finished product batch number, receiving number, and container number. Secondly, based on the process control set, the finished product batch number, corresponding acceptance number, and corresponding container number in each finished product batch object item are read sequentially, and grouped according to the three fields of finished product batch number, acceptance number, and container number; when multiple finished product batch object items have the same finished product batch number, the same acceptance number, and the same container number, these finished product batch object items are merged into a recall unit, and a recall unit number is generated for the recall unit; after the recall unit division is completed, the frozen objects and review objects in the process control set are read sequentially, and then the traceability knowledge graph is searched to see if there is a continuous connection path between each frozen object or review object and the finished product batch in each recall unit; When a continuous connection path exists, the recall unit is written into the recall object, and the recall unit number, finished product batch number, acceptance number, container number, and the number of the frozen object or review object connected to it are written into the recall object; when the same recall unit is connected to multiple frozen objects or multiple review objects respectively, all connected object numbers are written into the same recall object one by one, without generating new recall objects repeatedly; after completing the above processing, the result control set is obtained; Next, based on the process control set and the result control set, the frozen objects, the review objects, and the recalled objects are read respectively, and classified writing is performed according to object type to generate raw material traceability management results. Specifically, for frozen objects, the object type, raw material batch number or process node number, combination number, finished product batch number, acceptance number, and container number are written; for review objects, the object type, raw material batch number or process node number, combination number, finished product batch number, acceptance number, and container number are written; for recalled objects, the object type, recall unit number, finished product batch number, acceptance number, container number, and the connected frozen object number or review object number are written. If the same raw material batch or the same process node is written to both the frozen object and the review object at the same time, only the writing result of the frozen object is retained, and the writing result of the corresponding review object is deleted; if the same finished product batch appears in multiple recall objects at the same time, they are merged according to the principle that the acceptance number and container number are the same, and one recall object is retained; finally, the raw material traceability management result is output. Through the above processing, the result control module first expands the combination of anomaly sources and impact boundaries into an object list, then performs combination number counting on raw material batches, continuous connection path counting on process nodes, and grouping finished product batches by finished product batch number, receiving number, and container number. Finally, it forms frozen objects, review objects, and recall objects respectively, and summarizes and generates raw material traceability management results. As a result, the control attribution of raw material batches, process nodes, and finished product batches has a clear value basis, statistical basis, and connection basis, which can avoid problems such as unclear object attribution, unclear counting caliber, and inability to define the recall scope. In practical applications: For example, if a batch of Tremella polysaccharide raw material appears in two abnormal source combinations with combination numbers Z1 and Z2, the system will group the batch by object number and count the number of combination numbers as two, and write the batch of Tremella polysaccharide raw material into the frozen object; if a batch of tea extract raw material appears only in one abnormal source combination with combination number Z1, the system will count the number of combination numbers as one, and write the batch of tea extract raw material into the review object; if a process node has a continuous connection path with two different finished product batches after being retrieved by the traceability knowledge graph, the process node will be written into the frozen object; if a finished product batch object item has the same finished product batch number, the same receiving number, and the same container number as another finished product batch object item, the two will be merged into one recall unit, and when a continuous connection path is found between the recall unit and the aforementioned frozen object or review object, the recall unit will be written into the recall object, and finally uniformly written into the raw material traceability management result.

[0022] Working principle: First, the scattered records of raw materials (such as Tremella polysaccharide, tea extract, production processes, and finished products) are organized into standardized, connectable units. Then, based on source numbers, receiving numbers, and occurrence times, the transmission relationships between raw materials, processes, and finished products are gradually connected to form a traceability knowledge graph. On this basis, the process traces back from the finished product batch to identify the raw material batches and process nodes actually connected to that batch, and organizes them into source combination paths according to their occurrence times. Subsequently, these source combination paths are checked sequentially to verify whether the numbers are consecutive, whether the times are progressive, and whether the containers are continuously connected. This is then combined with the detection records... The system identifies the time point when the anomaly first appears, removes path nodes later than that time point, and thus obtains the combination of anomaly sources truly related to the anomaly result and its impact boundary. Finally, based on the anomaly source combination and impact boundary, it distinguishes the raw material batches and process nodes that need to be frozen, the raw material batches and process nodes that need to be reviewed, and the finished product batches that need to be recalled, forming the raw material traceability management results. In other words, this solution does not only look at a single ledger record or a single batch number, but connects the raw material flow, production transmission, result formation, and anomaly exposure into a complete link, and then gradually filters out the truly relevant anomaly sources and impact ranges along this link. For example, in the production of fermented beverages, a batch of Tremella polysaccharide and a batch of tea extract undergo weighing, feeding, replenishment, and fermentation to form several batches of finished products. Subsequently, the company discovers that the stability test results of one of these batches are abnormal. At this point, the system first organizes the corresponding raw material units, process units, and result units from the raw material records, process records, and finished product records. Then, it establishes connections between the Tremella polysaccharide raw material batch, the tea extract raw material batch, the feeding nodes, the replenishment nodes, the fermentation nodes, and the finished product batches. Next, the system traces backward from this abnormal batch to find the upstream raw material batches and process nodes involved. The system checks each node segment to ensure consistency in numbering, reasonable timing, and corresponding transfer records for container switching. When the system reads the test records corresponding to the finished product batch, it can pinpoint the exact time point from which the anomaly began and eliminate subsequent paths after that time point, retaining only the raw materials and process chains truly related to the anomaly. In this way, the company can clearly identify which batches of Tremella polysaccharide raw materials, which batches of tea extract raw materials, which feeding or fermentation process nodes need to be frozen or reviewed, and which finished product batches need to be recalled, thus narrowing the scope of investigation and handling to the actual affected objects.

[0023] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A raw material traceability management system for the mixed fermentation of Tremella polysaccharide and tea extract, characterized in that, include: The data consolidation module acquires raw material unit data, process unit data, and result unit data. It performs corresponding association and time sorting on the batches of Tremella polysaccharide raw materials, tea extract raw materials, mixed fermentation process nodes, and finished product batches according to the unit number and occurrence time in each data, and outputs a traceability unit set. The graph construction module, based on the traceability unit set, establishes raw material batches, process nodes, and finished product batches as knowledge graph nodes, and establishes knowledge graph edges according to the correspondence between raw materials entering the process, process transfer process, and process forming finished product, and outputs the traceability knowledge graph; The path tracing module, based on the source tracing knowledge graph, searches backward level by level along the edge of the knowledge graph, starting from the finished product batch. The raw material batches and process nodes connected to the finished product batch are arranged in the order of occurrence time to obtain the source combination path corresponding to the finished product batch, and output the source combination path set. The boundary determination module, based on the source combination path set, performs item-by-item comparison of the occurrence time and acceptance number of the adjacent nodes in each source combination path, retains the source combination paths with the occurrence time progressively and the corresponding acceptance numbers, and extracts the retained source combination paths before the first abnormal result of the corresponding finished product batch, and outputs the abnormal source combination and its impact boundary. The results control module extracts the corresponding raw material batches, process nodes, and finished product batches based on the combination of anomaly sources and impact boundaries, generates frozen objects, review objects, and recall objects, and outputs the raw material traceability management results.

2. The raw material traceability management system for the mixed fermentation of Tremella polysaccharide and tea extract according to claim 1, characterized in that: The data consolidation module includes: Acquire raw material unit data, process unit data, and result unit data. Extract unit number, source number, receiving number, and occurrence time from each data. Generate Tremella polysaccharide raw material unit, tea extract raw material unit, process unit, and result unit according to the unit number, and output the unit detail set. Based on the unit detail set, the source number and occurrence time of each process unit are read sequentially. The source number is compared with the successor number of each Tremella polysaccharide raw material unit, each tea extract raw material unit, and the preceding process unit. The Tremella polysaccharide raw material unit, tea extract raw material unit, and preceding process unit that match the comparison and whose occurrence time is earlier than the occurrence time of the process unit are connected to the process unit, and the unit connection set is output. Based on the unit connection set, the source number and occurrence time of each result unit are read sequentially. The source number is compared with the successor number of each process unit item by item. Process units that match the source number and whose occurrence time is earlier than the occurrence time of the result unit are connected to the result unit. Then, the connected units are sorted from earliest to latest according to their occurrence time, and the source unit set is output.

3. The raw material traceability management system for the mixed fermentation of Tremella polysaccharide and tea extract according to claim 2, characterized in that: The map construction module includes: Obtain the traceability unit set, read the unit number, source number, acceptance number and occurrence time of each raw material batch, each process node and each finished product batch, and establish raw material node, process node and finished product node using the unit number of each raw material batch, process node and finished product batch as node identifier, and output the graph node set; Based on the graph node set, the source number and acceptance number of each process node are read sequentially. The source number of each process node is compared with the acceptance number of each raw material node, and the raw material nodes with the same comparison are connected to the corresponding process node. Then, the source number of each process node is compared with the acceptance number of the preceding process node, and the preceding process nodes with the same comparison are connected to the corresponding process node. The process association edge set is output. Based on the process association edge set, the source number of each finished product node is read sequentially. The source number of each finished product node is compared with the acceptance number of each process node one by one. Process nodes with the same comparison are connected to the corresponding finished product node. Then, the graph node set, process association edge set and connection relationship from process node to finished product node are merged to output the traceability knowledge graph.

4. The raw material traceability management system for the mixed fermentation of Tremella polysaccharide and tea extract according to claim 3, characterized in that: The path tracing module includes: Obtain the traceability knowledge graph, read the node number of each finished product node, and generate a reverse retrieval sequence one by one with the node number of each finished product node as the retrieval starting point, and output the starting point sequence set; Based on the starting sequence set, read the process nodes and raw material nodes directly connected to the current node level by level along the incoming edges of each finished product node. The read process nodes and raw material nodes are appended to the end of the corresponding reverse search sequence in sequence. The same reading is performed along the incoming edges for the appended process nodes until no new nodes appear in the appended nodes. The connected node sequence set is then output.

5. The raw material traceability management system for the mixed fermentation of Tremella polysaccharide and tea extract according to claim 4, characterized in that: The path tracing module also includes: Based on the set of connected node sequences, the occurrence time of each process node and each raw material node in each connected node sequence is extracted, and the process nodes and each raw material node are rearranged from earliest to latest according to the occurrence time, and the sorted node sequence set is output. Based on the sorted node sequence set, the raw material nodes and process nodes corresponding to the same finished product node are connected end to end in the sorted order to generate the source combination path corresponding to the finished product node. The source combination paths corresponding to each finished product node are summarized and the source combination path set is output.

6. The raw material traceability management system for the mixed fermentation of Tremella polysaccharide and tea extract according to claim 5, characterized in that: The boundary determination module includes: Obtain the source combination path set and detection records, extract the node number, source number, successor number, occurrence time and container number of each node in each source combination path in sequence, and generate path node pairs in the order of the previous node and the next node in the same path, and output the path node pair set. Based on the path node pair set, the receiving number and occurrence time of the previous node and the source number and occurrence time of the next node in each path node pair are read sequentially. The source number of the next node is compared with the receiving number of the previous node item by item, and the occurrence time of the next node is compared with the occurrence time of the previous node item by item. Path node pairs with the same number and the occurrence time of the next node is later than the occurrence time of the previous node are written into the first pass flag, and the remaining path node pairs are written into the first block flag. The first comparison result set is output.

7. The raw material traceability management system for the mixed fermentation of Tremella polysaccharide and tea extract according to claim 6, characterized in that: The boundary determination module also includes: Based on the first comparison result set, the path node pairs marked with the first pass are collected and written according to each source combination path. The node number of the same node in the middle of two adjacent path node pairs, the successor number of the previous node, the source number and successor number of the same node, and the source number of the next node are read in sequence. The same comparison is performed on the node number of the same node in the two path node pairs, the same comparison is performed on the successor number of the previous node and the source number of the same node, and the same comparison is performed on the successor number of the same node and the source number of the next node. The source combination path with the same three comparisons is written with the through mark, and the source combination path with any one of the comparisons is written with the back mark. The path through result set is output. Based on the path connectivity result set, the container numbers of each node in each source combination path written with the connectivity mark are read sequentially. The container number of the previous node is compared with the container number of the next node item by item. For path node pairs with the same container number, the container connectivity mark is written. For path node pairs with different container numbers, the outgoing container number and the incoming container number in the transfer record are read. The container number of the previous node is compared with the outgoing container number item by item, and the container number of the next node is compared with the incoming container number item by item. For path node pairs with the same two comparisons, the container connectivity mark is written. For the remaining path node pairs, the container blocking mark is written. The container comparison result set is output.

8. The raw material traceability management system for the mixed fermentation of Tremella polysaccharide and tea extract according to claim 7, characterized in that: The boundary determination module also includes: Based on the container comparison result set, the source combination path written to the container connection mark is read for each finished product batch, and the detection records corresponding to the finished product batch are arranged from earliest to latest according to the occurrence time. The detection results in each detection record are read one by one, and the occurrence time of the first detection record with an abnormal detection result is determined as the abnormal cutoff point. Then, the nodes in each source combination path whose occurrence time is later than the abnormal cutoff point and their subsequent path nodes are deleted, and the candidate abnormal path set is output. Based on the candidate abnormal path set, the source combination paths that are retained after deletion are collected separately for each finished product batch. The batches of Tremella polysaccharide raw materials, tea extract raw materials, and process nodes in each source combination path are extracted to form an abnormal source combination. The process node with the latest occurrence time in each abnormal source combination is combined with the acceptance number, container number, and finished product batch number as the influence boundary, and the abnormal source combination and influence boundary are output.

9. A raw material traceability management system for the mixed fermentation of Tremella polysaccharide and tea extract according to claim 8, characterized in that: The result control module includes: Obtain the combination of abnormal sources and their impact boundaries. Extract the raw material batches of Tremella polysaccharide, raw material batches of tea extract, process nodes, and finished product batches from each abnormal source combination. Write the raw material batches, process nodes, and finished product batches from each abnormal source combination into the object list according to the combination attribution relationship, and output the control object set. Based on the control object set, the occurrence count of each raw material batch in each abnormal source combination, the number of connections between each process node and each finished product batch, and the corresponding number between each finished product batch and each influence boundary are read sequentially. Raw material batches with more than one occurrence count and process nodes with more than one connection count are written into the freeze object, and raw material batches and process nodes with a corresponding count of one are written into the review object. The process control set is then output.

10. A raw material traceability management system for the mixed fermentation of Tremella polysaccharide and tea extract according to claim 9, characterized in that: The result control module also includes: Based on the process control set, the batch number of each finished product batch and the acceptance number in the corresponding impact boundary are read sequentially. Finished product batches with the same batch number and the acceptance number located in the same impact boundary are merged into recall units. Each recall unit connected to the frozen object or the review object is written into the recall object, and the result control set is output. Based on the process control set and the result control set, the frozen objects, the review objects and the recalled objects are summarized. The raw material batch number and process node number corresponding to each frozen object, the raw material batch number and process node number corresponding to each review object, and the finished product batch number and impact boundary corresponding to each recalled object are written into the raw material traceability management result, and the raw material traceability management result is output.

Images