A device manufacturing quality early warning method and system based on multi-source manufacturing data fusion

Abstract

The application discloses a device manufacturing quality early warning method and system based on multi-source manufacturing data fusion, relates to the technical field of manufacturing quality early warning, and comprises the following steps: cross-closing judgment is performed on a manufacturing track deviation list and a reverse support fracture list, manufacturing process abnormal risk and monitoring conclusion support insufficient risk are distinguished, and a two-way closed early warning list with risk sources and monitoring point positioning is generated; whether the current risk has been sufficiently proved is judged according to the two-way closed early warning list, if the risk still cannot be positioned, the least supplementary evidence action capable of verifying the authenticity of the low-risk conclusion is selected, and a supplementary evidence early warning instruction for supplementary detection and record verification is generated. Through the two-way closed early warning list and the supplementary evidence early warning instruction, the application converts the insufficiently proved risk items into supplementary detection and record verification tasks, so that key evidence can be directly obtained by early warning disposal, and the effects of improving risk identification accuracy and monitoring closed-loop processing efficiency are achieved.

Description

Technical Field

[0001] This invention relates to the field of manufacturing quality early warning technology, and in particular to a method and system for equipment manufacturing quality early warning based on multi-source manufacturing data fusion. Background Technology

[0002] Equipment manufacturing process management typically revolves around process flow, process parameters, quality inspection, and supervision. It tracks the manufacturing status of equipment components by collecting manufacturing process records, inspection records, and supervision confirmation records, and combines these with historical qualified batches, quality acceptance rules, and supervision point records to form a basis for quality judgment. With the improvement of data acquisition capabilities at the manufacturing site, multi-source manufacturing data fusion is increasingly being used for manufacturing process management, quality traceability, and risk alerts.

[0003] The aforementioned conventional methods still have two limitations in equipment manufacturing quality early warning: First, there is a lack of two-way verification between the manufacturing process and the supervision conclusion, making it difficult to determine whether the quality risk comes from deviations in the manufacturing process or a breakdown in the support of the conclusion; second, when the risk is unclear, it usually relies on adding inspection items, lacking a minimum supplementary evidence selection mechanism for verifying the authenticity of low-risk conclusions. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for early warning of equipment manufacturing quality based on multi-source manufacturing data fusion to solve the problems of difficulty in bidirectionally locating the sources of equipment manufacturing quality risks and the lack of targeted supplementary evidence actions.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for early warning of equipment manufacturing quality based on multi-source manufacturing data fusion, which includes: collecting equipment manufacturing process data and equipment manufacturing business data, and performing object binding and time standardization processing to generate a standardized manufacturing event stream; Based on the time sequence and object attribution in the standardized manufacturing supervision event flow, the current equipment manufacturing process is restored into a real-time manufacturing event trajectory, and it is matched and verified with the qualified manufacturing event trajectory formed by historical qualified batches to generate a manufacturing trajectory deviation list. Based on the deviation process and deviation monitoring point in the manufacturing trajectory deviation list, trace back to the upstream manufacturing events related to the corresponding monitoring conclusions, verify the supporting relationship between the manufacturing process and the monitoring conclusions, and generate a reverse support failure list. Cross-close the manufacturing trajectory deviation list and the reverse support failure list to distinguish between abnormal risks in the manufacturing process and insufficient support in the supervision conclusions, and generate a two-way closed early warning list with risk sources and supervision point locations. Based on the two-way closed early warning list, determine whether the current risk has been sufficiently proven. If the risk still cannot be located, select the minimum supplementary evidence action that can verify the authenticity of the low-risk conclusion, and generate supplementary evidence early warning instructions for supplementary testing and record verification.

[0007] As a preferred embodiment of the equipment manufacturing quality early warning method based on multi-source manufacturing data fusion described in this invention, the steps of performing object binding and time standardization processing to generate a standardized manufacturing supervision event stream are as follows: The equipment manufacturing process data and equipment supervision business data are merged into the same processing queue according to the collection source and generation time, and object binding is performed according to the equipment number, component name, process name and supervision point name to generate an object binding supervision dataset. Time standardization processing is performed on the collection time and manufacturing stage sequence in the object-bound supervision dataset to generate a time-standardized supervision dataset. The time-standardized monitoring dataset is encapsulated into events according to equipment objects and manufacturing stages to generate a standardized monitoring event stream.

[0008] As a preferred embodiment of the equipment manufacturing quality early warning method based on multi-source manufacturing data fusion described in this invention, the step of restoring the current equipment manufacturing process into a real-time manufacturing event trajectory includes the following steps: Extract manufacturing events and supervision events corresponding to the same current equipment from the standardized supervision event stream, and establish event connection relationships according to time sequence and object affiliation to generate the current equipment manufacturing event sequence; Based on the current equipment manufacturing event sequence, the flow path of the current equipment in each manufacturing stage is reconstructed, and the process events, inspection events and monitoring point events are connected to form a real-time manufacturing event trajectory.

[0009] As a preferred embodiment of the equipment manufacturing quality early warning method based on multi-source manufacturing data fusion described in this invention, the steps of verifying the conformity with the qualified manufacturing event trajectory formed by historical qualified batches to generate a manufacturing trajectory deviation list are as follows: The real-time manufacturing event trajectory is matched and verified with the qualified manufacturing event trajectory to determine the deviation of the current equipment in terms of process sequence, event occurrence time and connection of supervision point, and generate trajectory matching and verification record; Based on the trajectory alignment verification record, the deviation process, deviation time point, and deviation monitoring point are extracted, and the deviation content is mapped to the specific manufacturing stage of the current equipment to generate a manufacturing trajectory deviation list.

[0010] As a preferred embodiment of the equipment manufacturing quality early warning method based on multi-source manufacturing data fusion described in this invention, the reverse tracing of the upstream manufacturing events associated with the manufacturing supervision conclusion refers to determining the manufacturing supervision conclusion to be traced back based on the deviation process and deviation supervision point in the manufacturing trajectory deviation list, and querying the upstream manufacturing events in the standardized manufacturing supervision event stream that correspond to the manufacturing supervision conclusion to be traced back, thereby generating an upstream manufacturing event matching set.

[0011] As a preferred embodiment of the equipment manufacturing quality early warning method based on multi-source manufacturing data fusion described in this invention, the steps of verifying the support relationship between the manufacturing process and the manufacturing conclusion, and generating a reverse support failure list, are as follows: Perform a quality contribution reachability analysis on the matching set of upstream manufacturing events to determine whether each upstream manufacturing event can be transmitted to the manufacturing supervision conclusion to be traced back along the process flow relationship, and generate a quality contribution support chain. The quality contribution chain breakpoint backtracking method is used to identify manufacturing event breakpoints in the quality contribution support chain that would cause the manufacturing supervision conclusion to lose support, and generate support relationship verification records. Based on the support relationship verification record, missing processes, missing evidence, and abnormal time relationships are extracted and mapped to the deviation monitoring points in the manufacturing trajectory deviation list to generate a reverse support fracture list.

[0012] As a preferred embodiment of the equipment manufacturing quality early warning method based on multi-source manufacturing data fusion described in this invention, the step of performing cross-closure judgment on the manufacturing trajectory deviation list and the reverse support fracture list is as follows: The manufacturing trajectory deviation list and the reverse support fracture list are correlated according to the equipment object, manufacturing stage and deviation from the monitoring point to generate a two-way risk correlation table. Perform yaw fracture closure matching on the two-way risk association table to determine whether the deviation process and the support fracture point to the same manufacturing stage, and generate manufacturing process abnormal risk items.

[0013] As a preferred embodiment of the equipment manufacturing quality early warning method based on multi-source manufacturing data fusion described in this invention, the steps of distinguishing between manufacturing process anomaly risks and insufficient support for monitoring conclusions, and generating a two-way closed early warning list with risk sources and monitoring point locations, are as follows: Using manufacturing process anomaly risk items to screen the remaining risks in the two-way risk association table, identify the monitoring points that are not explained by manufacturing process anomalies but have support fractures, and generate the monitoring conclusion of insufficient support risk items; By integrating the abnormal risk items in the manufacturing process with the risk items with insufficient support in the supervision conclusions according to the source of risk and the location of the supervision point, a two-way closed early warning list is generated.

[0014] As a preferred embodiment of the equipment manufacturing quality early warning method based on multi-source manufacturing data fusion described in this invention, the step of determining whether the current risk has been sufficiently proven based on the two-way closed early warning list, and if the risk still cannot be located, selecting the minimum supplementary evidence action that can verify the authenticity of the low-risk conclusion, and generating a supplementary evidence early warning instruction for supplementary inspection and record verification, is as follows: Based on the two-way closed early warning list, verify whether the abnormal risk items in the manufacturing process and the risk items with insufficient support in the supervision conclusion form a mutual corroboration relationship, and generate a risk proof status table; Based on the risk proof status table, filter out the risk items that are not sufficiently proven from the two-way closed early warning list, and extract the corresponding manufacturing events and supervision events to generate a set of risk events to be verified; Based on the set of risk events to be verified, a set of candidate supplementary evidence actions is constructed, and the minimum falsifiable path screening is performed to generate a combination of supplementary evidence actions that can verify the authenticity of low-risk conclusions with the fewest number of actions. Based on the combination of supplementary evidence actions, the evidence gaps at the corresponding monitoring points are mapped into tasks, and supplementary evidence warning instructions are generated for the corresponding monitoring points. Based on the supplementary evidence warning instructions, supplementary testing and record verification are carried out to generate equipment manufacturing quality risk warning information.

[0015] Secondly, the present invention provides an equipment manufacturing quality early warning system based on multi-source manufacturing data fusion, including: an event stream generation module, which collects equipment manufacturing process data and equipment manufacturing business data, performs object binding and time standardization processing, and generates a standardized manufacturing event stream; The trajectory deviation identification module uses the time sequence and object attribution in the standardized manufacturing supervision event flow as a basis to restore the current equipment manufacturing flow process into a real-time manufacturing event trajectory, and performs a fitting and verification with the qualified manufacturing event trajectory formed by historical qualified batches to generate a manufacturing trajectory deviation list. The support fracture tracing module traces the upstream manufacturing events associated with the corresponding supervision conclusions by going back to the deviated processes and deviations from the supervision points in the manufacturing trajectory deviation list, verifies the support relationship between the manufacturing process and the supervision conclusions, and generates a reverse support fracture list. The two-way closed early warning module cross-closes the manufacturing trajectory deviation list and the reverse support failure list to distinguish between abnormal risks in the manufacturing process and insufficient support in the supervision conclusions, and generates a two-way closed early warning list with the risk source and supervision point location. The verifiability and supplementary evidence module determines whether the current risk has been sufficiently proven based on the two-way closed early warning list. If the risk still cannot be located, it selects the minimum supplementary evidence action that can verify the authenticity of the low-risk conclusion and generates supplementary evidence early warning instructions for supplementary testing and record verification.

[0016] The beneficial effects of the invention are as follows: by generating a manufacturing trajectory deviation list and forming a reverse support failure list, a two-way correlation is achieved between manufacturing flow deviation and support gaps in the supervision conclusion, enabling the equipment supervision quality risk to be located at a specific manufacturing stage and supervision point, which facilitates the identification of the risk source; through a two-way closed early warning list and supplementary evidence early warning instructions, insufficiently proven risk items are transformed into supplementary testing and record verification tasks, enabling early warning handling to directly obtain key evidence, thereby improving the accuracy of risk identification and the efficiency of closed-loop supervision. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a method for early warning of equipment manufacturing quality based on the fusion of multi-source manufacturing data.

[0019] Figure 2 This is a schematic diagram of an equipment manufacturing quality early warning system based on the fusion of multi-source manufacturing data.

[0020] Figure 3 A flowchart for generating a two-way closed early warning list.

[0021] Figure 4 A flowchart for generating a list of trajectory yaws. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figures 1-4This is one embodiment of the present invention, which provides a method for early warning of equipment manufacturing quality based on multi-source manufacturing data fusion, comprising the following steps: S1. Collect equipment manufacturing process data and equipment supervision business data, and perform object binding and time standardization processing to generate a standardized supervision event flow.

[0026] Equipment manufacturing process data refers to the process names, process times, process parameter values, inspection items, inspection times, and inspection conclusions generated during the manufacturing and testing of equipment components.

[0027] Equipment manufacturing supervision data refers to the names of supervision points, witnessing times, supervision conclusions, rectification records, and re-inspection conclusions formed by supervision personnel during the equipment manufacturing quality confirmation process.

[0028] It should be noted that the equipment manufacturing process data is collected by the processing equipment controller, the workstation scanning and reporting records, and the testing records of the testing instruments; the equipment supervision business data is automatically obtained by the on-site records triggered by the supervision point sensing tags, the witness confirmation records generated by the electronic signature equipment, the rectification closed-loop document circulation records, and the return records of the re-inspection and testing equipment.

[0029] The equipment manufacturing process data and equipment supervision business data are merged into the same processing queue according to the collection source and generation time. Based on the preset equipment object mapping table, each record is mapped to the corresponding equipment object, component object, manufacturing stage and supervision point object, generating an object-bound supervision dataset.

[0030] It should be noted that the equipment object mapping table is set up based on the equipment ledger, component list, manufacturing process route, and supervision point configuration relationship by establishing a unique correspondence between equipment objects, component objects, manufacturing stages, and supervision point objects.

[0031] Time standardization is performed on the collection time and manufacturing stage sequence in the object-bound supervision dataset to generate a time-standardized supervision dataset.

[0032] Furthermore, based on the collection time in the object-bound supervision dataset, the occurrence time of each manufacturing event and supervision event is extracted, and a manufacturing stage timeline is established according to the manufacturing stage sequence. Based on the manufacturing stage timeline, the time of each manufacturing event and supervision event is realigned to form an event time realignment dataset. Based on the event time realignment dataset, the sequential relationship between manufacturing events, detection events, and supervision events under the same equipment object is corrected to form a time-aligned supervision dataset. Abnormal time points, duplicate time points, and missing time points in the time-aligned supervision dataset are uniformly marked and their order is corrected to generate a time-standardized supervision dataset.

[0033] The time-standardized monitoring dataset is encapsulated into events according to equipment objects and manufacturing stages to generate a standardized monitoring event stream.

[0034] Furthermore, event encapsulation units are established based on the equipment objects and manufacturing stages in the time-standardized supervision dataset. The process execution records, inspection records, and supervision records of the same equipment object within the same manufacturing stage are converted into manufacturing events, inspection events, and supervision events. For each event encapsulation unit, the event subject, event type, occurrence time, stage location, associated components, supervision point objects, and event content are written, and the succession relationship between upstream and downstream events is established according to the manufacturing stage timeline. Event encapsulation units that occur consecutively under the same equipment object and point to the same manufacturing stage are sequentially connected to generate a standardized supervision event flow.

[0035] S2. Based on the time sequence and object attribution in the standardized manufacturing event flow, the current equipment manufacturing process is restored to a real-time manufacturing event trajectory, and it is matched and verified with the qualified manufacturing event trajectory formed by historical qualified batches to generate a manufacturing trajectory deviation list.

[0036] Extract manufacturing events and supervision events corresponding to the same current equipment from the standardized manufacturing supervision event stream, and establish event connection relationships according to time sequence and object affiliation to generate the current equipment manufacturing event sequence.

[0037] Furthermore, from the standardized monitoring event flow, using the equipment object of the current equipment as the search condition, manufacturing events and monitoring events whose main body belongs to the same current equipment are filtered out, and the manufacturing stage, related components, monitoring point objects, occurrence time, and event content are retained. Based on the manufacturing stage and related components, it is verified whether the manufacturing events and monitoring events belong to the same manufacturing flow link, and events that do not belong to the current equipment manufacturing process are eliminated. The time succession relationship between adjacent events is established according to the occurrence time. When the occurrence time is the same, the succession order is determined according to the manufacturing stage order. The object affiliation connection relationship is established by combining the equipment object, related components, and monitoring point objects to generate a current equipment manufacturing event sequence that can continuously represent the current equipment manufacturing flow process.

[0038] Based on the current equipment manufacturing event sequence, the flow path of the current equipment in each manufacturing stage is reconstructed, and the process events, inspection events and monitoring point events are connected to form a real-time manufacturing event trajectory.

[0039] Furthermore, based on the manufacturing stage, occurrence time, associated components, and event content in the current equipment manufacturing event sequence, process events, inspection events, and supervision point events belonging to the same manufacturing stage are grouped into stage event groups. Process events serve as the starting point for stage flow, inspection events as quality confirmation nodes, and supervision point events as supervision confirmation nodes. The direction of succession within each stage event group is determined according to the occurrence time, and when occurrence times are parallel, the succession relationship is determined according to the manufacturing stage order and the affiliation of associated components. Each stage event group is converted into a stage flow node, and the output of the stage flow node of the previous manufacturing stage is connected to the input of the stage flow node of the next manufacturing stage, forming a continuous flow path for the current equipment's manufacturing process in each manufacturing stage. Process events, inspection events, and supervision point events in the continuous flow path are then chained together according to the stage succession relationship to generate a real-time manufacturing event trajectory that reflects the actual manufacturing progress and supervision intervention position of the current equipment.

[0040] The real-time manufacturing event trajectory is matched and verified with the qualified manufacturing event trajectory to determine the deviation of the current equipment in terms of process sequence, event occurrence time and monitoring point connection, and generate trajectory matching and verification record.

[0041] Furthermore, the real-time manufacturing event trajectory is divided into trajectory nodes to be verified according to the manufacturing stage, associated components, and monitoring points. Qualified trajectory nodes corresponding to the same manufacturing stage, associated components, and monitoring points in the qualified manufacturing event trajectory are used as the fitting benchmark. A one-to-one fitting relationship is established according to the process sequence, standard occurrence time, and monitoring point connection relationship of the trajectory nodes. Trajectory nodes that cannot be matched are marked as missing nodes or abnormally inserted nodes. For matched trajectory nodes, the process sequence is compared to see if it is reversed, whether the event occurrence time exceeds the time range of qualified trajectory nodes, and whether the monitoring point event is later than the detection confirmation or earlier than the process completion. This determines the deviation of the current equipment in terms of process sequence, event occurrence time, and monitoring point connection, and generates a trajectory fitting verification record.

[0042] Based on the trajectory alignment verification record, the deviation process, deviation time point, and deviation monitoring point are extracted, and the deviation content is mapped to the specific manufacturing stage of the current equipment to generate a manufacturing trajectory deviation list.

[0043] Furthermore, based on the node correspondence in the trajectory alignment verification record, the process names corresponding to reversed process order, missing process nodes, and abnormally inserted nodes are extracted to form deviated processes. Based on the comparison between the event occurrence time in the trajectory alignment verification record and the standard occurrence time of qualified trajectory nodes, the occurrence times corresponding to premature occurrence, delayed occurrence, and time sequence conflicts are extracted to form deviation time points. Based on the connection relationship of monitoring points in the trajectory alignment verification record, monitoring point objects that are later than inspection confirmation, earlier than process completion, or not connected to the corresponding process are extracted to form deviation monitoring points. The deviated processes, deviation time points, and deviation monitoring points are mapped to the specific manufacturing stage of the current equipment according to the equipment object, associated components, and stage position in the trajectory alignment verification record, and organized into a manufacturing trajectory deviation list according to the manufacturing stage order. The manufacturing trajectory deviation list clearly locates the deviated processes, deviation time points, and deviation monitoring points of the current equipment manufacturing process relative to the qualified manufacturing event trajectory, enabling quality risks to be identified early from the manufacturing flow process itself, rather than solely relying on inspection conclusions.

[0044] S3. Based on the deviation process and deviation monitoring point in the manufacturing trajectory deviation list, trace back to the upstream manufacturing events related to the corresponding monitoring conclusions, verify the supporting relationship between the manufacturing process and the monitoring conclusions, and generate a reverse support failure list.

[0045] Based on the deviation process and deviation monitoring point in the manufacturing trajectory deviation list, determine the monitoring conclusion to be traced back, and query the upstream manufacturing events in the standardized monitoring event flow that correspond to the monitoring conclusion to be traced back, and generate an upstream manufacturing event matching set.

[0046] Furthermore, based on the deviation monitoring points in the manufacturing trajectory deviation list, the monitoring events corresponding to the same equipment object and the same manufacturing stage are located in the standardized monitoring event flow, and the monitoring conclusions already formed in the monitoring events are read as the monitoring conclusions to be traced back. When the deviation monitoring point does not directly form a monitoring conclusion, the nearest monitoring event is searched backward according to the manufacturing stage to which the deviation process belongs, and the monitoring conclusion in the nearest monitoring event is determined as the monitoring conclusion to be traced back. Using the equipment object, associated parts, manufacturing stage, and formation time corresponding to the monitoring conclusion to be traced back as search conditions, the process events, inspection events, and manufacturing stage events that occurred earlier than the monitoring conclusion to be traced back and belong to the same manufacturing flow link are searched in the standardized monitoring event flow, and events that have no quality association with the deviation process and deviation monitoring point are excluded, generating an upstream manufacturing event matching set that can support the monitoring conclusion to be traced back.

[0047] It should be noted that upstream manufacturing events refer to process events, inspection events, and manufacturing stage events that belong to the same equipment object, the same related component, and the same manufacturing flow link before the conclusion of the retrospective supervision is formed, and that can support the conclusion of the supervision.

[0048] A quality contribution reachability analysis is performed on the matching set of upstream manufacturing events to determine whether each upstream manufacturing event can be transmitted to the manufacturing supervision conclusion to be traced back along the process flow relationship, thereby generating a quality contribution support chain.

[0049] Furthermore, taking the pending retrospective monitoring conclusion as the endpoint, the degree of support for each upstream manufacturing event in terms of process flow relationship, manufacturing stage sequence, consistency of related components, continuity of occurrence time, and correspondence of quality indicators is evaluated. Through multi-factor product fusion, minimum support item penalty, and time deviation attenuation processing, the achievable quality contribution value for each upstream manufacturing event is calculated, expressed as a percentage. When an upstream manufacturing event and the pending retrospective monitoring conclusion belong to the same equipment object and the same related component, and the manufacturing stage can continuously reach the pending retrospective monitoring conclusion along the process flow relationship, and the occurrence time is earlier than the pending retrospective monitoring conclusion while the quality indicators can support the pending retrospective monitoring conclusion, the upstream manufacturing event is determined to be achievable and written into the quality contribution support chain. When the achievable quality contribution value is lower than a preset achievable threshold, the upstream manufacturing event is determined to be unable to effectively transmit to the pending retrospective monitoring conclusion and is marked as a weak support event in the quality contribution support chain.

[0050] The expression for calculating the attainable value of the quality contribution is: ; in, Indicates the first The quality contribution of an upstream manufacturing event can reach a certain value. Indicates the first The first upstream manufacturing event in Scores under supporting factors Indicates a very small correction value. Indicates the first The weight of supporting factors This represents the penalty coefficient for the lowest supporting term. Indicates the first The degree of time deviation between an upstream manufacturing event and the conclusions of the retrospective monitoring. Indicates the time deviation penalty coefficient. It is a time deviation decay term, used to indicate that the greater the time deviation between the upstream manufacturing event and the conclusion of the retrospective manufacturing process, the weaker the supporting role of the upstream manufacturing event in the conclusion of the retrospective manufacturing process.

[0051] It should be noted that supporting factors refer to the evaluation factors used to measure whether upstream manufacturing events can effectively support the conclusions of retrospective supervision, such as process flow relationships, manufacturing stage sequence, consistency of related components, continuity of occurrence time, and correspondence of quality indicators.

[0052] The scores under the supporting factors are obtained by normalizing the degree of matching between upstream manufacturing events and the supervision conclusions to be reviewed in the five categories of supporting factors.

[0053] The degree of time deviation is obtained by normalizing the deviation distance between the occurrence time of the upstream manufacturing event and the standard support time window corresponding to the manufacturing supervision conclusion to be reviewed.

[0054] The weights of supporting factors are set based on the degree of contribution of the five types of supporting factors in the historical qualified batches to the accuracy of the supervision conclusions.

[0055] The penalty coefficient for the minimum support item is set based on the intensity of the impact of weak support factors in historical abnormal batches that lead to inaccurate supervision conclusions.

[0056] The time deviation penalty coefficient is set based on the degree to which time deviation in historical qualified batches and historical abnormal batches reduces the reliability of the supervision conclusion.

[0057] The quality contribution chain breakpoint backtracking method is used to identify manufacturing event breakpoints in the quality contribution support chain that would cause the manufacturing supervision conclusion to lose support, and generate support relationship verification records.

[0058] Furthermore, the quality contribution chain breakpoint backtracking method uses the manufacturing supervision conclusion to be backtracked as the backtracking endpoint. It reads downstream support events, current support events, and upstream support events level by level along the quality contribution support chain according to the reverse flow relationship of the manufacturing stages. Weak support events with quality contribution reachable values ​​lower than a preset reachable threshold are identified as breakpoint candidate events. For each breakpoint candidate event, the adjacent transferable support events before and after the breakpoint candidate event are extracted as the preceding and following support boundaries. The method then searches the upstream manufacturing event matching set for events that simultaneously satisfy the requirements of consistent related components, corresponding quality indicators, and continuous manufacturing stages. The process continues with the replacement manufacturing event that occurs in the same order as the previous support boundary. If the replacement manufacturing event can reconnect the previous support boundary with the subsequent support boundary, the candidate breakpoint event is marked as a non-breakpoint event. If the replacement manufacturing event cannot reconnect the previous support boundary with the subsequent support boundary, the candidate breakpoint event is determined as a manufacturing event breakpoint. The breakpoint cause is generated based on the decline in the achievable value of the quality contribution corresponding to the manufacturing event breakpoint, the breakpoint location, and the retrospective monitoring conclusions of the impact. The process name, occurrence time, associated components, deviation from the monitoring point, and breakpoint cause corresponding to the manufacturing event breakpoint are written into the support relationship verification record.

[0059] Based on the support relationship verification record, missing processes, missing evidence, and abnormal time relationships are extracted and mapped to the deviation monitoring points in the manufacturing trajectory deviation list to generate a reverse support fracture list.

[0060] Furthermore, based on the support relationship verification record, the manufacturing event breakpoint, breakpoint cause, and impact conclusions to be reviewed are read. The names of processes that have not formed effective support are extracted from the manufacturing stage corresponding to the manufacturing event breakpoint as missing processes. The contents of missing inspection records, witness records, and re-inspection records are extracted from the quality indicators and supervision conclusions corresponding to the manufacturing event breakpoint as missing evidence. The relationship between the process completion time being later than the witness time, the inspection time being earlier than the process time, and the re-inspection time being earlier than the rectification record time is extracted from the occurrence time before and after the manufacturing event breakpoint as abnormal time relationships. The missing processes, missing evidence, and abnormal time relationships are mapped to the deviation supervision points in the manufacturing trajectory deviation list according to the equipment object, related components, manufacturing stage, and deviation process, generating a reverse support break list.

[0061] S4. Cross-close the manufacturing trajectory deviation list and the reverse support failure list to distinguish between abnormal risks in the manufacturing process and insufficient support in the supervision conclusion, and generate a two-way closed early warning list with risk source and supervision point location.

[0062] The manufacturing trajectory deviation list and the reverse support failure list are correlated according to the equipment object, manufacturing stage and deviation from the monitoring point to generate a two-way risk correlation table.

[0063] Furthermore, the equipment objects, manufacturing stages, and deviation monitoring points in the manufacturing trajectory deviation list are used as the primary keys for association. Support breakage records with the same equipment objects, manufacturing stages, and deviation monitoring points are searched in the reverse support breakage list. Records with completely identical primary keys are directly associated with the same point. Records with the same manufacturing stage but with upstream and downstream relationships at the deviation monitoring point are associated with adjacent points according to the manufacturing stage timeline. The associated deviation processes, missing processes, missing evidence, and abnormal time relationships are merged into the same risk unit to generate a two-way risk association table. This two-way risk association table associates process deviations in the manufacturing trajectory deviation list and support gaps in the reverse support breakage list with the same equipment object, the same manufacturing stage, and the same deviation monitoring point. This allows early warning judgments to simultaneously cover manufacturing process anomalies and insufficient support in monitoring conclusions, and provides a basis for subsequent risk source identification, monitoring point location, and generation of supplementary evidence early warning instructions.

[0064] Perform yaw fracture closure matching on the two-way risk association table to determine whether the deviation process and the support fracture point to the same manufacturing stage, and generate manufacturing process abnormal risk items.

[0065] Furthermore, taking each associated risk unit as the judgment object, the deviation process, deviation monitoring point, missing process, missing evidence, and abnormal time relationship in the risk unit are read, and the deviation process is checked along the manufacturing stage time axis to see if it is located in the same manufacturing stage corresponding to the support fracture. If the deviation process and the support fracture are located in the same manufacturing stage, it is determined that the manufacturing process deviation has destroyed the support relationship of the manufacturing stage to the monitoring conclusion, and the risk unit is marked as a manufacturing process abnormal risk item. If the deviation process and the support fracture are not located in the same manufacturing stage but have an upstream and downstream continuous transmission relationship, it is further checked whether the deviation process is transmitted to the manufacturing stage where the support fracture is located through subsequent detection events or monitoring point events. If there is a continuous transmission relationship, it is also marked as a manufacturing process abnormal risk item. The marked manufacturing process abnormal risk item is written into the corresponding equipment object, manufacturing stage, deviation monitoring point, risk source, and associated fracture content for subsequent differentiation of the risk of insufficient support for the monitoring conclusion.

[0066] By using the manufacturing process anomaly risk items to screen the remaining risks in the two-way risk association table, the monitoring points that are not explained by manufacturing process anomalies but have support fractures are identified, and the monitoring conclusion of insufficient support risk items is generated.

[0067] Furthermore, the equipment object, manufacturing stage, deviation monitoring point, and deviation process in the manufacturing process anomaly risk item are used as the explained risk scope, and the explained risk scope is compared with each risk unit in the two-way risk association table. When the support fracture content in the risk unit can be explained by the deviation process within the same manufacturing stage, or can be transmitted from the upstream deviation process to the corresponding deviation monitoring point along the manufacturing stage time axis, the risk unit is classified into the manufacturing process anomaly risk item. When the risk unit does not fall into the explained risk scope, and the risk unit still has missing processes, missing evidence, and abnormal time relationships in the reverse support fracture list, the deviation monitoring point corresponding to the risk unit is determined as a monitoring point with support fracture that is not explained by the manufacturing process anomaly, and a monitoring conclusion of insufficient support risk item is generated based on the support fracture content corresponding to the deviation monitoring point.

[0068] By integrating the abnormal risk items in the manufacturing process with the risk items with insufficient support in the supervision conclusions according to the source of risk and the location of the supervision point, a two-way closed early warning list is generated.

[0069] Furthermore, the manufacturing process anomaly risk items and insufficient support for supervision conclusions risk items are categorized according to equipment object, manufacturing stage, and deviation from supervision point, and a corresponding risk source is written for each risk item. Risk items categorized to the same equipment object, manufacturing stage, and deviation from supervision point are merged, retaining the deviation process and transmission relationship in the manufacturing process anomaly risk items, while retaining the missing evidence and abnormal time relationship in the insufficient support for supervision conclusions risk items. Supervision points with only manufacturing process anomaly risk items are marked as manufacturing process anomaly warnings, supervision points with only insufficient support for supervision conclusions are marked as insufficient support for supervision conclusions warnings, and supervision points with both types of risk items are marked as two-way closed warnings. A two-way closed warning list is generated according to risk source, supervision point location, deviation process, support failure content, and warning type. The two-way closed warning list uniformly locates manufacturing process anomaly risks and insufficient support for supervision conclusions to specific equipment objects, manufacturing stages, and supervision points, enabling quality warnings to simultaneously explain whether the risk originates from manufacturing process deviation or conclusion support failure, and providing a clear basis for subsequent supplementary evidence warning instructions.

[0070] S5. Based on the two-way closed early warning list, determine whether the current risk has been sufficiently proven. If the risk still cannot be located, select the minimum supplementary evidence action that can verify the authenticity of the low-risk conclusion, and generate a supplementary evidence early warning instruction for supplementary testing and record verification.

[0071] Based on the two-way closed early warning list, verify whether the abnormal risk items in the manufacturing process and the risk items with insufficient support in the supervision conclusion form a mutual corroboration relationship, and generate a risk proof status table.

[0072] Furthermore, based on the manufacturing process anomaly risk items and insufficient support risk items of the supervision conclusion under the same equipment object, same manufacturing stage, and same supervision point location in the two-way closed early warning list, the deviation process, deviation time point, and transmission path corresponding to the manufacturing process anomaly risk items are extracted respectively, and the missing process, missing evidence, and abnormal time relationship corresponding to the insufficient support risk items of the supervision conclusion are extracted; the deviation process and missing process are checked for stage correspondence, the deviation time point and abnormal time relationship are checked for time closure, and the transmission path and missing evidence are checked for support chain closure; risk items that form a correspondence in all three types of checks are marked as fully proven risk items, and risk items that cannot form a correspondence in any check are marked as insufficiently proven risk items, and a risk proof status table is generated according to the risk item marking results.

[0073] It should be noted that the mutual corroboration relationship refers to the fact that the deviation process, deviation time point, and transmission path in the abnormal risk items of the manufacturing process can form a corresponding relationship with the missing process, missing evidence, and abnormal time relationship in the risk items of insufficient support for the supervision conclusion under the same equipment object, the same manufacturing stage, and the same supervision point.

[0074] Based on the risk proof status table, filter out the risk items that are not sufficiently proven from the two-way closed early warning list, and extract the corresponding manufacturing events and supervision events to generate a set of risk events to be verified.

[0075] Furthermore, based on the risk proof status table, the stage-related verification status, time-closed verification status, and support chain-closed verification status of each risk item in the two-way closed early warning list are read. When a manufacturing process abnormal risk item and a risk item with insufficient support for the supervision conclusion do not form a corresponding relationship in any verification, the corresponding risk item in the two-way closed early warning list is identified as an insufficiently proven risk item. Using the equipment object, manufacturing stage, and supervision point location corresponding to the insufficiently proven risk item as search conditions, manufacturing events corresponding to deviation process, deviation time point, and transmission path are extracted from the standardized supervision event flow. Supervision events corresponding to supervision conclusion, abnormal time relationship, and missing evidence are also extracted. These events are encapsulated according to the same equipment object, the same manufacturing stage, and the same supervision point to generate a set of risk events to be verified.

[0076] Based on the set of risk events to be verified, a set of candidate supplementary evidence actions is constructed, and the minimum falsifiable path screening is performed to generate a combination of supplementary evidence actions that can verify the authenticity of low-risk conclusions with the fewest number of actions.

[0077] Furthermore, the process involves reading insufficiently proven risk items from the set of risk events to be verified, determining whether the lack of sufficient proven risk items stems from missing manufacturing process evidence, inspection evidence, or supervision record evidence. Manufacturing process evidence is then mapped to process parameter review actions, inspection evidence to supplementary inspection actions, and supervision record evidence to record verification actions. For each action, a corresponding supervision point and verifiable risk gap are written, generating a candidate supplementary evidence action set. When performing minimum falsifiable path screening on the candidate supplementary evidence action set, each action is sorted according to its ability to refute low-risk conclusions and the number of risk gaps it covers. Priority is given to candidate supplementary evidence actions that cover the most risk gaps and have the fewest actions, while eliminating those that only repeatedly prove the same risk gap. This generates a supplementary evidence action combination that can verify the authenticity of low-risk conclusions with the fewest actions. This supplementary evidence action combination transforms insufficiently proven risk items into the minimum number of supplementary inspection and record verification actions required to verify the authenticity of low-risk conclusions. This eliminates the need for manual experience to increase inspection items during the supplementary verification process, instead focusing on precisely obtaining key evidence around the risk gaps. This reduces the cost of ineffective supplementary verification and improves the reliability of equipment supervision quality early warning judgments.

[0078] It should be noted that a low-risk conclusion refers to a judgment state in which, among the manufacturing events, inspection events, and supervision events that have been formed in the standardized supervision event flow, the corresponding supervision point in the two-way closed early warning list has not yet been jointly proven to have a quality risk by the manufacturing trajectory deviation list and the reverse support failure list.

[0079] Based on the combination of supplementary evidence actions, the evidence gaps at the corresponding monitoring points are mapped to tasks, and supplementary evidence warning instructions are generated for the corresponding monitoring points.

[0080] Furthermore, based on the supplementary evidence action combination, the evidence gaps at the corresponding supervision points are broken down into manufacturing process review tasks, inspection evidence reinforcement tasks, and supervision record verification tasks according to the objects of proof. The manufacturing process review task is bound to the process stage and parameters to be reviewed; the inspection evidence reinforcement task is bound to the inspection items and inspection locations to be supplemented; and the supervision record verification task is bound to the witnessing nodes and review records to be verified. The execution priority is determined according to the verification role of each task in the authenticity of the low-risk conclusion, and completion conditions, evidence return formats, and anomaly triggering conditions are configured for each task, generating supplementary evidence early warning instructions for the corresponding supervision points. These supplementary evidence early warning instructions transform the supplementary evidence action combination into executable supplementary inspection, process review, and record verification tasks at the corresponding supervision point, enabling early warning handling to go beyond risk alerts and directly obtain the key evidence needed to verify the authenticity of the low-risk conclusion.

[0081] Based on the supplementary evidence warning instructions, supplementary testing and record verification are carried out to generate equipment manufacturing quality risk warning information.

[0082] Furthermore, based on the supplementary evidence warning instruction, supplementary testing tasks and record verification tasks are determined under the corresponding supervision point. The testing items, testing locations, and verification processes associated with the supplementary testing tasks are retested, and the testing time, testing values, and testing conclusions obtained from the supplementary testing are backfilled into the corresponding manufacturing event. The witness nodes, rectification records, and re-inspection conclusions associated with the record verification tasks are verified in terms of source, time, and object, and the verified content is backfilled into the corresponding supervision event. The backfilled manufacturing event and supervision event are then compared with the two-way closed warning list to confirm whether the abnormal risk items in the manufacturing process and the risk items with insufficient support for the supervision conclusions have been resolved, maintained, or upgraded. Equipment supervision quality risk warning information containing risk sources, risk levels, supervision point locations, and handling suggestions is generated.

[0083] This embodiment also provides an equipment manufacturing quality early warning system based on multi-source manufacturing data fusion, including: an event stream generation module, which collects equipment manufacturing process data and equipment manufacturing business data, performs object binding and time standardization processing, and generates a standardized manufacturing event stream; The trajectory deviation identification module uses the time sequence and object attribution in the standardized manufacturing supervision event flow as a basis to restore the current equipment manufacturing flow process into a real-time manufacturing event trajectory, and performs a fitting and verification with the qualified manufacturing event trajectory formed by historical qualified batches to generate a manufacturing trajectory deviation list. The support fracture tracing module traces the upstream manufacturing events associated with the corresponding supervision conclusions by going back to the deviated processes and deviations from the supervision points in the manufacturing trajectory deviation list, verifies the support relationship between the manufacturing process and the supervision conclusions, and generates a reverse support fracture list. The two-way closed early warning module cross-closes the manufacturing trajectory deviation list and the reverse support failure list to distinguish between abnormal risks in the manufacturing process and insufficient support in the supervision conclusions, and generates a two-way closed early warning list with the risk source and supervision point location. The verifiability and supplementary evidence module determines whether the current risk has been sufficiently proven based on the two-way closed early warning list. If the risk still cannot be located, it selects the minimum supplementary evidence action that can verify the authenticity of the low-risk conclusion and generates supplementary evidence early warning instructions for supplementary testing and record verification.

[0084] In summary, this invention achieves a two-way correlation between manufacturing flow deviations and gaps in supervision conclusions by generating a manufacturing trajectory deviation list and forming a reverse support failure list. This allows the quality risks of equipment supervision to be located at specific manufacturing stages and supervision points, facilitating the identification of risk sources. Furthermore, through a two-way closed early warning list and supplementary evidence early warning instructions, insufficiently proven risk items are transformed into supplementary testing and record verification tasks, enabling early warning handling to directly obtain key evidence. This results in improved accuracy of risk identification and efficiency of closed-loop supervision.

[0085] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for early warning of equipment manufacturing quality based on multi-source manufacturing data fusion, characterized in that, include: Collect equipment manufacturing process data and equipment supervision business data, and perform object binding and time standardization processing to generate a standardized supervision event flow; Based on the time sequence and object attribution in the standardized manufacturing supervision event flow, the current equipment manufacturing process is restored into a real-time manufacturing event trajectory, and it is matched and verified with the qualified manufacturing event trajectory formed by historical qualified batches to generate a manufacturing trajectory deviation list. Based on the deviation process and deviation monitoring point in the manufacturing trajectory deviation list, trace back to the upstream manufacturing events related to the corresponding monitoring conclusions, verify the supporting relationship between the manufacturing process and the monitoring conclusions, and generate a reverse support failure list. Cross-close the manufacturing trajectory deviation list and the reverse support failure list to distinguish between abnormal risks in the manufacturing process and insufficient support in the supervision conclusions, and generate a two-way closed early warning list with risk sources and supervision point locations. Based on the two-way closed early warning list, determine whether the current risk has been sufficiently proven. If the risk still cannot be located, select the minimum supplementary evidence action that can verify the authenticity of the low-risk conclusion, and generate supplementary evidence early warning instructions for supplementary testing and record verification.

2. The equipment manufacturing quality early warning method based on multi-source manufacturing data fusion as described in claim 1, characterized in that: The steps for performing object binding and time standardization to generate a standardized monitoring event stream are as follows: The equipment manufacturing process data and equipment supervision business data are merged into the same processing queue according to the collection source and generation time, and object binding is performed according to the equipment number, component name, process name and supervision point name to generate an object binding supervision dataset. Time standardization processing is performed on the collection time and manufacturing stage sequence in the object-bound supervision dataset to generate a time-standardized supervision dataset. The time-standardized monitoring dataset is encapsulated into events according to equipment objects and manufacturing stages to generate a standardized monitoring event stream.

3. The equipment manufacturing quality early warning method based on multi-source manufacturing data fusion as described in claim 1, characterized in that, The steps to restore the current equipment manufacturing process into a real-time manufacturing event trajectory are as follows: Extract manufacturing events and supervision events corresponding to the same current equipment from the standardized supervision event stream, and establish event connection relationships according to time sequence and object affiliation to generate the current equipment manufacturing event sequence; Based on the current equipment manufacturing event sequence, the flow path of the current equipment in each manufacturing stage is reconstructed, and the process events, inspection events and monitoring point events are connected to form a real-time manufacturing event trajectory.

4. The equipment manufacturing quality early warning method based on multi-source manufacturing data fusion as described in claim 3, characterized in that, The process involves verifying the alignment of manufacturing event trajectories with historical qualified batches to generate a manufacturing trajectory deviation list. The steps are as follows: The real-time manufacturing event trajectory is matched and verified with the qualified manufacturing event trajectory to determine the deviation of the current equipment in terms of process sequence, event occurrence time and connection of supervision point, and generate trajectory matching and verification record; Based on the trajectory alignment verification record, the deviation process, deviation time point, and deviation monitoring point are extracted, and the deviation content is mapped to the specific manufacturing stage of the current equipment to generate a manufacturing trajectory deviation list.

5. The equipment manufacturing quality early warning method based on multi-source manufacturing data fusion as described in claim 1, characterized in that, The upstream manufacturing events associated with the reverse traceability of the supervision conclusion refer to determining the supervision conclusion to be traced back based on the deviation process and deviation supervision point in the manufacturing trajectory deviation list, and querying the upstream manufacturing events corresponding to the supervision conclusion to be traced back in the standardized supervision event flow to generate an upstream manufacturing event matching set.

6. The equipment manufacturing quality early warning method based on multi-source manufacturing data fusion as described in claim 5, characterized in that, The verification of the supporting relationship between the manufacturing process and the supervision conclusion is used to generate a reverse support failure list. The steps are as follows: Perform a quality contribution reachability analysis on the matching set of upstream manufacturing events to determine whether each upstream manufacturing event can be transmitted to the manufacturing supervision conclusion to be traced back along the process flow relationship, and generate a quality contribution support chain. The quality contribution chain breakpoint backtracking method is used to identify manufacturing event breakpoints in the quality contribution support chain that would cause the manufacturing supervision conclusion to lose support, and generate support relationship verification records. Based on the support relationship verification record, missing processes, missing evidence, and abnormal time relationships are extracted and mapped to the deviation monitoring points in the manufacturing trajectory deviation list to generate a reverse support fracture list.

7. The equipment manufacturing quality early warning method based on multi-source manufacturing data fusion as described in claim 1, characterized in that, The steps for cross-closing the manufacturing trajectory yaw list and the reverse support fracture list are as follows: The manufacturing trajectory deviation list and the reverse support fracture list are correlated according to the equipment object, manufacturing stage and deviation from the monitoring point to generate a two-way risk correlation table. Perform yaw fracture closure matching on the two-way risk association table to determine whether the deviation process and the support fracture point to the same manufacturing stage, and generate manufacturing process abnormal risk items.

8. The equipment manufacturing quality early warning method based on multi-source manufacturing data fusion as described in claim 7, characterized in that, The steps for distinguishing between manufacturing process anomaly risks and risks with insufficient support from supervision conclusions, and generating a two-way closed-loop early warning list with risk sources and supervision point locations, are as follows: Using manufacturing process anomaly risk items, the two-way risk association table is screened for residual risks to identify monitoring points that are not explained by manufacturing process anomalies but have support fractures, and monitoring conclusions of insufficient support risk items are generated. By integrating the abnormal risk items in the manufacturing process with the risk items with insufficient support in the supervision conclusions according to the source of risk and the location of the supervision point, a two-way closed early warning list is generated.

9. The equipment manufacturing quality early warning method based on multi-source manufacturing data fusion as described in claim 1, characterized in that, The process involves determining whether the current risk has been sufficiently proven based on the two-way closed early warning list. If the risk still cannot be located, the minimum supplementary evidence action that can verify the authenticity of the low-risk conclusion is selected, and supplementary evidence early warning instructions for supplementary testing and record verification are generated. The steps are as follows: Based on the two-way closed early warning list, verify whether the abnormal risk items in the manufacturing process and the risk items with insufficient support in the supervision conclusion form a mutual corroboration relationship, and generate a risk proof status table; Based on the risk proof status table, filter out the risk items that are not sufficiently proven from the two-way closed early warning list, and extract the corresponding manufacturing events and supervision events to generate a set of risk events to be verified; Based on the set of risk events to be verified, a set of candidate supplementary evidence actions is constructed, and the minimum falsifiable path screening is performed to generate a combination of supplementary evidence actions that can verify the authenticity of low-risk conclusions with the fewest number of actions. Based on the combination of supplementary evidence actions, the evidence gaps at the corresponding monitoring points are mapped into tasks, and supplementary evidence warning instructions are generated for the corresponding monitoring points. Based on the supplementary evidence warning instructions, supplementary testing and record verification are carried out to generate equipment manufacturing quality risk warning information.

10. A manufacturing quality early warning system based on multi-source manufacturing data fusion, based on the manufacturing quality early warning method based on multi-source manufacturing data fusion as described in any one of claims 1 to 9, characterized in that, include: The event stream generation module collects equipment manufacturing process data and equipment supervision business data, performs object binding and time standardization processing, and generates standardized supervision event streams. The trajectory deviation identification module uses the time sequence and object attribution in the standardized manufacturing event flow as a basis to restore the current equipment manufacturing process into a real-time manufacturing event trajectory, and performs a matching and verification with the qualified manufacturing event trajectory formed by historical qualified batches to generate a manufacturing trajectory deviation list. The support fracture tracing module traces the upstream manufacturing events associated with the corresponding supervision conclusions by going back to the deviated processes and deviations from the supervision points in the manufacturing trajectory deviation list, verifies the support relationship between the manufacturing process and the supervision conclusions, and generates a reverse support fracture list. The two-way closed early warning module cross-closes the manufacturing trajectory deviation list and the reverse support failure list to distinguish between abnormal risks in the manufacturing process and insufficient support in the supervision conclusions, and generates a two-way closed early warning list with the risk source and supervision point location. The verifiability and supplementary evidence module determines whether the current risk has been sufficiently proven based on the two-way closed early warning list. If the risk still cannot be located, it selects the minimum supplementary evidence action that can verify the authenticity of the low-risk conclusion and generates supplementary evidence early warning instructions for supplementary testing and record verification.

Images