A production management method for a silicon-based display screen

Abstract

The application discloses a production management method of a silicon-based display screen and relates to the technical field of production management. The method comprises the following steps: collecting real-time process data of the silicon-based display screen and preprocessing the real-time process data to obtain a process key parameter record set; obtaining yield drift early warning through similarity comparison based on the process key parameter record set, reserving traceable process batch records, and generating a process drift diagnosis sheet; performing controlled inspection on the process drift diagnosis sheet, generating a structured quality vector, and obtaining a production path atlas through deterministic risk assessment; determining deterministic dispatching through a dynamic priority scheduling algorithm according to the production path atlas to generate dynamic dispatching instructions; executing the dynamic dispatching instructions, recording dispatching order abstracts, and obtaining a shift compliance summary report through bidirectional back-link solidification. The application converts the station and time window constraints of risk assessment items in the production path atlas into a unique executable dispatching result, thereby improving equipment utilization and reducing comprehensive cost.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of production management, and particularly relates to a production management method of a silicon-based display screen. BACKGROUND

[0002] The silicon-based display screen integrates a pixel driving circuit with a sodium calcium silicate glass backplane, cooperates with a functional film system on the photoelectric glass side and a laminated packaging to realize low power consumption and fast response, and has been widely used in outdoor commercial advertising display and building beautification applications. The production chain link for the photoelectric glass process usually includes wafer pretreatment, electrostatic dust removal, circuit coating, initial curing, line inspection, high-temperature sintering, circuit detection, solder paste coating, SPI detection, mounting, reflow soldering, X-ray detection, pre-aging, laminated bonding, backside aging, structure and electrical control installation, and packaging into the warehouse. With the increase of pixel density and the complexity of optical stacking, production management is moving from static parameter care to whole-process data-driven governance with event sequence, short-time sensitive window and in-process path as the core.

[0003] The existing silicon-based display screen production management mainly has two deficiencies. On the one hand, the process data is mainly single-index out-of-bound alarm, and lacks the sequential characterization and batch-level evidence deposition of the event sequence of wafer pretreatment / circuit coating / initial curing / high-temperature sintering / mounting / reflow soldering / laminated curing, and it is difficult to identify the process drift of "same value different sequence" through similarity comparison and form traceable warning basis. On the other hand, the risk assessment result is difficult to be determinedly translated into an executable production path facing specific sites and time windows, lacking of deterministic dispatching under dynamic priority scheduling and execution restriction, and lacking of bidirectional back-link solidification with the previous basis, resulting in the interruption of the link from judgment to dispatching to audit, and the lack of review and auditability. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides a production management method of a silicon-based display screen to solve the problems of lack of batch-level evidence deposition of process event sequence and inability of risk assessment to be determinedly translated.

[0006] To solve the above technical problems, the present application provides the following technical scheme:

[0007] The present application provides a production management method of a silicon-based display screen, which comprises,

[0008] The real-time process data of the silicon-based display screen is collected and preprocessed to obtain a process key parameter record set; based on the process key parameter record set, a yield drift early warning is obtained through similarity comparison, a traceable process batch record is retained, and a process drift diagnosis sheet is generated; the process drift diagnosis sheet is subjected to controlled inspection to generate a structured quality vector, and through deterministic risk assessment, a production path map is obtained; based on the production path map, a deterministic dispatching is performed through a dynamic priority scheduling algorithm to generate a dynamic dispatching instruction; the dynamic dispatching instruction is executed, a dispatching order abstract is recorded, and a shift compliance summary report is obtained through bidirectional back-link solidification.

[0009] As a preferred scheme of the production management method of the silicon-based display screen, the real-time process data of the silicon-based display screen includes a timestamp, a device batch identifier, a process parameter, an online quality feature, and a process event record.

[0010] The preprocessing includes timestamp alignment, device batch association, range unification, event deduplication, baseline field mapping, and evidence anchor generation.

[0011] As a preferred scheme of the production management method of the silicon-based display screen, the process key parameter record set includes a device identifier evidence anchor, a serialized process event, and an online quality feature. As a preferred scheme of the production management method of the silicon-based display screen, based on the process key parameter record set, a yield drift early warning is obtained through similarity comparison, and the specific steps are as follows,

[0012] Based on the process key parameter record set, a fixed sequence splicing is performed through a time sequence anchoring splicing method to generate a traceable process batch record.

[0013] The traceable process batch record is subjected to double-track similarity comparison, and a yield drift early warning is obtained through similarity score calculation.

[0014] As a preferred scheme of the production management method of the silicon-based display screen, the generation of the process drift diagnosis sheet refers to bidirectional reference solidification of the yield drift early warning and the traceable process batch record through an evidence back-link solidification method based on the yield drift early warning to generate the process drift diagnosis sheet.

[0015] As a preferred scheme of the production management method of the silicon-based display screen, the controlled inspection of the process drift diagnosis sheet to generate a structured quality vector includes the following specific steps,

[0016] Based on the device identifier evidence anchor, the process drift diagnosis sheet is subjected to time window alignment controlled inspection to generate a controlled inspection guide.

[0017] According to the controlled inspection guide, the online quality features in the process key parameter record set are extracted, and the structured quality vector is obtained through segment alignment verification.

[0018] As a preferred scheme of the production management method of the silicon-based display screen, the production path atlas is obtained through the deterministic risk assessment, and the specific steps are as follows,

[0019] Based on the structured quality vector and the process drift diagnosis table, the gated risk score is calculated by using the gated risk assessment algorithm, and the risk assessment item is obtained.

[0020] The risk assessment item is gated to the station translation to generate the production path constraint set, and the deterministic path arrangement is performed to obtain the production path atlas.

[0021] As a preferred scheme of the production management method of the silicon-based display screen, the production path atlas is obtained according to the production path atlas, and the specific steps are as follows,

[0022] The risk assessment item level in the production path atlas is extracted, and the risk gated priority mapping is performed to obtain the priority label set;

[0023] The time window in the priority label set is discretized into continuous grids, and the station time slot occupancy table is generated by sequentially filling the time slot grids;

[0024] Based on the station time slot occupancy table and the production path atlas, the equipment identification evidence anchor is deterministically dispatched to obtain the dynamic dispatching instruction.

[0025] As a preferred scheme of the production management method of the silicon-based display screen, the specific steps of executing the dynamic dispatching instruction to record the dispatching order abstract are as follows,

[0026] The instruction field in the dynamic dispatching instruction is flattened and sequenced to obtain the execution event list;

[0027] Each execution matter in the execution event list is time-pointed, and the execution limit fulfillment mark is recorded to obtain the dispatching order abstract.

[0028] As a preferred scheme of the production management method of the silicon-based display screen, the specific steps of obtaining the shift compliance summary report are as follows,

[0029] Based on the dispatching order abstract, the back chain corresponding to the equipment identification evidence anchor and the production path atlas is generated to obtain the back chain solidification item set;

[0030] The back chain solidification item set is sequentially spliced, and the shift compliance summary report is generated by continuously connecting the continuous fingerprints.

[0031] The present application has the advantages that: by directly converting the risk assessment items at the site in the production path map and the time window constraint into a unique and executable dispatching result, the certainty of the sequential and time window is realized when conflicts occur, and the execution limit is fixed with the dispatching instruction, which improves the equipment utilization and beat stability, reduces waiting and rework, reduces the comprehensive cost, and strengthens the auditability and reviewability of the whole dispatching process. BRIEF DESCRIPTION OF DRAWINGS

[0032] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0033] Fig. 1 The flow chart of the production management method of the silicon-based display screen.

[0034] Fig. 2 The flow chart of generating the process drift diagnosis sheet.

[0035] Fig. 3 The flow chart of obtaining the production path map.

[0036] Fig. 4 The flow chart of generating the site time slot placeholder table. DETAILED DESCRIPTION

[0037] In order to make the above-mentioned purposes, features and advantages of the present application more apparent and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings.

[0038] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the present application, therefore the present application is not limited to the specific embodiments disclosed below.

[0039] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. "In one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is it an independent or alternative embodiment that excludes other embodiments.

[0040] REFERENCE Figs. 1-4 For one embodiment of the present application, the embodiment provides a production management method of a silicon-based display screen, comprising the following steps:

[0041] S1: Collecting real-time process data of silicon-based display screen and preprocessing to obtain a set of process key parameter records.

[0042] S1.1: The real-time process data of silicon-based display screen includes timestamp, equipment batch identification, process parameter and process event record;

[0043] Specifically, the timestamp is the local clock time written by the clock of the equipment controller and detection station of each equipment and each detection station when a record is generated;

[0044] The equipment batch identification is the combination of batch number and equipment number generated by scanning the barcode / QR code at the time of feeding;

[0045] The process parameter is collected by the configured sensor, such as coating thickness and uniformity, screen printing line width and opening rate, reflow soldering temperature, mounting deviation and adhesive curing dose and time;

[0046] The process event record is triggered by the state sensor and travel switch on the equipment (such as cleaning, maintenance completion and alignment re-marking) and merged with the current timestamp and equipment batch identification at the triggering moment to form a continuously referenced process event record.

[0047] S1.2: The preprocessing includes timestamp alignment, equipment batch association, range unification, event deduplication, baseline field mapping and evidence anchor generation;

[0048] Specifically, the real-time process data is timestamped, the latest time point in the same batch is taken or linearly interpolated to a unified time point, then the equipment batch association is performed according to “equipment number + batch number + process code”, and the event deduplication is performed on the repeated records at the same time point to generate the equipment identification evidence anchor.

[0049] The events belonging to the same equipment and batch are merged into the same time axis starting from the process event record, the repeated triggers at the same time are deduplicated, the parameter values bound to the events are unified, and the baseline field mapping is performed, and the serialized process events are output according to the sequence of fields on the time axis.

[0050] Starting from the real-time process data of silicon-based display screen, the equipment batch association is performed to combine the real-time process data of silicon-based display screen with the records of the same batch and the same equipment to obtain the spatial distribution index, then the range unification is performed (such as colorimetric, luminance, etc. converted according to unified unit and range), and the event deduplication is performed on the repeated detection results at the same time point; the luminance, colorimetric and spatial distribution indexes are summarized to output the online quality characteristics including luminance uniformity, color purity, defect point number, defect spatial distribution and related detection position information.

[0051] S1.3: The process key parameter record set includes a device identification evidence anchor, a serialized process event, and an online quality feature.

[0052] Specifically, the device identification evidence anchor serves as the unique identification and traceable basis of batch-level data, and fixes the real-time process data of the silicon-based display screen in sequence as a summary for reference and verification between subsequent similarity comparison, controlled inspection, production path atlas, dynamic dispatching instruction, and shift compliance summary report, ensuring consistent data source, complete link, and reproducible for the same batch.

[0053] The serialized process event, as a time sequence framework reflecting the sequence of process actions, arranges actions such as cleaning, target replacement, maintenance completion, and alignment and re-marking in chronological order, providing direct evidence for sensitive segment identification in similarity comparison, yield drift early warning trigger, time window selection for controlled inspection, and arrival order in production path atlas, avoiding judgment bias caused by "same value different sequence".

[0054] The online quality feature, as an objective measurement set reflecting the trend of the current batch of finished products, integrates results such as brightness, chroma, and spatial distribution within the same batch and time window as the serialized process event, providing quantitative support for generating structured quality vectors for controlled inspection, determining risk assessment items for risk assessment, and setting execution limits in dynamic dispatching instructions.

[0055] S2: Based on the process key parameter record set, a yield drift early warning is obtained through similarity comparison, and traceable process batch records are retained to generate a process drift diagnosis sheet.

[0056] S2.1: Based on the process key parameter record set, fixed sequence splicing is performed through the time sequence anchoring splicing method to generate traceable process batch records.

[0057] It should be noted that based on the process key parameter record set, the corresponding records are filtered according to the device batch identifier and a continuous time axis is established in ascending order of time stamp, the serialized process event is used as an anchor point to divide segments, and within each segment, the fields are spliced in the order of "time stamp → device batch identifier → process code → process event record → key process parameter → online quality feature", and the key process parameters and online quality features are aligned to the anchor point according to the continuous time axis to generate spliced rows; at the same time, the device identification evidence anchor is attached to each spliced row to complete the fixed sequence splicing of the entire batch-level continuous record, and the traceable process batch records are output.

[0058] S2.2: Double-track similarity comparison is performed on the traceable process batch records, and yield drift early warning is obtained by calculating the similarity score.

[0059] It should be noted that based on the traceable process batch record, the double-track similarity comparison (including full sequence similarity comparison and fragment sequence similarity comparison) is performed in the process key parameter record set according to the equipment batch identifier and the process code;

[0060] The full sequence similarity comparison aligns the key process parameters and the online quality characteristics field by field in the fixed field order on the common time axis of the two batch-level continuous records, and performs dimension unification and accumulates the closeness; the fragment sequence similarity comparison takes the sequenced process events in the traceable process batch record as an anchor point to demarcate a fragment window, checks the consistency of the event sequence and the closeness of the quality characteristics in each fragment window, and summarizes the fragment similarity;

[0061] The similarity score is calculated based on the fragment sequence similarity comparison as the priority; based on the similarity score, all fragment events are marked as "normal" if the sequence is consistent and the fragment similarity is consistent with the full sequence similarity, marked as "warning" if the fragment similarity is lower than the full sequence similarity, and marked as "serious" if the sequence of any fragment event is inconsistent, to obtain the yield drift warning. The expression for calculating the similarity score is:

[0062] ;

[0063] Wherein, is the similarity score, is the fragment similarity weight, is the number of fragment windows, is the fragment window index value, is the fragment similarity of the th window, is the total number of key process parameters, is the index value of the total number of key process parameters, is the weight of the th key process parameter, is the sequence similarity of the th key process parameter.

[0064] The fragment similarity weight is obtained by the proportion of the number of consistent fragments in the total number of fragments in the fragment sequence similarity comparison, and the exemplary value range is: ;

[0065] The weight of the key process parameter is obtained by the proportion of the consistent direction of the parameter change and the online quality characteristic change in the fragment window, and the exemplary value range is: ;

[0066] More preferably, by simultaneously performing full-sequence similarity comparison and fragment sequence similarity comparison on the traceable process batch record, and unifying fragment sensitive window information and full-sequence multi-parameter proximity into a single similarity score, false positives and false negatives are reduced, triggering controlled inspection and subsequent production path arrangement in advance, improving feedforward control efficiency and batch-level traceability.

[0067] S2.3: Based on the yield drift warning, the evidence back chain solidification method is used to solidify the yield drift warning and the traceable process batch record in both directions, generating a process drift diagnosis sheet.

[0068] It should be noted that based on the yield drift warning and the traceable process batch record, the device identification evidence anchor is established to establish the reference site; the similarity level in the yield drift warning is written into the reference site and points to the traceable process batch record;

[0069] The yield drift warning and the device identification evidence anchor are written into the traceable process batch record to form a forward back chain; the traceable process batch record and the device identification evidence anchor are written into the yield drift warning to form a reverse back chain;

[0070] The forward back chain and the reverse back chain are frozen in time sequence as the same batch link, and are aligned with the serialized process events and online quality characteristics as the reference basis, and the device identification evidence anchor, the similarity level, the forward back chain and the reverse back chain are output in the process drift diagnosis sheet. Fixed field order.

[0071] S3: Controlled inspection is performed on the process drift diagnosis sheet to generate a structured quality vector, and through deterministic risk assessment, a production path map is obtained.

[0072] S3.1: Based on the device identification evidence anchor, the process drift diagnosis sheet is subjected to time window alignment controlled inspection to generate a controlled inspection guide;

[0073] It should be noted that based on the device identification evidence anchor and the traceable process batch record given in the process drift diagnosis sheet, the continuous time axis is positioned in the process key parameter record set according to the traceable process batch record; and the start event and the end event in the traceable process batch record are taken as the time sequence anchor points, and the time stamp is mapped to a sampling time window list;

[0074] Under the constraint of the sampling time window list, the online quality characteristics consistent with the traceable process batch record are extracted from the process key parameter record set, and whether all the time points of each online quality characteristic fall between the corresponding time sequence anchor points is checked according to the time stamp of the serialized process event, to obtain a fragment alignment check conclusion; the device identification evidence anchor, the traceable process batch record, the sampling time window list and the fragment alignment check conclusion are summarized in fixed field order to generate a controlled inspection guide.

[0075] S3.2: Extracting online quality features in the process key parameter record set according to the controlled inspection guide, and obtaining a structured quality vector through segment alignment verification;

[0076] It should be noted that according to the identification of the traceable process batch record in the controlled inspection guide, the detection items corresponding to the online quality feature names are retrieved in the process key parameter record set according to the sampling time window list listed in the controlled inspection guide; the segment alignment verification conclusion given by the controlled inspection guide is called in each sampling time window, and the timestamps and serialized process event anchors of the online quality features are compared one by one, only the detection values whose time points completely fall within the corresponding anchor interval are retained, and the repeated written records are removed;

[0077] The detection values such as brightness, chroma and spatial distribution in each sampling time window are integrated into single-row entries in the order of "identification of traceable process batch record → device identification evidence anchor → identification of sampling time window → name of online quality feature → value of online quality feature → segment alignment verification conclusion", and are sequentially spliced into a structured quality vector according to the sampling time window order.

[0078] S3.3: Based on the structured quality vector and the process drift diagnosis sheet, a gated risk assessment algorithm is used to calculate the gated risk score, and a risk assessment item is obtained;

[0079] It should be noted that the structured quality vector and the process drift diagnosis sheet are the only input sources, and the continuous time axis is located according to the device identification evidence anchor in the process drift diagnosis sheet and the identification of the traceable process batch record; the continuous time axis and the sampling time window in the structured quality vector are matched, and for the online quality features in each sampling time window, according to the single priority order of "prohibition first, review second, and pass last", if any "prohibition" appears, the risk level of the time window is marked as high, if there is "review" but no "prohibition", the risk level of the time window is marked as medium, and if all are "pass", the risk level of the time window is marked as low. The gated risk assessment algorithm is used to calculate the gated risk score; and the gated risk score, together with the device identification evidence anchor, the traceable process batch record identification, the trigger segment and the time range, are written into a unified record to generate a risk assessment item. The expression for calculating the gated risk score is:

[0080] ;

[0081] Wherein, is the gated risk score, is the number of sampling time windows, is the index value of the number of sampling time windows, is the risk level of the time window, is the reference weight coefficient of prohibition, is the reference weight coefficient of review, Passing reference weight coefficient.

[0082] Prohibited reference weight coefficient, obtained by the correlation intensity statistics of risk events and final yield loss, the exemplary value range is: ;

[0083] Review reference weight coefficient, directly determined by the proportion of "review" gating conclusions in all gating conclusions within the same batch sampling time window, the exemplary value range is: [1.5, 2.5];

[0084] Passing reference weight coefficient, determined by the quality feature distribution of normal production batches, the exemplary value range is: [0.8, 1.2];

[0085] Further, the gating risk assessment algorithm locates the sampling time window in the structured quality vector according to the trigger segment and time range in the process drift diagnosis sheet, and then checks the consistency of each index such as brightness uniformity, color purity, defect point number and spatial distribution with the allowed range of the trigger segment in each sampling time window, forming the segment gating conclusions of "passing, reviewing, and prohibiting". The segment gating conclusions of the same batch are summarized in the fixed order of "prohibition first, review second, and passing last" to calculate the gating risk score (high, medium, and low), and are translated into the three judgment fields of "whether to review at the station, whether to limit speed at the station, and whether to release at the station" in the risk assessment item, which are directly referenced by the subsequent production path constraint set and deterministic path arrangement.

[0086] Preferably, the gating risk score is calculated based on the structured quality vector and the process drift diagnosis sheet, the review, speed limit, and release thresholds are determined item by item according to the fixed rules, and the risk assessment item that can be directly landed at the station and the time window is output, realizing the deterministic translation from the measurement result to the production path constraint; ensuring that the subsequent production path map and dynamic dispatching instruction have a unified standard and auditable basis.

[0087] S3.4: Translate the risk assessment item through the gate to the station, generate the production path constraint set, and arrange the deterministic path to obtain the production path map.

[0088] It should be noted that according to the equipment identification evidence anchor recorded in the risk assessment item and the identification of the trigger segment corresponding to the process station recorded in the traceable process batch record, the gating to the station is translated, "whether to review at the station" is translated into a review node in front of the trigger segment corresponding process station, "whether to limit speed at the station" is translated into parameter restriction set at the trigger segment corresponding process station, "whether to release at the station" is translated into release range marked at the trigger segment corresponding process station, and the sampling time window in the structured quality vector is mapped to the station execution time window to form the production path constraint set;

[0089] After the production path constraint set is generated, the time window overlap and site conflict are processed in the fixed order of "review first, speed limit second, and release last" along the continuous time axis of the traceable process batch record. For example, when there is a "review" time window [10:00, 10:15] overlapping with a "release" time window [10:05, 10:20] on the same time axis of the traceable process batch record, the [10:00, 10:15] is occupied first for "review", and the "release" is sequentially filled into the next available interval [10:15, 10:30]. If there is also a "speed limit" time window [10:12, 10:18] at the same time, the "speed limit" is placed after "review", and then "release" is placed, and all process sites, time windows, parameter limits, and review nodes are counted to generate a production path map.

[0090] S4: According to the production path map, a deterministic dispatching is performed through a dynamic priority scheduling algorithm to generate a dynamic dispatching instruction.

[0091] S4.1: Extract the risk assessment item level in the production path map and perform risk gate priority mapping to obtain a priority label set;

[0092] It should be noted that the site identifier, time window, risk assessment item level, parameter limit, review node, and device identifier evidence anchor are read in sequence according to the process site in the production path map, and "review" is mapped to the highest priority, "speed limit" is mapped to the second highest priority, and "release" is mapped to the lowest priority. When multiple gate requirements appear at the same site and at the same time window, the single priority label is reserved in the fixed order of "review priority over speed limit, speed limit priority over release";

[0093] According to the fixed field order of "device identifier evidence anchor→site identifier→time window→priority label→execution limit specification→review node reference→risk assessment item level", a single record is generated, and the priority label set is sorted according to the time window and site order.

[0094] S4.2: Discretize the time window in the priority label set into continuous grids, and sequentially fill the time slot grids to generate a site time slot occupancy table;

[0095] It should be noted that the start time of each record time window in the priority label set is aligned as the alignment point, and the continuous sub-time windows are uniformly divided by advancing step by step from the start time. Each time window in the divided granularity is written into the site identifier, device identifier evidence anchor, priority label, execution limit specification, and review node reference, and a sequence number is assigned to generate continuous time slot grids;

[0096] In each time slot grid, write site identification, device identification evidence anchor, priority label, execution limit description and review node reference; according to the single path of "priority label from high to low → site order in production path map from front to back → time window starting point from early to late", generate processing occupation demand piece by piece, and record the starting time slot from the beginning of the time window. After completing the priority occupation and conflict elimination, the remaining continuous idle time period and the starting time slot and ending time slot are recorded. When encountering occupation conflict, move to the continuous idle time period of the same time window backward in the same time window. If there is no available continuous idle time period in the same time window, move to the continuous idle segment at the starting point of the nearest subsequent time window and synchronously retain the execution limit conclusion and review node reference, and generate a site time slot occupation table.

[0097] S4.3: Based on the site time slot occupation table and the production path map, determine the device identification evidence anchor, and obtain the dynamic dispatching instruction.

[0098] It should be noted that the starting time slot in the site time slot occupation table is from early to late, and the traversal sequence is established in the order of site order in the production path map. Extract each occupation record in the traversal time slot occupation table in the traversal sequence, locate the corresponding constraint item in the production path map according to the site identification and time window, and read the parameter limit, review node and device identification evidence anchor in the constraint item and pair them with the starting time slot and ending time slot of the occupation record one by one to generate a single instruction item.

[0099] For multiple occupation records of the same device identification evidence anchor, concatenate them according to the site order, and only retain the continuous interval of the first starting time slot and the last ending time slot, while retaining the parameter limit, review node and release marker. Perform one-time sorting and concatenate as dynamic dispatching instruction in the order of fixed fields "device identification evidence anchor → site identification → starting time → ending time → parameter limit → review node → release marker".

[0100] Preferably, the dynamic priority scheduling algorithm is based on the production path map and the priority label set, and processes site conflicts in the time grid according to the order of "review priority, speed limit second, release last" and performs time slot order compensation, so that high-risk segments can obtain available time slots first and execution limits are solidified. Therefore, the risk threshold is directly converted into executable site time slot occupation and deterministic dispatching, which improves the utilization of bottleneck equipment and the stability of shift rhythm, reduces waiting and rework caused by time window conflicts, and provides clear and auditable time site basis for subsequent dispatching order summary and bidirectional backtracking.

[0101] S5: Execute the dynamic dispatching instruction, record the dispatching order summary, and solidify it through bidirectional backtracking to obtain a shift compliance summary report.

[0102] S5.1: Flatten and sequence the instruction field in the dynamic dispatch instruction, obtain the execution event list;

[0103] It should be noted that based on the dynamic dispatch instruction, the instruction entries are read one by one in the order of "device identification evidence anchor→site identification→start time→end time→parameter limit→review node→release mark", and are uniformly sorted in ascending order of start time and in the order of site sequence in the production path map. Each sorted instruction entry is solidified as an execution matter, and all execution matters are collected to form an execution event list containing only the instruction fields of "device identification evidence anchor, site identification, start time, end time, parameter limit, review node, and release mark".

[0104] S5.2: Time point dotting is performed on each execution matter in the execution event list, and the execution limit fulfillment mark is recorded, and a dispatch order abstract is obtained;

[0105] It should be noted that based on the execution event list, the execution matters are processed one by one in ascending order of start time. On each execution matter, time point dotting is performed on the start time and the end time respectively, and "start time point dotting" and "end time point dotting" are written.

[0106] Between "start time point dotting" and "end time point dotting", it is checked whether the execution process meets the parameter limit and "parameter limit fulfillment mark" is written. For example, if the recorded coating thickness and uniformity between "start time point dotting" and "end time point dotting" are equal to the target setting in the parameter limit, it is recorded as "satisfies". If the recorded coating thickness and uniformity at any time are inconsistent with the target setting in the parameter limit, it is recorded as "does not satisfy". At the same time, it is checked whether the action name and site identification in the review node field match between the start time point dotting and the end time point dotting, and the completion record is obtained. If there is one and only one complete match, it is judged as satisfying, otherwise it is judged as not satisfying, and "review node fulfillment mark" is written. The release mark is checked whether it has met the execution event list and "release execution fulfillment mark" is written. For example, if there is one and only one completion record named "complete site A release" between "start time point dotting" and "end time point dotting" and it is consistent with the site identification, it is recorded as "satisfies". If it does not appear, appears multiple times, or is inconsistent with the site identification, it is recorded as "does not satisfy". Subsequently, it is solidified as a single row record in the order of "device identification evidence anchor→site identification→start time→end time→start time point dotting→end time point dotting→parameter limit fulfillment mark→review node fulfillment mark→release execution fulfillment mark", and all single row records are spliced to obtain a dispatch order abstract.

[0107] S5.3: Based on the dispatch order abstract, a back chain corresponding to the device identification evidence anchor and the production path map is referenced back, and a back chain solidification entry set is generated;

[0108] It should be noted that based on the start time point dotting, the "device identification evidence anchor, site identification, start time point dotting, end time point dotting, parameter limit fulfillment mark, review node fulfillment mark, release execution fulfillment mark" in the dispatch order abstract are processed in ascending order row by row, and based on the device identification evidence anchor and the site identification, the corresponding constraint items and time windows of the same batch in the production path map are located, and then a pair of corresponding references is generated for each row: the path side reference is written to the located production path map item identification (a unique number formed by splicing the traceable process batch record identification, site identification and time window start and end time), and the dispatch side reference is written to the current dispatch order abstract row identification; the device identification evidence anchor, site identification, start time point dotting, end time point dotting, parameter limit fulfillment mark, review node fulfillment mark, release execution fulfillment mark and the above path side reference and dispatch side reference are fixed in the order of fixed fields as a single back chain fixed item in the same row; all single back chain fixed items are output as a back chain fixed item set in the order of start time point dotting, and the back chain fixed item set is the only input for generating a subsequent shift compliance summary report.

[0109] S5.4: The back chain fixed item set is sequentially spliced and connected by continuous fingerprint to generate a shift compliance summary report.

[0110] It should be noted that based on the back chain fixed item set, a unified sequence is established from early to late according to the start time point dotting, and is spliced into a shift link; the device identification evidence anchor, site identification and end time point dotting in the previous item are concatenated in a fixed order, and then the device identification evidence anchor, site identification and start time point dotting in the next item are written together into the connection site between the two shift links to generate a continuous fingerprint connection identification;

[0111] The continuous fingerprint connection identification is repeatedly written for each connection site of the whole link to ensure that the head and tail are connected and the connection sequence number is recorded; the shift link, continuous fingerprint connection identification and connection sequence number are sequentially fixed as a shift compliance summary report.

[0112] In summary, the present application directly converts the site and time window constraints of the risk assessment item in the production path map into a unique and executable dispatch result, realizes the deterministic compensation according to the sequence and time window when there is a conflict, and makes the execution limit fixed together with the dispatch instruction, which improves the device utilization and beat stability, reduces waiting and rework, reduces the comprehensive cost, and strengthens the auditability and reviewability of the whole dispatch process.

[0113] It should be noted that the above examples are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced, without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. A method of managing production of a silicon-based display screen, characterized by: The application relates to a silicon-based display screen real-time process data collection and preprocessing method. Key process parameter records are obtained by collecting and preprocessing silicon-based display screen real-time process data; Based on the key process parameter records, a yield drift early warning is obtained through similarity comparison, and the specific steps are as follows, Based on the key process parameter records, a fixed sequence splicing is performed through a time sequence anchoring splicing method to generate traceable process batch records; A double-track similarity comparison is performed on the traceable process batch records, and a yield drift early warning is obtained by calculating a similarity score; A process drift diagnosis sheet is generated by retaining the traceable process batch records; A structured quality vector is generated by performing a controlled inspection on the process drift diagnosis sheet; A production path atlas is obtained through deterministic risk assessment, and the specific steps are as follows, Based on the structured quality vector and the process drift diagnosis sheet, a gated risk score is calculated by using a gated risk assessment algorithm to obtain a risk assessment item; The risk assessment item is gated to a station translation to generate a production path constraint set, and a deterministic path arrangement is performed to obtain a production path atlas; A dynamic dispatch instruction is generated by performing a deterministic dispatching on the production path atlas through a dynamic priority scheduling algorithm; A shift compliance summary report is obtained by recording a dispatch order abstract and performing a bidirectional back-link solidification.

2. The production management method of a silicon-based display screen according to claim 1, characterized by: The silicon-based display screen real-time process data comprises a timestamp, a device batch identifier, a process parameter, an online quality feature and a process event record. The preprocessing comprises timestamp alignment, device batch association, scale unification, event deduplication, baseline field mapping and evidence anchor generation.

3. The production management method of a silicon-based display panel according to claim 2, wherein: The key process parameter records comprise a device identifier evidence anchor, a serialized process event and an online quality feature.

4. The production management method of a silicon-based display panel according to claim 3, wherein: The process drift diagnosis sheet is generated by bidirectional reference solidification of the yield drift early warning and the traceable process batch records through an evidence back-link solidification method based on the yield drift early warning.

5. The production management method of a silicon-based display panel according to Claim 4, wherein: The controlled inspection on the process drift diagnosis sheet to generate a structured quality vector is performed according to the following steps, Based on the device identifier evidence anchor, a time window alignment controlled inspection is performed on the process drift diagnosis sheet to generate a controlled inspection guide; According to the controlled inspection guide, the online quality features in the key process parameter records are extracted, and a structured quality vector is obtained through fragment alignment verification.

6. The production management method of a silicon-based display screen according to Claim 5, wherein: The dynamic dispatch instruction is generated by performing a deterministic dispatching on the production path atlas through a dynamic priority scheduling algorithm according to the production path atlas, and the specific steps are as follows, The risk assessment item level in the production path atlas is extracted, and a risk gated priority mapping is performed to obtain a priority label set; The time window in the priority label set is discretized into a continuous grid, and a station time slot placeholder table is generated by performing a time slot grid order filling; Based on the station time slot placeholder table and the production path atlas, a deterministic dispatching is performed on the device identifier evidence anchor to obtain the dynamic dispatch instruction.

7. The production management method of a silicon-based display panel according to Claim 6, wherein: The dynamic dispatch instruction is executed, and a dispatch order abstract is recorded, and the specific steps are as follows, The instruction fields in the dynamic dispatch instruction are flattened and sequenced to obtain an execution event list; Each execution matter in the execution event list is time-pointed, and an execution limit fulfillment mark is recorded to obtain a dispatch order abstract.

8. The production management method of a silicon-based display screen according to Claim 7, characterized by: The shift compliance summary report is obtained according to the following steps, Based on the dispatch order abstract, the back chain solidification item set is generated by referencing back chain corresponding to the device identification evidence anchor and the production path map; The back chain solidification item set is sequentially spliced, and the shift compliance summary report is generated by continuous fingerprint connection.

Images