Ceramic body glaze firing correlation tracing method and system

Abstract

The application discloses a kind of ceramic body glaze firing inspection correlation traceability method and system, comprising the following steps, S1, with single piece product or continuous body group that keeps internal order establishes minimum traceability unit, obtains the identity, time and location anchor point of each production stage and establishes cross-stage candidate relationship;S2, identify the batch change event of body or glaze, determine the transition interval and material batch contribution relationship according to material balance, input, return material, output and product dosage.The ceramic body glaze firing inspection correlation traceability method and system are based on the identity, time, location, material and equipment record left by product in each production stage, connect the body, glaze, glazing, firing and inspection information in the same evidence chain, and keep candidate range and pending state for the relationship that cannot be determined temporarily, so that the traceability result is closer to the material actually used by product and the production process actually experienced.

Description

Technical Field

[0001] This invention relates to the field of ceramic processing technology, specifically to a method and system for tracing the correlation between firing inspection and testing of ceramic blanks and glazes. Background Technology

[0002] Ceramic products typically undergo a continuous process including batching, molding, drying, glazing, firing, and inspection. Existing production management systems often organize data using work orders, planned batches, equipment tasks, and time records, and can also identify products using QR codes, RFID tags, or visual characters. While these methods are suitable for recording planned information and process results, when changing batches of raw materials or glazes, the previous batch of materials is often retained in silos, mixing containers, conveying pipelines, and return systems; products within the same changeover period may be affected by both old and new batches simultaneously. If batches are only classified based on the changeover time recorded by the operator, the resulting source relationships may not match the actual materials used in the product.

[0003] Application CN117371474A discloses a one-item-one-code full lifecycle traceability scheme for ceramic production, which can link product identification with production, circulation and quality information. This type of scheme illustrates the role of identification and information collection in ceramic traceability. However, relying solely on product codes, work order batches or process completion records is still difficult to directly represent the contribution relationship of multiple material batches to the product during the batch transition period, and it is also difficult to distinguish the actual firing experience of products in the same furnace in terms of channels, loading positions, residence time or abnormal events.

[0004] Defects after firing are generally only discovered during the inspection stage. Current practices often link inspection results to products or work orders, and then rely on experience to look up possible raw material batches, glaze batches, or kiln parameters. Once cross-process identification is interrupted due to damage, rejection, merging, splitting, missed reading, or high-temperature deformation, the traceability chain can easily be directly severed. At the same time, inspection results are rarely used to verify the reasonableness of batch change boundaries, firing paths, and candidate associations, which may result in the defect source range being too broad or too narrow, and it is also difficult to positively determine the affected products from known abnormal materials or abnormal process ranges. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for tracing the correlation between ceramic body and glaze firing inspection, in order to solve the problems raised in the background art: how to establish a chain of evidence that is consistent with the actual material flow and actual equipment experience when ceramic production spans multiple stages, including body, glazing, firing and inspection, and the identity records may be discontinuous; how to preserve the contribution relationship of multiple batches to the product during the batch transition period; how to preserve the actual thermal experience of each piece, group or loading position during the firing stage; and how to use the defect cluster boundary obtained from inspection to verify and correct the aforementioned correlation, so that the traceability scope, evidence status and on-site verification basis can be interpreted and verified.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for tracing and linking the firing inspection of ceramic bodies and glazes, comprising: S1, establish the smallest traceability unit with a single product or a continuous blank group that maintains internal order, obtain the identity, time and location anchor points of each production stage and establish cross-stage candidate relationships; S2 identifies batch change events for blanks or glazes, and determines the transition range and material batch contribution relationship based on material balance, input amount, return amount, output amount and product usage. S3, based on the time, location, device status and available glaze material of the traceability unit passing through the glazing device, actual glazing evidence is formed; S4, based on the kiln's spatiotemporal path or loading topology, forms the actual firing experience; S5, Related Inspection Records, checks cross-stage relationships based on time sequence, physical accessibility, equipment status and material balance. When conflicts or anomalies occur, the original relationship is invalidated and candidate relationships are rebuilt according to the anchor points before and after the event. S6 forms a defect clustering region. By comparing its boundary with the batch transition boundary, firing anomaly boundary, and candidate relationship boundary, correction candidates are generated and re-verified. The reverse source range and positive influence range are output along the same evidence chain that has not failed.

[0007] As a further step, the anchor points include unit identifiers, workstation identifiers, arrival or departure times, channels or locations, adjacent units, quantities, and equipment status; in adjacent stages, candidate relationships with formation basis are first established and saved. When damage, rejection, merging, splitting, missing reading, identifier deformation, rework, or re-burning occurs, a finite candidate set is formed based on the identity anchor points before and after the event, quantity conservation, order, and positional adjacency relationships.

[0008] As a further step, in S2, when the materials in the feeding container are approximately fully mixed, the remaining amount and feeding ratio of each batch are recursively calculated based on the initial balance, added amount, returned amount and output amount of each batch of materials according to the sampling interval. The participation amount and normalized contribution ratio of each batch are determined by combining the effective material consumption rate or adhesion rate of the smallest traceability unit. When the input data is an interval value, the interval recursion results are saved and candidate contribution relationships are formed.

[0009] As a further step, in S2, when the conveying pipeline is close to the first-in-first-out or piston flow state, the average residence time of the material is determined according to the effective volume of the pipeline and the average net flow rate of the corresponding time period. The outlet material is initially matched with the corresponding inlet feeding batch, and the batch change transition interval is expanded according to the pipeline residue, flow dispersion and cleaning and emptying records.

[0010] As a further step, in S3, the actual glazing candidate relationships are checked sequentially: the passage interval of the smallest traceability unit intersects with the effective working interval of the glazing device within the calibrated time tolerance, the location is accessible to the device, the device is running and not in a cleaning or disabled state, and there is available glaze during the intersecting period; when there is only one candidate and there is no conflict, they are bound together; when there are multiple candidates, they are retained in parallel; when key records are missing and a finite candidate set cannot be formed, they are recorded as unresolved; when there are conflicts in time, location, device status, or material supply, the relationship is invalidated; when rework or repeated glazing is performed, the glazing experience is added.

[0011] As a further step, in S4, when using a continuous roller kiln, the kiln position is recursively calculated based on the kiln entry sequence, channel, product spacing, section length, and speed record. The kiln is divided into sections at the boundaries of kiln stop, speed reduction, brick chasing, brick jamming, empty space, or manual intervention. The entry and exit times of each section are combined with the concurrent process records to form the actual firing experience. When using kiln cars, saggers, or firing plates for loading, a loading topology is established according to the kiln car number, layer, grid position or sagger position, loading event, and firing cycle. The loading and unloading records are used to jointly verify the position node of the smallest traceability unit.

[0012] As a further development, in S5, the evidence status of cross-stage relationships includes verified, candidate, pending, and invalid. Verified is defined as when all verification conditions are met and the independent records are complete; candidate is defined as when all verification conditions are met but the independent records are incomplete; pending is defined as when key records are missing and a finite set of candidates cannot be formed; invalid is defined as when a clear conflict occurs. When an anomaly occurs, the original relationship affected by it becomes invalid, the relationship is rebuilt as candidate, and the original relationship, original status, supporting records, conflict records, and reasons for change are retained.

[0013] As a further step, in S6, the smallest traceability units are arranged according to the production sequence. Adjacent defect units that meet the calibration conditions in terms of product serial number interval, inspection or production time interval, and defect type are connected to the same judgment interval. When the calibration quantity condition is reached, a defect cluster interval is formed. Its boundary is compared with the boundary of batch change transition, firing abnormality, and candidate related interval. When the deviation exceeds the calibration allowable range, a boundary candidate is generated to move forward, backward, expand, shrink, or remain unchanged. For each candidate, the material balance, path, and cross-stage relationship verification are re-executed. Defect evidence does not directly cover the original relationship.

[0014] As a further step, in S6, each stage record is represented as a node, and cross-stage relationships with evidence status and supporting records are represented as edges; reverse tracing returns to the source of material, glazing, and firing from the defect node along the non-failed edges, and forward tracing reaches the smallest traceability unit that may be affected from the abnormal material batch or abnormal process interval along the same set of edges; candidate edges and unresolved edges are retained and marked with the reasons to be verified, and failed edges stop participating in traversal and are saved in the relationship history.

[0015] A ceramic body glaze firing inspection correlation traceability system includes a data acquisition module, a traceability unit management module, a batch change contribution module, a glazing performance module, a firing experience module, a relationship verification module, a test feedback module, and a traceability output module. Each module works together to acquire identity anchor points, material supply, equipment status, loading position, and inspection results, establish the smallest traceability unit and cross-stage relationships, determine material batch contribution relationships, actual glazing evidence, and actual firing experience, verify cross-stage relationships and reconstruct candidate relationships in case of anomalies, correct correlation boundaries based on defect clustering intervals, and output the reverse source range and positive influence range along the same evidence chain that has not failed.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: The ceramic body glaze firing inspection correlation traceability method and system, based on the identity, time, location, material and equipment records left by the product at each production stage, connects the body, glaze, glazing, firing and inspection information in the same evidence chain, and retains the candidate range and unresolved status for relationships that cannot be determined temporarily, so that the traceability results are closer to the materials actually used by the product and the actual production process experienced, as shown below.

[0017] 1. During the batch changeover of blanks or glazes, the contribution relationship between the old and new batches is determined based on the material balance, input, return, output and product usage. Container mixing, pipeline residue and material delivery time are also taken into account in the judgment. This can reduce the misjudgment of the source caused by dividing material batches solely based on the batch changeover time, and can also explain why the transitional product is related to one or more material batches.

[0018] 2. Based on the actual time, location, device status, and available glaze materials of the product passing through the glazing device, determine the actual glazing evidence, and form the actual firing experience based on the kiln spatiotemporal path or loading topology; when damage, rejection, line merging, line splitting, missed reading, rework, or re-firing occurs, stop using the original relationship affected, and then establish a limited candidate set based on the reliable anchor points before and after the anomaly. This can reduce mismatches caused by judging the product experience according to the planned record, and reduce the possibility of a wrong relationship being continuously transmitted to subsequent production stages or a missed reading causing the entire traceability chain to be interrupted.

[0019] 3. Compare the boundary of the defect cluster area with the batch change transition boundary, firing anomaly boundary, and candidate relationship boundary. For the formed correction candidates, recheck the material balance, glazing path, firing path, and cross-stage relationship. Then, trace the reverse source and the positive impact range along the same unfailed evidence chain. This can reduce new errors caused by directly modifying the production relationship based solely on the defect distribution, and reduce the situation where the affected product range is determined to be too large, too small, or the location is off. At the same time, it is convenient for on-site personnel to distinguish between the confirmed, still need to be verified, and traceability results that cannot be determined by the current records. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall method flow of the present invention; Figure 2 This is a schematic diagram illustrating the system composition and data relationships of the present invention; Figure 3 This is a schematic diagram of the smallest traceability unit and cross-stage evidence chain of the present invention; Figure 4 This is a schematic diagram illustrating the contribution relationship between the batch change of the blank and the glaze in this invention; Figure 5 This is a schematic diagram illustrating the establishment of actual glazing evidence for this invention; Figure 6 This is a schematic diagram reconstructing the actual firing process of the present invention; Figure 7 This is a schematic diagram illustrating the verification, failure, and repair of cross-stage relationships in this invention; Figure 8 This is a schematic diagram illustrating the feedback correction of the evidence chain based on the test results of this invention. Figure 9 This is a schematic diagram of the bidirectional traceability and comparison verification design of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Please see Figures 1-9 The present invention provides the following technical solution: In this embodiment, the ceramic body glaze firing inspection and traceability system includes a data acquisition module, a traceability unit management module, a batch change contribution module, a glazing performance module, a firing experience module, a relationship verification module, a testing feedback module, and a traceability output module. Each module exchanges processing results through the same data storage and organizes data according to a unified traceability unit identifier and time benchmark, such as... Figure 2 As shown, each system module works collaboratively around the same data storage. The data acquisition module provides identity anchors, material supply, equipment status, loading location, and inspection results. Other modules use this information to complete the management of the smallest traceability unit, the determination of batch change contribution, the formation of actual glazing evidence, the reconstruction of actual firing experience, relationship verification, inspection feedback, and traceability output. The data storage preserves production stage nodes, cross-stage relationships, the basis for relationship formation, supporting records, conflict records, failure causes, and evidence status, enabling the green body, material supply, glazing, firing, and inspection records to be continuously organized along the same association chain.

[0023] The data acquisition module receives identity anchor points, workstation time, location, material supply, equipment status, loading location, and inspection results. The traceability unit management module uses this information to establish the smallest traceability unit and candidate relationships between adjacent production stages. The batch change contribution module determines the batch change transition interval and material batch contribution relationships based on material balance, input quantity, return quantity, output quantity, and product usage. The glazing performance module combines the actual time, location, device status, and available glaze material of the traceability unit passing through the glazing device to form actual glazing evidence. The firing experience module forms the actual firing experience based on the kiln's spatiotemporal path or loading topology. The relationship verification module verifies the data according to time sequence... The physical accessibility, equipment status, and material balance are checked across stages. When conflicts or anomalies occur, the original relationships are invalidated, and candidate relationships are reconstructed based on the identity, quantity, sequence, and location anchors before and after the event. The detection feedback module forms defect clustering intervals based on inspection results, compares its boundaries with batch change transition boundaries, firing anomaly boundaries, and candidate relationship boundaries to generate correction candidates and trigger re-checking. The traceability output module outputs the source range obtained by reverse investigation of defective products and the impact range determined by abnormal material batches or abnormal process intervals along the same evidence chain that has not been invalidated, along with candidate relationships, unresolved causes, and on-site verification information.

[0024] like Figure 1 As shown, the ceramic body glaze firing inspection correlation traceability method of this embodiment executes S1 to S6 in sequence. First, the smallest traceability unit and cross-stage candidate relationship are established. Then, the material batch contribution relationship, actual glazing evidence and actual firing experience are determined. Subsequently, the cross-stage relationship is verified and repaired. Finally, the correlation boundary is corrected based on the defect clustering interval and the bidirectional traceability result is output. The specific implementation is as follows.

[0025] S1. As Figure 3 As shown, the minimum traceability unit is established using a single product or a continuous group of blanks that maintains internal order. The identity, time, and location anchor points of each production stage are obtained, and cross-stage candidate relationships are established. In specific implementation, the system creates the minimum traceability unit at the forming process or the earliest station where products can be stably identified. When the conditions for identifying each product are met, a single ceramic product is used as a unit. When only continuous identification is possible but not stable identification of each product, a group of blanks with continuous forming time, adjacent positions, and maintaining internal order in subsequent processes is used as a unit. At the same time, the quantity and arrangement direction within the group are recorded. In S1, the anchor points include unit identifier, station identifier, arrival or departure time, channel or position, adjacent units, quantity, and equipment status. Adjacent stages first establish and save candidate relationships with formation basis. When damage, rejection, merging, splitting, missed reading, identifier deformation, rework, or re-firing occurs, a finite candidate set is formed based on the identity anchor points before and after the event, quantity conservation, order, and positional adjacency relationship.

[0026] The system generates stage nodes for each of the following stages: blank preparation, glazing, kiln entry, kiln exit, and inspection. Each stage node is indexed by a unit identifier and includes the corresponding workstation identifier, arrival or departure time, channel or location, adjacent units, quantity, and equipment status, while also saving the record source. For adjacent stage nodes of the same minimum traceability unit, the system first establishes candidate relationships based on identifier correspondence, time sequence, location reachability, and production order. Simultaneously, it saves the identity, time, location, and equipment records used to form the relationship. Before completing subsequent verification, the candidate relationship is not directly confirmed as a fixed one-to-one relationship. This ensures that each cross-stage candidate relationship can explain which field records were formed. If discontinuous time or impossible location is found later, the corresponding stage node and original record can be directly located, reducing mismatches caused by forced association based on the entire batch or a single time window.

[0027] After an abnormal event occurs, the system extracts the last reliable anchor point before the event and the first reliable anchor point after the event, using the time and location of the abnormality as boundaries. Within the product range defined by the two anchor points, impossible correspondences are eliminated item by item according to quantity conservation, sequence, channel flow, and positional adjacency. If only one candidate relationship remains and there are no record conflicts, the candidate relationship is retained. If multiple relationships remain, they are saved side by side as a finite candidate set, and the basis for their formation is recorded. If a key record is missing and the candidate range cannot be limited, the reason for the need for review is recorded. This processing method helps to limit uncertainty to a period of products before and after the abnormality. Even if there are omissions, rejections, or changes in sequence in the middle, the candidate range that can continue to be checked can be retained, reducing the transmission of erroneous relationships to subsequent processes.

[0028] The smallest traceability unit, stage node, identity anchor, candidate relationship, abnormal event record and finite candidate set formed by S1 are written into the same data storage and used as a unified index for subsequent determination of material batch contribution relationship, actual glazing evidence, actual firing experience and inspection association.

[0029] S2. Identify batch change events for blanks or glazes, and determine the transition range and material batch contribution relationship based on material balance, input quantity, return quantity, output quantity, and product usage. The system records the material being supplied before the batch change as the old batch and the material continuously added after the batch change as the new batch. The system uses the occurrence time of silo switching, formula switching, mixing tank switching, glaze tank replenishment, supply pipeline switching, or manual confirmation records as the calculation starting point, and then combines this with the actual material flow after the batch change to determine the material source corresponding to each smallest traceability unit, such as... Figure 4As shown in S2, when the materials in the feeding container are approximately fully mixed, the remaining amount and feeding ratio of each batch are recursively calculated based on the initial remaining amount, added amount, returned amount and output amount of each batch of materials according to the sampling interval. The participation amount and normalized contribution ratio of each batch are determined by combining the effective material consumption rate or adhesion rate of the smallest traceability unit. When the input data is an interval value, the interval recursion results are saved and candidate contribution relationships are formed.

[0030] like Figure 4 As shown, whether to adopt a fully mixed recursive method is determined by combining the structure diagram of the glaze tank or mixing tank, the agitator operation record, the flow direction of the circulating pump, the position of the glaze return inlet and outlet, and the batch change sampling record. When the material difference between different sampling positions at the same time is within the allowable range predetermined by the production line, the fully mixed recursive method is adopted. When there are stable sequential arrivals, stratification, or partitioning phenomena, the capacity integration zone method or the average residence time method is adopted. When the batch of returned material cannot be distinguished, the amount of returned material is used as the interval input and the candidate contribution relationship is retained.

[0031] Under the above-mentioned conditions of thorough mixing, the remaining quantity, supply ratio, and contribution to the smallest traceability unit of each material batch are calculated according to the following relationship: M(k) = Σ_b M_b(k) M_b(k+1)=M_b(k)+I_b(k)+R_b(k)-O(k)×M_b(k) / M(k) p_b(k) = M_b(k) / M(k) m_u,b=Σ_k[q_u(k)×p_b(k)×Δt_k] c_u,b=m_u,b / Σ_b m_u,b In the formula, b represents the material batch, k represents the sampling interval, M_b(k) represents the amount of material of batch b in the container at the beginning of the sampling interval, M(k) represents the effective total amount of material of each batch in the container at that moment, I_b(k) represents the amount of material of batch b added in the sampling interval, R_b(k) represents the amount of material of batch b returned to the container in the sampling interval, O(k) represents the amount of material output or consumed in the sampling interval, p_b(k) represents the proportion of batch b in the feeding, u represents the minimum traceability unit, q_u(k) represents the effective material consumption rate or adhesion rate of the minimum traceability unit u in the sampling interval, Δt_k represents the duration of the sampling interval, m_u,b represents the participation of batch b in the minimum traceability unit u, and c_u,b represents the normalized batch contribution ratio.

[0032] When only the upper and lower limits of liquid level, flow rate, return material amount, or product usage can be determined, the system will use the corresponding data as interval input, so that the remaining amount, supply ratio, participation amount, and normalized contribution ratio of each batch continue to be pushed forward in the form of intervals. When the existing data is insufficient to support the precise ratio, the contribution results can be saved in an ordered state of old batch as the main component, old batch and new batch participating together, or new batch as the main component. At the same time, the material batches that may participate and the data source are retained. This can preserve the situation of gradual mixing and replacement of materials during the batch change process, and reduce the misjudgment of product source caused by directly dividing the batch according to the batch change time.

[0033] When using contribution status to express the status, two configurable boundaries are set to satisfy 0≤θ_L<θ_H≤1. When the contribution ratio of the old batch is not less than θ_H, it is recorded as the old batch being the main batch. When the contribution ratio of the old batch is greater than θ_L and less than θ_H, it is recorded as the old batch and the new batch participating together. When the contribution ratio of the old batch is not greater than θ_L, it is recorded as the new batch being the main batch.

[0034] θ_L and θ_H are calibrated by continuously changing batches of samples. The contribution ratio of each sample from the old batch is correlated with the batch judgment results of color, composition, solid content or on-site approval. The boundaries for entering the new batch-dominant state and maintaining the old batch-dominant state are determined respectively. When the samples are insufficient to form a stable boundary, the ordered state and proportion range of old batch-dominant, co-participation and new batch-dominant are output. Fixed boundary values ​​are not filled in.

[0035] In S2, when the conveying pipeline is close to the first-in-first-out or piston flow state, the average residence time of the material is determined according to the effective volume of the pipeline and the average net flow rate of the corresponding time period. The outlet material is initially matched with the corresponding inlet feeding batch, and the batch change transition interval is expanded according to the pipeline residue, flow dispersion and cleaning and emptying records.

[0036] For conveying pipelines approaching FIFO or piston flow, the average residence time of materials is calculated as τ = V_eff / Q_avg, where τ represents the average residence time of materials from the feed inlet to the outlet, V_eff represents the effective volume of the pipeline involved in the conveying, and Q_avg represents the average net flow rate for the corresponding time period. The material output at the outlet at time t is first corresponding to the inlet feed batch at approximately time t-τ. Then, based on the pipeline residue that cannot be emptied, the dispersion of materials before and after the flow process, and the cleaning and emptying records, the single corresponding time is extended forward or backward into a batch change transition interval. This allows the conveying delay experienced by the material from the inlet to the outlet to be included in the calculation, making the actual batch supplied at the outlet correspond more closely to the products passing through at that time.

[0037] V_eff is determined based on the geometric volume of each section of the pipeline from the material inlet to the glazing outlet and the residual volume that cannot be emptied. Q_avg is preferentially used to estimate the average net flow rate recorded by the flow meter in the corresponding time period. If the flow meter is missing, it is estimated by combining the material replenishment amount, liquid level change and glaze return amount, and the candidate interval of τ is expanded according to the data error.

[0038] When the material supply path involves both container mixing and pipeline transportation, the fully mixed recursive method and the average residence time method are used respectively according to the material state of each segment, and then the results of each segment are connected according to the material flow sequence. When the liquid level, flow rate, cumulative usage and actual sampling support each other, the corresponding contribution relationship is recorded as verified. When only equipment capacity and production sequence support each other, it is recorded as candidate. When there is data missing, unclear source of return material or conflict of cleaning records, it is recorded as unresolved and the on-site verification item is saved.

[0039] The final output of S2 is the contribution relationship between the smallest traceability unit and one or more blank batches and glaze batches, the batch transition boundary, the candidate batch set, the corresponding data source, and the evidence status.

[0040] S3. Based on the time, location, device status, and available glaze material of the traceability unit passing through the glazing device, actual glazing evidence is formed. In specific implementation, the system determines the glazing line, spray gun, roller, glazing device, or other glazing device actually passed by the smallest traceability unit based on the arrival time, departure time, channel, running direction, and adjacent units of the smallest traceability unit at the glazing station. Equipment tasks and planned production records are only used to provide a range of candidate devices. The final glazing relationship must also match the actual station time and location records. In S3, the actual glazing candidate relationships are checked in sequence: the passage of the smallest traceability unit intersects with the effective working range of the glazing device within the calibrated time tolerance, the location is accessible to the device, the device is running and not in a cleaning or disabled state, and there is available glaze material during the intersection period; when there is only one candidate and no conflict, it is bound; when there are multiple candidates, they are retained in parallel; when key records are missing and a finite candidate set cannot be formed, it is recorded as unresolved; when there is a conflict in time, location, device status, or material supply, the relationship is invalidated; when rework or repeated glazing is performed, the glazing experience is added.

[0041] The calibration time tolerance is determined by the clock synchronization error of each acquisition device, the scanning or visual recognition delay, the station cycle time and the actuator response time. The channel and device position mapping record includes the channel number, the reachable glazing device, the split or parallel flow position and the running direction. When there are multiple candidates, the station direct reading record, equipment operation log, position continuity, front and rear unit sequence and quantity conservation are checked in turn. If they still cannot be distinguished, the candidates are retained in parallel.

[0042] like Figure 5As shown, the system first forms the actual passage interval based on the records of the smallest traceability unit entering and leaving the glazing area. Then, it obtains the glaze tank batch, liquid level, material flow rate, glaze return status, start-up and shutdown status, cleaning records, batch change records, rework records, and repeated glazing records corresponding to the glazing device within this interval. The glaze contribution interval formed by S2 intersects with this actual passage interval, and the glaze contribution that can be supplied by the corresponding device within the intersection period is used as the actual glazing candidate. In this way, the time when the product actually passes through the glazing device can be matched with the glaze that can actually be supplied at that time, reducing the deviation caused by directly judging the source of glaze according to the planned formula, work order batch, or equipment task.

[0043] When forming a candidate set, the system first reads the actual traversed range of the smallest traceability unit, filters the reachable glazing devices according to the channel, location, and direction of operation, and then excludes devices that are in shutdown, cleaning, or disabled periods. The effective working range of the remaining glazing devices is cross-checked with the glaze contribution range. When there is only one relationship in the candidate set and there is no conflict, it is bound as the actual glazing relationship of the smallest traceability unit. When there are multiple valid relationships, they are retained in parallel. When there is a clear conflict in time, location, device status, or material supply record, the corresponding relationship is invalidated. When a key record is missing and the scope cannot be limited to a limited number of candidates, it is recorded as unresolved. When reworking or repeating glazing, a new glazing experience with actual time, glazing device, and glaze contribution is added, and the original experience is retained. In this way, the verifiable actual range can still be retained in the case of paralleling, reworking, restarting after shutdown, or missing records, reducing the situation of connecting the product to the wrong glaze batch due to forcibly selecting a single glazing relationship.

[0044] S3 ultimately outputs the actual glazing device, the actual glazing time interval, the corresponding glaze contribution relationship, rework or repeated glazing records, the basis for the relationship formation and the status of evidence, and binds the above content to the corresponding smallest traceability unit.

[0045] S4. Based on the kiln's spatiotemporal path or loading topology, the actual firing experience is formed. The system reads the kiln entry anchor point, entry sequence, channel, product spacing, loading position, and firing cycle of the smallest traceability unit, and obtains the kiln operating speed, section length, temperature, pressure, atmosphere, combustion status, and abnormal event records. The actual firing position and firing period of the smallest traceability unit are reconstructed according to the continuous operation mode or loading operation mode. In S4, when using a continuous roller kiln, the kiln position is recursively deduced based on the kiln entry sequence, channel, product spacing, section length, and speed records. The kiln is divided into sections at the boundaries of kiln stop, speed reduction, brick chasing, brick jamming, empty space, or manual intervention. The entry and exit times of each section are combined with the process records of the same period to form the actual firing experience. When using kiln cars, saggers, or firing plates for loading, a loading topology is established according to the kiln car number, layer, grid position or sagger position, loading event, and firing cycle. The loading and unloading records are used to jointly verify the position node of the smallest traceability unit.

[0046] like Figure 6 As shown, for a continuous roller kiln, the system recursively calculates the cumulative position of the minimum traceability unit within the kiln according to the sampling interval of the speed record, and determines the segment in which the minimum traceability unit is located in each sampling interval based on the segment boundary: s_u(k)=s_u(k-1)+v_c(k)×Δt_k L_z,start≤s_u(k) <L_z,end In the formula, s_u(k) represents the cumulative displacement of the minimum traceability unit u relative to the kiln entry position at the end of the kth sampling interval, s_u(k-1) represents the cumulative displacement of the minimum traceability unit at the end of the previous sampling interval, c represents the channel where the minimum traceability unit is located, v_c(k) represents the measured speed of channel c in the kth sampling interval or the effective speed determined by the equipment recording, Δt_k represents the duration of the sampling interval, z represents the kiln section, and L_z,start and L_z,end represent the start and end points of section z relative to the kiln entry position, respectively.

[0047] Kiln speed is primarily taken from the drive encoder, frequency converter, or PLC operation log. Multiple channels driven by the same drive mechanism and verified on-site to have no independent slippage share the speed record. The effective length of each section is determined based on the kiln layout diagram and on-site measurement of the kiln entry reference point. The sampling frequency of speed and process logs should cover kiln shutdown, speed reduction, and recovery events. When the recording interval is too large, the kiln is segmented according to the event boundary, and the time period that cannot be further subdivided is saved as a candidate interval.

[0048] When a kiln shutdown occurs, the system records the effective speed within the corresponding sampling interval as zero. When a speed reduction occurs, the speed record of the corresponding time period is used to continue the recursion. When there are issues such as brick chasing, brick jamming, empty spaces, manual intervention, or abnormal section parameters, the start and end times of the event are used as the segment boundaries to determine the entry time, exit time, and location of the smallest traceability unit in each segment. Then, the temperature, pressure, atmosphere, or combustion status records of the same period are combined into the corresponding segment, thus forming the actual firing experience in the continuous kiln. This allows for the differentiation of the actual time and dwell time of different products in each segment under the same kiln setting conditions, making it easier to identify which segment of the product is actually affected by the kiln shutdown, speed reduction, or local anomaly.

[0049] When using kiln cars, saggers, or firing plates for loading, the system establishes a hierarchical loading topology based on kiln car number, layer, grid or sagger number, loading direction, loading event, firing cycle, and location verification records. The smallest traceability unit is bound to a location node that is jointly verified by loading and unloading records, and the loading, kiln entry, kiln exit, and unloading times are recorded. When the production line has sample pieces or temperature measuring rings, the carrier number, placement location, bound product range, placement and retrieval time points, and test results are recorded simultaneously, and these are used as location-level firing witnesses. When the production line does not have such witness carriers, the loading and unloading sequence and adjacent product records are used for optional verification. Firing witnesses are not set as a necessary condition for forming an actual firing experience. This allows products in different layers, grids, and sagger positions within the same kiln car or the same firing cycle to be distinguished, reducing the expansion of the traceability scope caused by treating all products in a kiln car as having the same firing experience.

[0050] When the kiln entry identification record is missing, the system establishes candidate firing relationships based on the last identity anchor point before entering the kiln, the first identity anchor point after exiting the kiln, the quantity changes before and after entering the kiln, the accessibility of the passage, and the location adjacency relationship. When the candidate relationships can be distinguished, they are bound to the corresponding spatiotemporal path or loading position. If there are still multiple reasonable correspondences, multiple actual firing experience candidates are retained in parallel and the reasons for verification are recorded.

[0051] S4 ultimately outputs the kiln path or loading position of the smallest traceability unit, the actual entry and exit times of each section, the concurrent process records, abnormal events, the basis for location verification, and the status of evidence. The above content will be used as firing evidence for subsequent cross-stage relationship verification and defect cause tracing.

[0052] S5. Linking inspection records, the system verifies cross-stage relationships based on time sequence, physical accessibility, equipment status, and material balance. In case of conflicts or anomalies, the original relationship is invalidated, and candidate relationships are reconstructed according to the anchor points before and after the event. The system binds inspection records to the smallest traceability unit after kiln exit. Inspection records include defect type, defect location, quality grade, inspection time, and inspection station. When image detection is available, defect areas or defect features can also be saved, and image detection results are used as one of the supporting records for cross-stage relationships. Figure 7 As shown in S5, the evidence status of cross-stage relationships includes verified, candidate, pending, and invalid. Verified is defined as when all verification conditions are met and the independent records are complete; candidate is defined as when all verification conditions are met but the independent records are incomplete; pending is defined as when key records are missing and a finite set of candidates cannot be formed; invalid is defined as when a clear conflict occurs. When an anomaly occurs, the original relationship affected by it becomes invalid, the relationship is rebuilt as a candidate, and the original relationship, original status, supporting records, conflict records, and reasons for change are retained.

[0053] The system performs unified verification of cross-stage relationships between the green body and glazing, glazing and kiln entry, kiln entry and firing, and firing and inspection. Among them, time sequence verification is used to determine whether the departure time of the previous stage is earlier than the arrival time of the next stage; physical accessibility verification is used to determine whether the channel, transfer direction or loading position can form a corresponding connection; equipment status verification is used to determine whether the relevant equipment is in an allowable operating state during the period when the relationship occurs; and material balance verification is used to determine whether the product quantity, material balance, input quantity, output quantity and product usage are consistent with each other. At the same time, it checks for any records of damage, rejection, paralleling, splitting, missing reading, rework or re-firing that conflict with the relationship. In this way, relationships with reversed time, inaccessible location or inconsistent material quantity can be eliminated in time during the evidence chain formation process, reducing the possibility of a single erroneous record causing multiple subsequent production stages to be linked incorrectly.

[0054] Independent records refer to records generated by different formation mechanisms and capable of mutual verification, including identity reading records, equipment operation logs, material receipt and expenditure records, location or sequence records, on-site sampling records, and manual verification records. Copies or transferred copies of the same original record are not considered new independent records. The required number of independent records is determined based on the reliability of various types of data. The system updates the evidence status according to the completeness of the currently obtained records. When new on-site sampling, firing witnessing, equipment logs, or manual verification records are added, the verification is re-executed. When new conflicting records appear or errors are found in the original records, the evidence status of the corresponding relationship is reduced or the relationship is invalidated, and a new relationship version is generated for each verification result.

[0055] When an abnormal event occurs, the system first determines the affected scope based on the event's start time, end time, involved channels, quantity changes, and upstream and downstream identity anchors. At the event boundary, it terminates the continued use of the affected original relationship. Then, starting from the last reliable anchor before the event and the first reliable anchor after the event, it re-searches for candidate relationships according to quantity conservation, production sequence, channel accessibility, and location adjacency. If there is only one relationship after reconstruction and all verification conditions are met, the candidate relationship is retained. If multiple reasonable relationships exist, they are saved in parallel. If a finite set of candidates cannot be formed, it is recorded as unresolved. When a one-to-many or many-to-one relationship is formed, all associated objects, the quantity distribution between each object, and the reasons for their formation are saved. The original relationship, original state, supporting records, conflict records, reconstruction results, and reasons for changes are all written into the relationship history. In this way, even if there are cases of missed reading, removal, merging, splitting, or rework, it can still explain why the original relationship stopped and how the replacement relationship was formed, which facilitates subsequent verification from the abnormal position without having to re-examine the entire production record.

[0056] S5 ultimately outputs the binding results of inspection records and the smallest traceability unit, the evidence status of each cross-stage relationship, supporting records, conflict records, the scope of abnormal impact, reconstructed candidate relationships and relationship change history, and saves the missing records corresponding to unresolved relationships and on-site review items.

[0057] S6. Form a defect clustering interval, compare its boundary with the batch change transition boundary, firing anomaly boundary, and candidate relationship boundary, generate correction candidates and re-verify, output the reverse source range and positive influence range along the same evidence chain that has not failed. In S6, arrange the smallest traceability units according to the production sequence, and connect adjacent defect units that meet the calibration conditions in terms of product serial number interval, inspection or production time interval, and defect type to the same judgment interval. When the calibration quantity condition is reached, a defect clustering interval is formed. Compare its boundary with the boundary of the batch change transition, firing anomaly, and candidate related interval. When the deviation exceeds the calibration allowable range, generate boundary candidates that move forward, backward, expand, shrink, or remain unchanged. For each candidate, re-execute the material balance, path, and cross-stage relationship verification. Defect evidence does not directly cover the original relationship.

[0058] The system first arranges the smallest traceability units according to the production sequence, and determines the same or related defect types that can be merged for judgment through the defect association table. When the product serial number interval between adjacent defect units is not greater than the calibrated serial number interval, the inspection or production time interval is not greater than the calibrated time interval, and the defect type meets the association conditions, the adjacent defect units are connected to the same judgment interval. When the number of defect units in the judgment interval reaches the calibrated quantity condition, a defect cluster interval is formed, and the starting boundary, ending boundary, internal defect distribution, product sequence, and inspection time of the interval are saved. Defects that do not meet the quantity condition are retained as single events or smaller candidate intervals.

[0059] Let G represent the maximum allowable product sequence number interval between adjacent defective units, ΔT represent the maximum allowable inspection time or production interval, and N represent the number of defective units required to form a defect cluster interval. G is calibrated based on the adjacent product sequence number interval under stable production conditions, ΔT is calibrated based on production cycle time, inspection lag, and transfer waiting time, and N is calibrated based on the distinction between historical defect intervals and isolated defects. The defect association table simultaneously records the defect type, location of occurrence, and possible shared material mechanism or firing mechanism. It allows intermittent rules to record the number of normal products that can be crossed within the interval and the applicable scenarios. When the sample is insufficient, it outputs a candidate cluster interval with a wider range.

[0060] like Figure 8As shown, the system compares the starting and ending boundaries of the defect clustering interval with the corresponding boundaries of the batch transition interval for billet or glaze, the firing anomaly interval, and the candidate correlation interval, respectively. The displacement of the starting boundary equals the starting index of the defect clustering interval minus the starting index of the candidate cause interval, and the displacement of the ending boundary equals the ending index of the defect clustering interval minus the ending index of the candidate cause interval. When the product serial number cannot be stably matched, the boundary displacement is calculated using the corresponding timestamp. When both boundary displacements are within the calibrated allowable range, the defect interval is used as mutually supporting verification evidence; if the boundary deviation exceeds... When the boundary falls outside the allowable range, candidate boundaries are generated for forward, backward, expansion, reduction, or no change. For each candidate, the material balance, glazing path, firing path, and cross-stage relationship checks from S2 to S5 are re-executed. Multiple candidates are retained in parallel during the check without directly overwriting the original relationship with defect evidence. This allows the start and end positions of defects found at the inspection end to help check whether the batch boundary and abnormal boundary at the front end are too early or too late. At the same time, the actual material, location, and equipment records still determine whether the relationship is valid, reducing new errors caused by directly modifying production records based solely on defect distribution.

[0061] When the production sequence can be maintained continuously, the product sequence number difference is used first. When there is paralleling, splitting, buffer accumulation, or incomplete sequence recording, the time difference is used instead. The allowable range of consistent boundaries is determined by a combination of identity reading error, sampling interval, conveying cycle time, and event recording delay calibration. Only when the boundary displacement with the same direction repeatedly appears in multiple independent batch change samples or abnormal samples and passes the physical verification again, is the displacement used as a stable offset for adjustment.

[0062] Each round of calibration saves the boundary positions, candidate relation sets, and relation statuses before and after adjustment. Calibration stops when any of the following conditions are met: the number of consecutive iterations determined based on the duration of historical sporadic fluctuations, sufficient on-site evidence to confirm the relation, existing data to no longer distinguish candidates, or the number of iterations reaches the configurable upper limit N_max. N_max is set according to the number of convergence rounds of the test samples. If multiple candidates still exist after reaching the upper limit, an unresolved status is output. When there is a clear physical conflict, missing key records, or the scope of impact exceeds the enterprise's review scope, on-site review is triggered, and the reasons for the review and on-site review items are output.

[0063] In S6, each stage record is represented as a node, and cross-stage relationships with evidence status and supporting records are represented as edges. Reverse tracing returns to the source of materials, glazing, and firing from defective nodes along non-failed edges, while forward tracing reaches the smallest traceability unit that may be affected from abnormal material batches or abnormal process intervals along the same set of edges. Candidate edges and unresolved edges are retained and marked with the reasons to be verified, while failed edges stop participating in traversal and are saved in the relationship history.

[0064] like Figure 9As shown, in specific tracing, reverse tracing starts with a single defective product or defect cluster, sequentially passing through inspection relationships, actual firing experiences, actual glazing evidence, and material batch contribution relationships, outputting the possible corresponding raw material batch, glaze batch, glazing device, glazing period, firing section, and abnormal firing period; forward tracing starts with abnormal material batches, batch change transition periods, abnormal glazing devices, or abnormal kiln periods, finding the smallest traceable unit and product range that may be affected along the same evidence chain, and listing the relationship paths traversed, the evidence status of each relationship, supporting records, conflict records, and on-site verification items in the output. In this way, users can both trace possible sources from defective products backward and delineate potentially affected products from known anomalies, while also distinguishing between confirmed ranges, candidate ranges that still need to be verified, and ranges that cannot be determined by current data.

[0065] In one specific embodiment, the glaze supply system uses a glaze tank with a nominal capacity of 800 L and an effective working capacity of 600 L. At 10:00, the old batch A is switched to the new batch B. The initial balance of batch A at the time of batch switch is 600 L, and the sampling interval is 10 min. From 10:00 to 10:10, the new batch B is replenished with 120 L, the old batch A is returned with 20 L of glaze, and the glazing output is 140 L. The corresponding average replenishment flow rate is 12 L / min, the return flow rate is 2 L / min, and the output flow rate is 14 L / min. According to the fully mixed recursive relationship of S2, at 10:10, M_A(1) = 480 L, M_B(1) = 120 L, p_A(1) = 0.80, and p_B(1) = 0.20. The effective volume V_eff of the glaze supply pipeline is 42 L, and the average net flow rate Q_avg is 14 L. L / min, according to τ=V_eff / Q_avg, the average dwell time τ is 3min. In this embodiment, the contribution state boundary is θ_L=0.20, θ_H=0.80.

[0066] In this embodiment, traceability units U21 to U32 sequentially pass through the glazing device G1 in channel 1 from 10:09:30 to 10:15:00, and sequentially enter the continuous roller kiln in the same channel from 10:18 to 10:24. The normal kiln speed of this kiln is 1.20 m / min. The kiln stops from 10:40 to 10:45. The effective lengths of the preheating zone, firing zone, and cooling zone are 36 m, 24 m, and 48 m, respectively. At the start of the kiln stoppage, U21 to U32 are located in the preheating zone, which is 19.2 m to 26.4 m from the kiln inlet. The traceability units that detect pinhole defects are U21, U22, U24, U25, U27, U28, U30, U31, and U32. The time interval between adjacent defect inspections does not exceed 4 minutes, and the defect correlation parameters are G=2 and ΔT=4. With min and N=6, the defect clustering interval from U21 to U32 is formed, and the candidate cause interval is from U19 to U30. The initial boundary displacement and the final boundary displacement are both +2.

[0067] In the on-site verification of this embodiment, the batch change operation record was executed at 10:00. The flow meter recorded the cumulative material replenishment and glazing output from 10:00 to 10:10 as 120 L and 140 L respectively, and the glaze return metering record was 20 L. The effective margin change calculated from the glaze tank level was consistent with the aforementioned material inflow and outflow. The workstation reading record showed that U21 to U32 all passed through the glazing device G1 in channel 1. The kiln speed log recorded the kiln stoppage period from 10:40 to 10:45. The inspection photos, product serial numbers, and verification signature sheets corresponded to the aforementioned pinhole defect sequence. After re-executing S2 to S5, no hard conflicts were found in time, location, equipment status, and material balance. Therefore, the current relationship was saved as a candidate, and a decision on whether to update it to verified would be made after obtaining the sample component test results.

[0068] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for tracing and linking the firing inspection of ceramic bodies and glazes, characterized in that, include: S1, establish the smallest traceability unit with a single product or a continuous blank group that maintains internal order, obtain the identity, time and location anchor points of each production stage and establish cross-stage candidate relationships; S2 identifies batch change events for blanks or glazes, and determines the transition range and material batch contribution relationship based on material balance, input amount, return amount, output amount and product usage. S3, based on the time, location, device status and available glaze material of the traceability unit passing through the glazing device, actual glazing evidence is formed; S4, based on the kiln's spatiotemporal path or loading topology, forms the actual firing experience; S5, Related Inspection Records, checks cross-stage relationships based on time sequence, physical accessibility, equipment status and material balance. When conflicts or anomalies occur, the original relationship is invalidated and candidate relationships are rebuilt according to the anchor points before and after the event. S6 forms a defect clustering region. By comparing its boundary with the batch transition boundary, firing anomaly boundary, and candidate relationship boundary, correction candidates are generated and re-verified. The reverse source range and positive influence range are output along the same evidence chain that has not failed.

2. The method for tracing and correlation of ceramic body and glaze firing inspection according to claim 1, characterized in that, In S1, the anchor points include unit identifier, workstation identifier, arrival or departure time, channel or location, adjacent units, quantity and equipment status; in the adjacent stage, candidate relationships with formation basis are first established and saved. When damage, rejection, merging, splitting, missing reading, identifier deformation, rework or reburning occurs, a finite candidate set is formed according to the identity anchor points before and after the event, quantity conservation, order and position adjacency relationship.

3. The method for tracing and correlation of ceramic body and glaze firing inspection according to claim 1, characterized in that, In S2, when the materials in the feeding container are approximately fully mixed, the remaining amount and feeding ratio of each batch are recursively calculated based on the initial balance, added amount, returned amount and output amount of each batch of materials according to the sampling interval. The participation amount and normalized contribution ratio of each batch are determined by combining the effective material consumption rate or adhesion rate of the smallest traceability unit. When the input data is an interval value, the interval recursion results are saved and candidate contribution relationships are formed.

4. The method for tracing and correlation of ceramic body and glaze firing inspection according to claim 1, characterized in that, In S2, when the conveying pipeline is close to the first-in-first-out or piston flow state, the average residence time of the material is determined according to the effective volume of the pipeline and the average net flow rate of the corresponding time period. The outlet material is initially matched with the corresponding inlet feeding batch, and the batch change transition interval is expanded according to the pipeline residue, flow dispersion and cleaning and emptying records.

5. The method for tracing and correlation of ceramic body and glaze firing inspection according to claim 1, characterized in that, In S3, the actual glazing candidate relationships are checked sequentially: the passage interval of the smallest traceability unit intersects with the effective working interval of the glazing device within the calibrated time tolerance, the location is reachable from the device, the device is running and not in a cleaning or disabled state, and there is available glaze during the intersecting period; when there is only one candidate and there is no conflict, they are bound together; when there are multiple candidates, they are retained in parallel; when key records are missing and a finite candidate set cannot be formed, they are recorded as unresolved; when there are conflicts in time, location, device status, or material supply, the relationship is invalidated; when rework or repeated glazing is performed, the glazing experience is added.

6. The method for tracing and correlation of ceramic body and glaze firing inspection according to claim 1, characterized in that, In S4, when using a continuous roller kiln, the kiln position is recursively calculated based on the kiln entry sequence, channel, product spacing, section length, and speed record. The kiln is divided into sections at the boundaries of kiln stop, speed reduction, brick chasing, brick jamming, empty space, or manual intervention. The entry and exit times of each section are combined with the concurrent process records to form the actual firing experience. When using kiln cars, saggers, or firing plates for loading, a loading topology is established according to the kiln car number, layer, grid position or sagger position, loading event, and firing cycle. The loading and unloading records are used to jointly verify the position node of the smallest traceability unit.

7. The method for tracing and correlation of ceramic body and glaze firing inspection according to claim 1, characterized in that, In S5, the evidence status of cross-stage relationships includes verified, candidate, pending, and invalid. Verified is defined as when all verification conditions are met and the independent records are complete; candidate is defined as when all verification conditions are met but the independent records are incomplete; pending is defined as when key records are missing and a finite set of candidates cannot be formed; invalid is defined as when a clear conflict occurs. When an anomaly occurs, the original relationship affected by it becomes invalid, the relationship is rebuilt as candidate, and the original relationship, original status, supporting records, conflict records, and reasons for change are retained.

8. The method for tracing and correlation of ceramic body and glaze firing inspection according to claim 1, characterized in that, In S6, the smallest traceability units are arranged according to the production sequence. Adjacent defect units that meet the calibration conditions in terms of product serial number interval, inspection or production time interval, and defect type are connected to the same judgment interval. When the calibration quantity condition is reached, a defect cluster interval is formed. Its boundary is compared with the boundary of batch change transition, firing abnormality, and candidate related interval. When the deviation exceeds the calibration allowable range, a boundary candidate is generated to move forward, backward, expand, shrink, or remain unchanged. For each candidate, the material balance, path, and cross-stage relationship verification are re-executed. Defect evidence does not directly cover the original relationship.

9. The method for tracing and correlation of ceramic body and glaze firing inspection according to claim 1, characterized in that, In S6, each stage record is represented as a node, and cross-stage relationships with evidence status and supporting records are represented as edges. Reverse tracing returns to the source of materials, glazing, and firing from defective nodes along non-failed edges, while forward tracing reaches the smallest traceability unit that may be affected from abnormal material batches or abnormal process intervals along the same set of edges. Candidate edges and unresolved edges are retained and marked with the reasons to be verified, while failed edges stop participating in traversal and are saved in the relationship history.

10. A ceramic body glaze firing inspection and traceability system, characterized in that, The system is used to operate the method described in any one of claims 1-9, comprising a data acquisition module, a traceability unit management module, a batch change contribution module, a glazing performance module, a firing experience module, a relationship verification module, a detection feedback module, and a traceability output module. Each module collaboratively acquires identity anchor points, material supply, equipment status, loading location, and inspection results; establishes the smallest traceability unit and cross-stage relationships; determines material batch contribution relationships, actual glazing evidence, and actual firing experience; verifies cross-stage relationships and reconstructs candidate relationships in case of anomalies; corrects associated boundaries based on defect clustering intervals; and outputs the reverse source range and positive influence range along the same evidence chain that has not failed.

Images