A method and system for quality control of a rubber pad production process

Abstract

This invention discloses a quality control method and system for the production process of rubber liners. It acquires production parameter data and process configuration parameters, generates process matching coefficients through cross-process linkage analysis, statistically analyzes defect distribution and maps it to a process control list to establish a hierarchical quality control scheme. Based on the process matching coefficients, it weights and adjusts the hierarchical quality control schemes to form a comprehensive control scheme, identifies deviation transmission amplification and complete absorption anomalies to generate detection anomaly alarms, and determines the graded quality control benchmark. It executes hierarchical corrections to generate monitoring feedback tables, tracks the narrowing rate differences between shifts to generate quality control trend charts, and triggers false-qualification warnings when detection parameters quickly reach the standard to form a qualification judgment level. It adjusts the detection frequency to generate process qualification certificates, screens multi-process synchronous critical batches to form quality stability labels, and outputs rubber liner production quality control instructions, realizing structured control of cross-process and cross-shift quality risks in the rubber liner production process.

Description

Technical Field

[0001] This invention relates to the field of quality control technology in rubber product manufacturing, and in particular to a quality control method and system for the production process of rubber lining sheets. Background Technology

[0002] The production process of rubber linings involves multiple steps such as mixing, extrusion, calendering and vulcanization. The processing parameters of each step are interconnected. Fluctuations in parameters of upstream steps are transmitted to downstream through materials and accumulate. Product defects do not always appear at the point of occurrence. Traditional single-point sampling inspection of each step is difficult to capture the cross-step linkage failure pattern. Quality abnormalities are often only discovered at the end inspection, resulting in batch rework.

[0003] On the other hand, production sites typically involve multiple shifts, and differences in operating habits and equipment status across shifts cause defects to cluster in a periodic manner. Existing quality control methods lack the ability to continuously track the differences in the rate at which parameters narrow between shifts. When parameters suddenly return from an out-of-range state to the acceptable range, existing detection mechanisms classify them as normal and acceptable, ignoring the risks of tooling wear or data distortion, leading to falsely qualified products flowing into downstream processes. Summary of the Invention

[0004] This invention discloses a quality control method and system for the production process of rubber liners, aiming to solve problems such as the difficulty in identifying cross-process linkage failures, the inability to track the distribution patterns of defects caused by shift differences, and the false acceptance and missed detection due to abnormally rapid parameter compliance. This invention achieves structured quality control over the entire rubber liner production process by establishing a complete quality control chain, including inter-process linkage matching assessment, defect sudden change tracing and hierarchical quality control deployment, deviation transmission anomaly alarm and grading benchmark determination, shift difference tracking and false acceptance early warning identification, and multi-process synchronous critical screening and control instruction output.

[0005] The first aspect of this invention provides a quality control method for the production process of rubber lining sheets, comprising the following steps: Acquire production parameter data and process configuration parameters, and conduct cross-process linkage analysis on the production parameter data and the process configuration parameters to form process matching coefficients; Based on the production parameter data, defect severity statistics are performed to generate a defect heat map. Based on the process configuration parameters, process hierarchy is extracted to generate a process control list. The defect heat map is mapped to the process control list to locate suspicious processes with sudden defect changes and establish a step-by-step quality control plan. Based on the process matching coefficient, a weighted adjustment is applied to the step-by-step quality control scheme to form a comprehensive control scheme. The comprehensive control scheme identifies process sections where deviations are amplified and completely absorbed, and generates detection anomaly alarms. The failure distribution law is extracted according to the detection anomaly alarms to determine the graded quality control benchmark. The production parameter data is corrected step by step using the graded quality control benchmark to generate a monitoring feedback table. Deviation monitoring is performed on the monitoring feedback table to generate a quality control trend chart. The quality control trend chart is used to detect parameter anomalies, quickly reach the standard, trigger a false pass warning, and form a pass judgment level. Based on the aforementioned qualification level, the frequency of testing is adjusted to generate process qualification certificates. The process qualification certificates are then subjected to multi-process synchronous critical batch screening to form quality stability labels. Based on the quality stability labels, rubber lining production quality control instructions are output.

[0006] A second aspect of this invention provides a quality control system for the rubber liner production process, comprising: An adaptation analysis unit is used to acquire production parameter data and process configuration parameters, and to perform cross-process linkage analysis on the production parameter data and the process configuration parameters to form a process matching coefficient. The quality control scheme unit is used to perform defect severity statistics and generate a defect heat map according to the production parameter data, extract the process hierarchy based on the process configuration parameters and generate a process control list, map the defect heat map to the process control list to locate suspicious processes with sudden defect changes and establish a step-by-step quality control scheme. The alarm control unit is used to apply weighted adjustment to the step-by-step quality control scheme based on the process matching coefficient to form a comprehensive control scheme, generate detection anomaly alarms for process sections where deviations are amplified and deviations are completely absorbed by the comprehensive control scheme, and determine the graded quality control benchmark by extracting the failure distribution law according to the detection anomaly alarms. The deviation monitoring unit is used to perform step-by-step correction on the production parameter data through the graded quality control benchmark to generate a monitoring feedback table, perform deviation monitoring on the monitoring feedback table to generate a quality control trend chart, and use the quality control trend chart to detect parameter abnormalities, quickly reach the standard, trigger a false pass warning, and form a pass judgment level. The result output unit is used to adjust the detection frequency based on the qualified judgment level to generate process qualified certificates, perform multi-process synchronous critical batch screening on the process qualified certificates to form quality stability labels, and output rubber liner production quality control instructions based on the quality stability labels.

[0007] The beneficial effects of this invention are reflected in the following points: 1. By extracting the fluctuation distribution of key parameters of each process and verifying it in conjunction with the upstream and downstream transmission constraints of process configuration parameters, the invention identifies vulnerable processes with persistently large linkage differences and recent additions, and generates process matching coefficients, thereby achieving a quantitative assessment of structural linkage defects between processes. Based on this, it distinguishes between sudden increases in operational standard defects caused by shift change nodes and random increases caused by equipment failures, classifying and mapping these to the process control list, thus achieving accurate attribution of defect source types and differentiated quality control deployment. 2. Through continuous analysis of the deviation transmission ratio between adjacent processes in the comprehensive control scheme, it distinguishes and identifies two types of structural anomalies: transmission amplification and transmission complete absorption, and jointly generates detection anomaly alarms. It extracts the failure distribution patterns of primary and secondary failure sources to determine graded quality control benchmarks, establishing a complete mapping mechanism from deviation transmission behavior to graded response benchmarks, thus solving the problem of missed detection of transmission-type anomalies by traditional single-threshold alarms. 3. Trend fitting and continuous difference identification were performed on the rate of deviation narrowing between shifts to distinguish between occasional and systematic shift operation differences; for batches whose parameters suddenly returned to the qualified range, the historical frequency of rapid compliance was traced by tooling number to identify two types of risks: false compliance due to tooling wear and false compliance due to unknown causes, thus avoiding the misjudgment and release of falsely qualified products; further screening was conducted on batches whose boundary distances of multiple processes were simultaneously lower than the safety margin and a joint risk score was calculated. Based on the quality stability label level, four types of control instructions were output: suspension, enhanced monitoring, release, or suspension, thus realizing hierarchical closed-loop control of three types of risks: shift differences, false compliance, and system drift. Attached Figure Description

[0008] Figure 1 This is a flowchart illustrating a quality control method for the production process of rubber linings according to the present invention.

[0009] Figure 2 This is a structural block diagram of a quality control system for the production process of rubber lining sheets according to the present invention. Detailed Implementation

[0010] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0011] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0012] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0013] The technical solutions of the embodiments of this application are described below.

[0014] like Figure 1 As shown, this embodiment of the invention provides a quality control method for the production process of rubber lining sheets, including the following steps S110-S150: Step S110: Obtain production parameter data and process configuration parameters, and conduct cross-process linkage analysis on the production parameter data and process configuration parameters to form process matching coefficients.

[0015] Specifically, production parameter data and process configuration parameters are acquired. The collection scope of production parameter data covers four main processes: mixing, extrusion, calendering, and vulcanization. The mixing process collects three continuous time-series data items: mixing temperature, rotor speed, and mixing time. The extrusion process collects four items: extrusion pressure, die temperature, screw speed, and Mooney viscosity of the rubber compound. The calendering process collects four items: roller temperature, roller spacing, calendering speed, and sheet thickness. Each item is recorded at a frequency of once every 30 seconds, with the equipment number and shift identifier appended. Among the production parameter data, the vulcanization temperature and vulcanization time parameters in the vulcanization process have the greatest impact on the final product performance. The vulcanization temperature is set at 175 degrees Celsius, and the measured data is collected synchronously at eight temperature measurement points using a thermocouple array. The collection cycle for each batch is the entire vulcanization process for that batch. The production parameter data is organized and stored by batch number. Each batch record covers the complete time-series array of each process's data items, the batch feed amount, and the production date. The batch number is generated by concatenating the process code, production date, and serial number. The process configuration parameters define the standard operating procedures for each process, including the nominal values ​​of key parameters, allowable deviation ranges, and parameter transfer constraints between upstream and downstream processes. In the process configuration parameters, the nominal value of the die temperature for the extrusion process is 145 degrees Celsius, with an allowable deviation range of ±8 degrees Celsius. The nominal value of the roller spacing for the calendering process is set in conjunction with the nominal value of the Mooney viscosity of the rubber compound in the extrusion process, with a transfer constraint coefficient of 0.42. The parameter transfer constraints of the process configuration parameters are recorded in the form of upstream and downstream process pairs. Each constraint record includes the upstream process number, downstream process number, associated parameter name pair, and transfer constraint coefficient. A total of 6 sets of transfer constraints between upstream and downstream process pairs are recorded in the process configuration parameters.

[0016] In some embodiments, the step of performing cross-process linkage analysis on the production parameter data and the process configuration parameters to form a process matching coefficient includes: extracting the fluctuation distribution of key parameters of each process based on the production parameter data to generate a fluctuation distribution table; verifying the upstream and downstream process associations of the fluctuation distribution table and the process configuration parameters to generate a process association difference table; identifying processes in the process association difference table that significantly influence the downstream from the upstream to generate vulnerable linkage markers; and weightedly integrating the vulnerable linkage markers and the process association difference table to generate a process matching coefficient.

[0017] A fluctuation distribution table was generated by extracting the fluctuation distribution of key parameters for each process based on production parameter data. In the internal mixing process, the measured values ​​of the mixing temperature over nearly 30 batches ranged from 168 to 189 degrees Celsius, with a standard deviation of 4.7 degrees Celsius. The fluctuation distribution was characterized by the mean, standard deviation, range, and interquartile range of each parameter over time. In the extrusion process, the batch-to-batch range of extrusion pressure within the same shift was 1.8 MPa, while the inter-shift range was 3.2 MPa. The inter-shift range was significantly larger than the intra-shift range, indicating that shift changes contributed more to the extrusion pressure fluctuation than the random fluctuations of the equipment itself. The fluctuation distribution table was organized using a two-level index based on process number and parameter name, with a table size of 13 rows (4 processes multiplied by the sum of the number of data points collected for each process). In the production parameter data, the average standard deviation of the vulcanization temperature at the eight measurement points in the vulcanization process was 2.1 degrees Celsius, and the average batch-to-batch range of the temperature difference between measurement points was 5.8 degrees Celsius. The fluctuation distribution table uses the average of the statistics at each measurement point as the representative statistic for that parameter. Parameters in the fluctuation distribution table whose standard deviation exceeds 50% of the allowable deviation range (half-width) are marked as high-fluctuation parameters. The allowable deviation range for the mixing temperature in the internal mixing process is ±8 degrees Celsius. A standard deviation of 4.7 degrees Celsius exceeds 50% of the allowable deviation range (half-width) threshold of 4.0 degrees Celsius and is marked as high-fluctuation. When the mixing temperature in the internal mixing process fluctuates excessively for a long period, the batch-to-batch differences in the degree of crosslinking of the rubber compound will increase accordingly, directly affecting the uniformity of hardness of the rubber linings produced under the same vulcanization conditions in subsequent vulcanization processes.

[0018] A process correlation difference table is generated by verifying the correlation between the fluctuation distribution table and the process configuration parameters for upstream and downstream processes. The standard deviation of each key parameter in the fluctuation distribution table is compared with the corresponding allowable deviation range of the process configuration parameters. Process parameters whose standard deviation exceeds half the allowable deviation range are marked as exceeding fluctuation limits in the correlation verification. The transmission constraint coefficient of the process configuration parameters defines the proportional relationship of upstream parameter fluctuations transmitted downstream. The correlation difference is calculated using the formula D=Gd-Gu×k, where D is the correlation difference, Gu is the measured standard deviation of the upstream parameter, k is the transmission constraint coefficient of the corresponding pair in the process configuration parameters, and Gd is the measured standard deviation of the downstream parameter. The standard deviations of upstream and downstream parameters are uniformly converted to the downstream parameter dimensions before being included in the calculation. The process correlation difference table uses six pairs of upstream and downstream pairings of process configuration parameters as row indexes. A positive correlation difference indicates that the actual fluctuation of the downstream process exceeds the upstream transmission expectation, while a negative value indicates that the fluctuation of the downstream process is lower than the transmission expectation. Positive value pairs are given priority in subsequent vulnerable linkage marker identification. The parameters marked with high volatility in the volatility distribution table usually have larger correlation difference values ​​in the process correlation difference table. The actual volatility transmitted by high volatility parameters to downstream processes exceeds the preset transmission range of process configuration parameters. The 6 records in the process correlation difference table are sorted from largest to smallest correlation difference value. The Mooney viscosity of the rubber compound in the extrusion process has the highest correlation difference value with the roller spacing in the calendering process.

[0019] For example, the step of identifying process pairs that significantly influence downstream processes in the process association difference table and generating vulnerable linkage markers includes: extracting the dispersion of each pairing difference from the process association difference table according to the process pairing type to generate a pairing dispersion table; identifying process pairs with abnormally large dispersion based on the pairing dispersion table to generate a suspected abnormal pairing table; identifying newly added linkage abnormal pairs that have appeared after historical stabilization in the suspected abnormal pairing table to generate new abnormal markers; and classifying and labeling the suspected abnormal pairing table based on the new abnormal markers to generate vulnerable linkage markers.

[0020] The process correlation difference table is generated by extracting the dispersion of each pairing difference according to the process pairing type. The correlation difference quantity in the process correlation difference table is based on the statistical results of the current batch window. The dispersion of the pairing difference further characterizes the stability of the correlation difference quantity over multiple recent batch windows. The dispersion is measured by the coefficient of variation of the correlation difference quantity in each batch window. The correlation difference quantity sequence of extrusion to calendering pairing in the process correlation difference table within the last 10 batch windows is 0.72, 0.85, 0.91, 0.88, 0.87, 0.93, 0.79, 0.86, 0.90, 0.87, with a mean of 0.858, a standard deviation of 0.060, and a coefficient of variation of 0.070, reflecting that the correlation difference quantity of this pairing fluctuates little and remains consistently high. The pairing dispersion table covers all 6 pairs in the process correlation difference table, and the coefficient of variation and the mean of the correlation difference quantity are calculated for each pairing before being written. Pairs with lower mean values ​​of correlation differences in the process correlation difference table may have higher coefficients of variation, indicating that the difference in the pair is unstable. The pair dispersion table adds a double-high annotation to pairs with coefficients of variation higher than 0.3 and mean values ​​exceeding 0.4. The lower the coefficient of variation in the pair dispersion table, the more stable the correlation difference behavior of the process pair. Stable, high dispersion characteristics are better at reflecting the risk of continuous overtransmission between processes than random fluctuations. In the extrusion process, the Mooney viscosity of the rubber compound is consistently high, and the batch-to-batch range is stable. In the calendering process, the roller spacing is systematically larger between batches, and the stability of the liner thickness continues to decline in this segment. This pair is not included in the suspected abnormal pairing table because its coefficient of variation is below the threshold, but it is included in the process matching coefficient calculation based on the mean value of correlation differences during subsequent weighted integration.

[0021] Based on the pairing dispersion table, process pairings with abnormally high dispersion are identified, generating a suspected abnormal pairing table. Pairings in the pairing dispersion table whose coefficient of variation exceeds the mean plus one standard deviation of all pairing coefficients of variation are defined as process pairings with abnormally high dispersion. The mean coefficient of variation reflects the average fluctuation stability level of each process pairing in the current production line. The coefficients of variation for the six pairs in the pairing dispersion table are: mixing to extrusion 0.41, extrusion to calendering 0.070, calendering to vulcanization 0.15, mixing to calendering 0.11, mixing to vulcanization 0.09, and extrusion to vulcanization 0.22, with a mean of 0.18 and a standard deviation of 0.13. The threshold is the mean plus one standard deviation, which equals 0.31. The mixing to extrusion pairing has a coefficient of variation of 0.41, exceeding the threshold of 0.31, and is therefore identified as a process pairing with abnormally high dispersion. The other five pairs are all below the threshold. The suspected abnormal pairing table extracts pairing records that meet the abnormal dispersion criteria from the pairing dispersion table and adds an anomaly identification timestamp. Currently, only the mixing to extrusion pairing meets the condition of exceeding the coefficient of variation limit. In the pairing dispersion table, pairings that simultaneously meet both the criteria of exceeding the standard coefficient of variation and having a mean correlation difference exceeding 0.5 are marked as double anomalies in the suspected anomaly pairing table. The pairing of internal mixing and extrusion has a coefficient of variation exceeding 0.41 and a mean correlation difference exceeding 0.5 (0.63), thus meeting the double anomaly criteria and being marked accordingly. Double anomaly pairings indicate that the process transmission relationship is both persistently biased and unstable. The pairing of extrusion and calendering has a coefficient of variation of 0.070, far below the threshold of 0.31. Although the mean correlation difference of 0.858 is the highest among the six groups, it does not meet the dispersion anomaly identification criteria and is not included in the suspected anomaly pairing table.

[0022] Newly emerging linked anomaly pairs are identified in the suspected anomaly pairing table after historical stability, generating new anomaly markers. The coefficient of variation (COP) of the mixing-to-extrusion pairing in the suspected anomaly pairing table is traced back to the last 30 batch windows. Historical tracing is performed by retrieving the correlation difference sequence for each historical window from the production record archive using the pairing number. The COP sequence for the mixing-to-extrusion pairing in the suspected anomaly pairing table has an average COP of 0.12 in batch windows 1 to 20, indicating historical stability. From batch window 21 onwards, the COP rises to 0.28, and after batch window 25, it consistently exceeds the current threshold of 0.31, indicating that this pairing has recently experienced a linked anomaly. The criteria for identifying new anomaly markers are that the historical COP of a pair in the suspected anomaly pairing table is below the threshold in the previous 20 batch windows, and the COP continuously exceeds the threshold in the last 5 consecutive batch windows. The starting batch window for the continuous exceeding of the threshold is defined as the starting point of the new anomaly. The pairing of internal mixing and extrusion processes has historically shown stable correlation across batches. However, recently, after a change in carbon black raw material specifications in a particular batch, the coefficient of variation for this pairing has been steadily increasing. The higher specific surface area of ​​the new carbon black batch has led to increased viscosity fluctuations in the internal mixing process. These viscosity fluctuations, amplified through the process transmission chain from internal mixing to extrusion, have significantly worsened the batch-to-batch dispersion of extrusion pressure. Adding anomaly markers can accurately pinpoint such abrupt changes in process transmission relationships caused by raw material batch changes. New anomaly markers are associated with the starting batch window number, the average coefficient of variation before the start, the average coefficient of variation after the start, and the number of consecutive batch windows exceeding the limit. Pairings in the suspected anomaly pairing table that do not meet the addition criteria indicate that the correlation anomaly is a long-standing systemic problem rather than a recent change, and no new anomaly marker record is generated for these pairs.

[0023] Based on the newly added anomaly markers, the suspected anomaly pairing table is classified and labeled to generate a vulnerable linkage marker. Process pairings with records in the newly added anomaly markers are classified as newly emerging vulnerable linkages in the vulnerable linkage markers, while process pairings without corresponding newly added anomaly marker records in the suspected anomaly pairing table are classified as ongoing vulnerable linkages. If the mixing-to-extrusion pairing in the newly added anomaly markers meets the criteria for a newly emerging vulnerable linkage, the corresponding record in the suspected anomaly pairing table is written into the vulnerable linkage marker and a newly emerging classification label is attached. If there are pairings in the suspected anomaly pairing table with a consistently excessive coefficient of variation but no corresponding newly added anomaly marker record, these pairings are written into the vulnerable linkage markers with the ongoing vulnerable linkage label. The difference between newly emerging and ongoing classifications directly affects subsequent handling strategies. Newly emerging vulnerable linkages indicate the need to prioritize checking whether recent triggering factors such as mixing rotor wear or changes in raw material specifications have occurred. Wear of the mixing rotor blades reduces cutting and mixing efficiency, leading to uneven carbon black dispersion. Unevenly dispersed rubber enters the extrusion process, causing increased extrusion pressure fluctuations due to local viscosity differences. On the other hand, ongoing vulnerable linkages more often indicate a long-term mismatch between the transfer constraint coefficient of the process pairing and the actual production conditions of the rubber liner, requiring an assessment at the process configuration parameter level to determine whether the constraint coefficient standard needs to be revised. The vulnerable linkage markers are associated with the process pairing number, classification label, pairing dispersion table coefficient of variation, and the starting batch window number of the newly added anomaly marker. The starting batch window number of the newly added anomaly marker serves as a key time node for tracing changes in the production process in the vulnerable linkage markers, indicating that the transmission relationship between the mixing and extrusion processes underwent a structural change near the 25th batch window.

[0024] The process matching coefficient is generated by weighted integration of the fragile linkage markers and the process association difference table. The classification label of each record in the fragile linkage markers determines the weighting factor: the weighting factor for newly occurring fragile linkage pairings is set to 1.5, the weighting factor for persistent fragile linkage pairings is set to 1.2, and the weighting factor for process pairings not included in the fragile linkage markers in the process association difference table is 1.0. The formula for calculating the process matching coefficient is M = 1 - α × (D × W) / max(D × W), where D is the association difference of the corresponding pair in the process association difference table, W is the corresponding weighting factor, max(D × W) is the maximum value of the association difference after weighting all 6 pairs, and α is the normalized adjustment coefficient with a value of 0.82. A lower M value indicates a more severe linkage deviation in the pairing. When the mixing temperature in the internal mixing process fluctuates continuously, if this fluctuation is consistently and excessively transmitted to the extrusion process, leading to a decrease in the uniformity of the extruded rubber compound, the corresponding process matching coefficient will be significantly lower than that of the pairings in the normally linked states of calendering and vulcanization. This directly reflects a weak link in the process connection between internal mixing and extrusion. Pairings with low process matching coefficients will trigger more frequent allocation of testing resources in subsequent quality control schemes. After the six pairs in the process correlation difference table are weighted and normalized, the process matching coefficients are output in the form of a 6-element sequence. The element order is consistent with the pairing order in the process correlation difference table. The process matching coefficient sequence is passed as a whole to the subsequent weighted adjustment calculation.

[0025] Step S120: Based on the production parameter data, perform defect severity statistics to generate a defect heat map, extract the process hierarchy based on the process configuration parameters to generate a process control list, map the defect heat map to the process control list to locate suspicious processes with sudden defect changes, and establish a step-by-step quality control plan.

[0026] Specifically, a defect heatmap is generated based on defect severity statistics using production parameter data. The quality inspection records for each process in the production parameter data include three defect fields: appearance defect rating, dimensional deviation mark, and physical property non-compliance mark. Appearance defect ratings are graded from 1 to 5, with 5 being the most severe. Defect severity statistics calculate a weighted defect density for each defect type in each process. The weighted defect density is calculated using the formula Wd=sum(wi×ri), where Wd is the weighted defect density, wi is the weight of the i-th type of defect, and ri is the occurrence rate of that type of defect in the corresponding batch. Physical property non-compliance has the highest weight of 3.0, appearance defects have a weight of 1.5, and dimensional deviations have a weight of 2.0. The defect heatmap uses process number as the row axis and batch sequence as the column axis, covering nearly 30 batches. Each cell corresponds to the weighted defect density value for that process in that batch. The color mapping of the defect heatmap is divided into four levels based on the quartiles of the weighted defect density. The fourth level corresponds to high defect locations with a weighted defect density exceeding 0.8. The cells in batches 18 to 22 of the vulcanization process fall into the fourth level, exhibiting a continuous high defect clustering characteristic. The process locations in the defect heatmap where the weighted defect density exceeds 0.6 for three consecutive batches are marked as persistent high defect areas. Batches 15 to 17 of the mixing process constitute a persistent high defect area.

[0027] A process control list is generated by extracting process hierarchy based on process configuration parameters. The process configuration parameters define the standard operating procedures and parameter transfer constraints for four processes: mixing, extrusion, calendering, and vulcanization. Process hierarchy is determined by the cumulative impact of each process's transfer constraint coefficient on the final product quality. Processes with higher transfer constraint coefficients have a greater downstream impact on quality, and therefore higher priority. The vulcanization process has no downstream transfer constraints but directly determines the final product properties, and is designated as a Level 1 control process. The extrusion process has transfer constraint coefficients of 0.42 and 0.27 for calendering and vulcanization, respectively, with a cumulative transfer impact coefficient of 0.69, and is designated as a Level 2 control process. The mixing process has transfer constraint coefficients of 0.35, 0.23, and 0.15 for extrusion, calendering, and vulcanization, respectively, with a cumulative transfer impact coefficient of 0.73, and is designated as a Level 2 control process. The calendering process's transfer constraint coefficient only affects vulcanization as a single downstream process, with a cumulative transfer impact coefficient of 0.27, and is designated as a Level 3 control process. The process control list is organized step by step. Each process is associated with a process name, control level, a list of nominal values ​​of process configuration parameters, allowable deviation range, and associated downstream process number group. The control level in the process control list determines the distribution of testing frequency in subsequent quality control plans. The first-level control process has the highest testing frequency and the strictest parameter tolerance. The process control list adds a continuous monitoring label for the first-level control process.

[0028] In some embodiments, mapping the defect heatmap to the process control list to locate processes with suspected sudden defect changes and establish a step-by-step quality control scheme includes: extracting the defect density of each process according to the process time period from the defect heatmap to generate a process defect density table; identifying processes with sudden increases in defect density before and after shift change nodes from the process defect density table to generate shift change sudden increase markers; identifying processes with random sudden increases in defect density caused by equipment failure from the process defect density table to generate equipment sudden increase markers; and classifying and mapping the shift change sudden increase markers and the equipment sudden increase markers to the process control list to establish a step-by-step quality control scheme.

[0029] The defect density of each process was extracted from the defect heatmap according to the process time period to generate a process defect density table. In the batch sequence of the column axis of the defect heatmap, each batch is labeled with a shift identifier: the batch number for the morning shift is suffixed with A, and the batch number for the evening shift is suffixed with B. The process time period is divided into two categories, morning and evening shifts, based on the shift boundaries. In the defect heatmap, the weighted average defect density of the mixing process during the morning shift is 0.31, and the average for the evening shift is 0.47, with a difference of 0.16. This difference indicates that the quality control level of the mixing process during the evening shift is lower than that during the morning shift. The decreased accuracy of material weighing and increased deviation in the determination of the mixing endpoint due to continuous high-temperature operation by evening mixing operators are common causes of shift quality differences. For the extrusion process, the average defect density during the morning shift is 0.28, and the average for the evening shift is 0.29, with a difference of only 0.01. The process defect density table is statistically analyzed by process and shift. It records the mean, standard deviation, and number of batches with sudden increases in weighted defect density for each batch within that shift. The rule for counting sudden increases is that a batch with a density difference exceeding 0.35 between adjacent batches is counted as one sudden increase. The number of sudden increases in the mixing process is 3 in the process defect density table. Process periods with higher standard deviations in the process defect density table reflect poorer defect stability during the corresponding shift. In the defect heatmap, the standard deviation for the mixing process during the evening shift is 0.18, significantly higher than the 0.09 for the morning shift. The number of sudden increases in batches corresponding to persistently high-defect areas in the defect heatmap is highlighted in the process defect density table. The process defect density table also adds a high-risk period label to process periods with persistently high defect areas.

[0030] The process defect density table identifies processes with a sudden increase in defect density before and after shift change nodes, generating shift change surge markers. A shift change node is defined as the boundary between adjacent batches in the batch sequence where the shift identifier changes from A to B or from B to A. There are 14 shift change nodes across approximately 30 batches. The weighted defect density difference between one batch before and after each shift change node in the process defect density table is used as a candidate value for a shift change surge. In the process defect density table, the surge candidate value for the mixing process at the 8th shift change node is 0.53, exceeding the shift change surge identification threshold of 0.35. This shift change node is therefore marked as the shift change surge position for the mixing process. The shift change surge threshold of 0.35 is set based on the 75th percentile of all shift change node surge candidate values ​​for all processes in the process defect density table; surges exceeding this percentile are considered statistically significant surges. In the process defect density table, if the same process experiences a sudden increase at multiple shift change nodes, the shift change surge marker records all shift change node numbers that meet the threshold and the corresponding shift change direction. The internal mixing process has two shift change nodes that meet the criteria. The frequent occurrence of shift change surges indicates a problem with the operational procedures being not properly followed during shift handover in the internal mixing process. The rotor temperature of the internal mixer drops due to heat dissipation during shift changes. If the incoming operator does not perform sufficient preheating according to procedures before feeding materials and starting mixing, the initial mixing temperature will be significantly lower, resulting in substandard rubber compound mixing uniformity. The shift change surge marker is associated with the process number, shift change node number, surge candidate value, and shift change direction. In the process defect density table, the process with the most hits of the shift change surge marker is the internal mixing process (2 hits), while the extrusion process (0 hits).

[0031] For processes with random spikes in defect density caused by equipment malfunctions, equipment spike markers are generated. The generation of these markers involves two stages: candidate identification and cross-validation. In the candidate identification stage, batches with spikes whose spike locations do not align with shift change nodes and whose spike duration does not exceed two are selected from the process defect density table. This is achieved by excluding nodes already marked with shift change spike markers. In the process defect density table, the extrusion process shows a spike in weighted defect density from 0.26 to 0.71 in batch 19. This batch does not align with any shift change node, and the density drops back to 0.31 in batch 20, meeting the characteristics of a random spike. Therefore, it is identified as a candidate location for an equipment spike. In the cross-validation stage, the time window corresponding to the candidate location is compared with the equipment alarm records in the production parameter data. Candidate locations with alarm records within the window are marked with a confirmed status in the equipment spike marker, while candidate locations without alarm records are marked with a "reason pending investigation" status. An alarm record for excessive screw speed was found within the time window corresponding to batch 19 of the extrusion process. During the period of excessive speed, the uncontrolled shear rate of the rubber compound caused sharkskin-like defects on the surface of the extrudate and out-of-tolerance die dimensions. This location was recorded as a confirmed status in the equipment surge marker. The equipment surge marker is associated with the process number, surge batch number, surge magnitude, and confirmation status label. No surges exceeding the threshold occurred in the vulcanization and calendering processes outside of shift change points, and the equipment surge marker only contains one confirmation record for the extrusion process.

[0032] A tiered quality control scheme is established based on the mapping of shift change escalation markers and equipment escalation markers to the process control list. The two records for the mixing process in the shift change escalation marker correspond to the second-level control level of the mixing process in the process control list. The quality control measures triggered by a shift change escalation are to perform 100% full inspection on three consecutive batches after the escalation shift change node and to conduct a special verification of the shift handover specifications. The one record for the extrusion process in the equipment escalation marker corresponds to the second-level control level of the extrusion process in the process control list. The quality control measures triggered by an equipment escalation are to perform 100% re-inspection on the escalation batch, trace the corresponding equipment maintenance records, and assess whether the batch of products needs to be downgraded. The control level of the process control list determines the upper limit of testing resource investment for each escalation trigger. When a first-level control process is triggered, it is upgraded to double sampling inspection of the entire batch. The vulcanization process, as a first-level control process, must perform double sampling inspection of the entire batch because there is no subsequent opportunity to correct the vulcanization quality of the finished lining. Second-level control processes are upgraded to single sampling inspection of the entire batch, and third-level control processes are upgraded to sampling inspection once every two batches. In the hierarchical quality control scheme, shift change surge markers and equipment surge markers are distinguished by different trigger type labels. Quality control items triggered by shift change surges include additional operation specification verification tasks, while quality control items triggered by equipment surges include additional equipment health assessment tasks. Process periods marked with high-risk periods in the process defect density table trigger continuous high-defect quality control items in the hierarchical quality control scheme. The continuous high-defect areas of batches 18 to 22 in the vulcanization process correspond to the first-level control process. Double sampling inspection of the entire batch is carried out according to the first-level control requirements, and physical property re-inspection is initiated. The hierarchical quality control scheme generates a total of 4 quality control items. Processes not marked by surge markers and high-risk period markers in the process control list are performed according to the regular inspection frequency.

[0033] Step S130: Apply weighted adjustment to the step-by-step quality control scheme based on the process matching coefficient to form a comprehensive control scheme. For the comprehensive control scheme, generate detection anomaly alarms for process sections where deviations are amplified and completely absorbed. Determine the graded quality control benchmark by extracting the failure distribution law according to the detection anomaly alarms.

[0034] Specifically, a comprehensive control plan is formed by applying weighted adjustments to the step-by-step quality control scheme based on the process matching coefficient. In the six-element sequence of process matching coefficients, the lowest matching coefficient for the mixing-to-extrusion pairing is 0.18. The weighted adjustment multiplies the upstream process inspection frequency involved in process pairings with a matching coefficient below 0.4 by an amplification factor, defined as 1 divided by the corresponding process matching coefficient. The amplification factor for the mixing process is 1 / 0.18, approximately 5.6. The original step-by-step quality control scheme required sampling once every 5 batches; multiplying by 5.6 and rounding to the nearest whole number results in sampling once per batch. The inspection frequency for process pairings with a matching coefficient above 0.7 remains unchanged from the original step-by-step quality control scheme. The inspection frequency for process pairings between 0.4 and 0.7 is multiplied by 1.5 and rounded to the nearest whole number. The sample size for process pairings with a matching coefficient below 0.3 is doubled based on the quantity specified in the step-by-step quality control scheme. The integrated control scheme inherits the entire record structure of the hierarchical quality control scheme. It updates the values ​​for detection frequency and sample size based on weighted adjustments and adds a process matching coefficient adjustment coefficient to record the actual amplification factor of each process. The shift change surge trigger records and equipment surge trigger records in the hierarchical quality control scheme retain trigger type labels in the integrated control scheme. The weighted adjustment of the process matching coefficient is further superimposed on the detection frequency of the trigger records. For the mixing process, due to its extremely low process matching coefficient and the fact that it also carries the shift change surge trigger label, when multiple rules are superimposed, the rule with the highest detection intensity is taken as the final execution standard. The integrated control scheme adjusts its detection frequency to double sampling for each batch. Double sampling requires that the same batch be independently inspected by two different inspectors, who then cross-compare the results to eliminate subjective bias from single-person inspection. The integrated control scheme uses a combination key of process number and trigger type as a unique index.

[0035] In some embodiments, the step of generating a detection anomaly alarm for process sections where deviation amplification and complete deviation absorption are identified by the integrated control scheme includes: extracting the deviation transmission ratio between adjacent processes from the integrated control scheme to generate a transmission ratio distribution table; identifying process sections where the transmission ratio is continuously greater than a reference value based on the transmission ratio distribution table to generate a transmission amplification marker; identifying process sections where the transmission ratio is continuously zero based on the transmission ratio distribution table to generate a complete absorption marker; and generating a detection anomaly alarm based on the combined transmission amplification marker and the complete absorption marker.

[0036] The comprehensive control scheme extracts the deviation transmission ratio between adjacent processes to generate a transmission ratio distribution table. The comprehensive control scheme covers the parameter deviation records of nearly 30 batches. The deviation transmission ratio of paired adjacent processes is calculated using the formula T=δd / δu, where δu and δd are the upstream and downstream parameter deviations of paired adjacent processes in the same batch, respectively. Both δu and δd are divided by the allowable deviation half-width of their respective processes to normalize to a dimensionless deviation rate before being included in the calculation. This is different from Gu and Gd, which represent standard deviation in S110. A T greater than 1 indicates that the deviation is amplified during transmission, while a T equal to 0 indicates that the deviation is completely absorbed. In the integrated control scheme, the mean of the conduction ratio sequence for nearly 30 batches of mixing-to-extrusion pairings in the integrated control scheme was 1.34. The maximum value appeared in batch 19 at 2.17, and the minimum value appeared in batch 6 at 0.71, with a sequence standard deviation of 0.38. This reflects that the deviation conduction behavior of this pairing fluctuates significantly between batches. The abnormally high conduction ratio in batch 19 coincided with the screw speed exceeding the limit alarm in the extrusion process of that batch. The excessive screw speed caused a shortened residence time of the rubber compound in the barrel, resulting in the temperature deviation in the upstream mixing process being transmitted to the extrusion die end without sufficient homogenization. The conduction ratio distribution table is organized with a two-level index of process pairing number and batch number, with a scale of 6 pairs multiplied by 30 batches, totaling 180 numerical units. In the integrated control scheme, process pairings with a process matching coefficient lower than 0.4 are marked with a low matching coefficient in the conduction ratio distribution table. When the upstream process deviation in the conduction ratio distribution table is zero, the conduction ratio cannot be calculated, and the corresponding position is marked as a null value. Null value positions are not involved in the continuous judgment in subsequent identification.

[0037] Based on the conduction ratio distribution table, conduction amplification markers are generated for process segments where the conduction ratio is consistently greater than a benchmark value. The identification condition for the conduction amplification marker is that the conduction ratio of paired products in the same process is greater than the benchmark value of 1.2 for five consecutive batch windows. The benchmark value of 1.2 is set based on the 75th percentile of the conduction ratios of all paired products in the conduction ratio distribution table. In the conduction ratio distribution table, the conduction ratio sequence for batches 15 to 22 of the mixing-to-extrusion pairing is 1.31, 1.45, 1.52, 2.17, 1.89, 1.63, 1.48, and 1.35, respectively. Eight consecutive batches in this segment exceed the benchmark value of 1.2, satisfying the conduction amplification identification condition. The consistently high mixing temperature in this segment of the mixing process leads to a widening of the batch-to-batch Mooney viscosity difference in the rubber compound. This high viscosity difference is further amplified by the pressure response characteristics of the extrusion process after being transmitted through the mixing-to-extrusion transfer chain. The batch-to-batch fluctuation of extrusion pressure exhibits a significant amplification effect in this segment. In the conduction ratio distribution table, the conduction ratio sequence of extrusion-to-calendering pairings exceeded the benchmark value of 1.2 for five consecutive batches from batch 17 to 21, with an average of 1.38 and a peak value of 1.61, thus triggering the write of the conduction amplification marker. The conduction amplification marker is associated with the process pairing number, the start and end batch numbers of the continuous exceedance, and the average and peak conduction ratios within the segment. If process pairings marked with low matching coefficients in the conduction ratio distribution table also trigger conduction amplification markers, a joint low matching coefficient marker is added to the conduction amplification marker. For mixing-to-extrusion pairings, both low matching coefficient markers and conduction amplification markers are present, and the joint marker is activated.

[0038] A complete absorption marker is generated for process segments in the conduction ratio distribution table where the conduction ratio is consistently zero. Complete absorption refers to the anomaly where upstream processes exhibit parameter deviations, but the corresponding parameter deviations in downstream processes remain consistently zero. This phenomenon is extremely rare under normal production conditions and usually indicates a false zero deviation caused by sensor malfunction or signal interruption in the downstream process. A zero conduction ratio needs to be distinguished from a null value caused by zero upstream deviation and the aforementioned complete absorption scenario; the complete absorption marker only identifies the latter. In the conduction ratio distribution table, the conduction ratio sequence for batches 23 to 27 of the calendering-vulcanization pairing is 0.00. The corresponding roller temperature deviations for these batches in the calendering process are 1.8, 2.3, 1.5, 2.1, and 1.9 degrees Celsius, respectively. Upstream processes all exhibit deviations, while the temperature deviations for the corresponding batches in the vulcanization process are all zero, confirming a complete absorption scenario. The identification criteria for the complete absorption marker are three consecutive batches with a zero conduction ratio in the same process pairing and a non-zero upstream deviation. The number of consecutive batches must be less than five for the conduction amplification marker, indicating a higher risk level for the complete absorption anomaly and requiring faster warning triggering. The data reliability rating is determined by the following factors: complete absorption is associated with the process pairing number, the start and end batch numbers that are continuously zero, the average upstream deviation within the section, and the data reliability rating. The data reliability rating is based on the magnitude of the average upstream deviation; complete absorption cases with an average exceeding 30% of the allowable deviation half-width are rated as low reliability. The average upstream deviation of calendering to vulcanization pairing is 1.92 degrees Celsius, and the allowable deviation range of the calendering process roller temperature is ±8 degrees Celsius. This does not exceed the 30% threshold of 2.4 degrees Celsius (8 degrees Celsius half-width), and is rated as moderate reliability.

[0039] Anomaly alarms are generated based on the combined use of conduction amplification and complete absorption markers. These markers are stored together in the alarm, with anomaly type tags categorized as either conduction amplification or complete absorption. These two types of anomalies have different handling priorities; complete absorption anomalies, due to questionable data reliability, have higher priority than conduction amplification anomalies. The low matching coefficient of the mixing-to-extrusion pairing in the conduction amplification marker is converted into a high-risk marker for that record in the alarm. A high-risk marker indicates that the anomaly possesses both conduction amplification and low matching coefficient characteristics, requiring priority handling. The alarm is associated with the process pairing number, anomaly type tag, anomaly batch segment, anomaly intensity value, high-risk marker status, and suggested handling measures. The anomaly intensity for conduction amplification anomalies is the average conduction ratio within the segment, while the anomaly intensity for complete absorption anomalies is the average upstream deviation within the segment. Conduction amplification and complete absorption markers are arranged in descending order of handling priority in the alarm. Calendering-to-vulcanization pairings hit by the complete absorption marker are ranked first; mixing-to-extrusion high-risk pairings hit by the conduction amplification marker are ranked second; and extrusion-to-calendering conduction amplification pairings are ranked third. Recommended handling measures are determined jointly by the abnormality type label and the high-risk label status. The recommended measures for high-risk records of conduction amplification are to immediately stop the line and investigate. The recommended measures for routine records of conduction amplification are to increase the frequency of random inspections. The recommended measures for records of complete absorption are to calibrate the testing instrument and re-inspect the corresponding batch.

[0040] The graded quality control benchmark was determined based on the failure distribution patterns extracted from the abnormal alarm detection. The abnormal alarm detection covered three process pairings: mixing to extrusion, extrusion to calendering, and calendering to vulcanization. The abnormal batch segment for mixing to extrusion pairing was batches 15 to 22, and for extrusion to calendering pairing, it was batches 17 to 21. These two segments highly overlapped, and the failure distribution patterns identified batches 17 to 21 as a high-incidence segment of simultaneous abnormalities across multiple pairings. The abnormal intensity of mixing to extrusion pairing (1.60) was higher than that of extrusion to calendering pairing (1.38). The former was designated as the primary failure source, and the latter as the secondary failure source. The graded quality control benchmark was divided into three quality control levels based on the primary and secondary failure sources. The primary failure source corresponds to the first-level quality control benchmark, requiring double sampling for each batch in the relevant process and extending the scope of enhanced testing to include three batches before and after the boundary of the abnormal batch segment. The secondary failure source corresponds to the second-level quality control benchmark, requiring single sampling for each batch in the relevant process and intensive sampling within the abnormal batch segment. Due to data reliability issues, a separate data verification benchmark is established for the calendering-vulcanization pairing corresponding to the complete absorption mark, requiring manual review of the test records for all recent batches of this pairing. The graded quality control benchmark is associated with the quality control level label, the corresponding process pairing number list, the test frequency requirements, and the sample size requirements. The strictness of the quality control measures is consistent with the priority of handling abnormal alarms.

[0041] Step S140: The production parameter data is corrected step by step according to the graded quality control benchmark to generate a monitoring feedback table. Deviation monitoring is performed on the monitoring feedback table to generate a quality control trend chart. The quality control trend chart is used to detect parameter anomalies and quickly reach the standard to trigger a false qualified early warning and form a qualified judgment level.

[0042] Specifically, a monitoring feedback table is generated by progressively correcting production parameter data using a tiered quality control benchmark. The three levels of the tiered quality control benchmark correspond to the first-level quality control benchmark for mixing and extrusion pairing, the second-level quality control benchmark for extrusion and calendering pairing, and the data verification benchmark for calendering and vulcanization pairing. The measured deviations of each process in the production parameter data are processed progressively according to the correction rules of the corresponding level of the tiered quality control benchmark. The first-level quality control benchmark requires correction labeling for batches where the mixing temperature deviation in the mixing process exceeds 60% of the allowable deviation half-width. In the production parameter data, the mixing temperature deviation of batch 19 in the mixing process is 6.8 degrees Celsius, exceeding the 60% threshold of 4.8 degrees Celsius (8 degrees Celsius half-width). This batch is written with a correction label in the monitoring feedback table. The measured deviations of each process in the production parameter data are compared batch by batch with the allowable deviation range of the corresponding process in the tiered quality control benchmark. Batches with deviations within the allowable range are labeled as compliant; batches exceeding the allowable range but not exceeding the correction trigger threshold are labeled as critical; and batches exceeding the correction trigger threshold are labeled as requiring correction. The monitoring feedback form covers all intersections of batches and processes, recording the deviation value and compliance status for each batch and process. The data verification benchmark for the graded quality control standard marks all recent batches paired with calendering and vulcanization in the monitoring feedback form as pending verification. In the production parameter data, the deviation record for the vulcanization process of this pairing is replaced with a pending verification label instead of a compliance status mark. All 30 batches covered by the production parameter data are written into the monitoring feedback form, with a total of 7 batch process combinations requiring correction.

[0043] In some embodiments, the step of performing deviation monitoring and generating a quality control trend chart on the monitoring feedback table includes: extracting historical deviation records of each key parameter from the monitoring feedback table according to the process to generate a deviation history table; fitting the deviation history table with the deviation trends of mixing temperature, extrusion pressure, and vulcanization time to generate a trend fitting table; identifying sections with significant differences in narrowing rate between different shifts of the same process from the trend fitting table to generate shift difference markers; and weightedly integrating the shift difference markers with the trend fitting table to generate a quality control trend chart.

[0044] A deviation history table is generated by extracting historical deviation records of key parameters from the monitoring feedback table according to each process. Each of the 30 batch records in the monitoring feedback table corresponds to a column of deviation data for each process number. Extraction is done by process number, with each column being extracted sequentially. For the mixing process, the deviation sequence of mixing temperature is extracted; for the extrusion process, the deviation sequence of extrusion pressure is extracted; and for the vulcanization process, the deviation sequence of vulcanization time is extracted. Among the 30 values ​​of the mixing temperature deviation sequence in the mixing process of the monitoring feedback table, the average deviation of compliant batches is 2.3 degrees Celsius, while the average deviation of batches requiring correction is 6.1 degrees Celsius. The difference in deviation distribution between the two types of batches is distinguished in the deviation history table by compliance status annotation. The deviation history table is organized process-by-process and batch-by-batch. Each deviation record is associated with the deviation value, compliance status annotation, batch / shift identifier, and graded quality control benchmark level label. The key parameter processes selected are the three processes involved in the graded quality control benchmark: mixing, extrusion, and vulcanization. The calendering process is not included in the trend analysis because it is currently under data verification and awaiting confirmation. The deviation history table covers 90 deviation records. Batch records in the monitoring feedback table that are pending verification are retained with independent labels in the deviation history table. In the deviation history table, the deviation values ​​of the records awaiting verification are marked in parentheses to indicate that the reliability of those values ​​is questionable. Seven batch process combinations marked as requiring correction in the monitoring feedback table are tagged with "Needs Correction" in the deviation history table; these seven records are treated as outliers in subsequent trend fitting. The batch shift identifier in the deviation history table is inherited from the batch number suffix in the monitoring feedback table. Grouping the deviation history table by shift identifier allows for comparison of the distribution differences in deviation values ​​between morning and evening shifts.

[0045] A trend fitting table was generated by fitting the deviation history table to the trends of mixing temperature, extrusion pressure, and vulcanization time deviations. Trend fitting was performed on the mixing temperature deviation sequence, extrusion pressure deviation sequence, and vulcanization time deviation sequence in the deviation history table, using the moving least squares method with a fitting window length of 8 batches and a step size of 1 batch. The fitting results of the mixing temperature deviation sequence in the deviation history table showed a slow decreasing trend from batches 1 to 14, with a mean fitting slope of -0.08 degrees Celsius / batch. From batches 15 to 22, the fitting slope turned positive with a mean of +0.31 degrees Celsius / batch. The trend reversal point was located between batches 14 and 15. The time of the trend reversal coincided with the initial period of the new fragile linkage between mixing and extrusion in the fragile linkage marker. After the change in carbon black raw material specifications, the mixing temperature of the mixing process lost its original batch-to-batch convergence trend and instead diverged. Each record in the trend fitting table corresponds to the fitting result of one process within a fitting window. It correlates the fitting slope, fitting intercept, mean deviation within the fitting window, goodness of fit R², and deviation narrowing rate. The deviation narrowing rate is expressed as the absolute value of the fitting slope; a negative slope corresponds to a positive narrowing rate, and a positive slope corresponds to a negative narrowing rate. The seven records in the deviation history table that require correction are weighted less in the trend fitting calculation, with a weighting coefficient set to 0.3 to avoid excessive bias in the fitting results from outliers. The fitting slope of the vulcanization time deviation sequence in the trend fitting table is close to zero across all batch windows, and the mean goodness of fit R² is 0.12. The fluctuations in the corresponding sequences in the deviation history table are randomly distributed. The trend fitting table generates a total of 69 records across 3 processes and 23 effective fitting windows.

[0046] For example, the step of identifying sections with significant differences in narrowing rates between different shifts of the same process using the trend fitting table and generating shift difference markers includes: generating a shift rate distribution table by statistically analyzing the deviation narrowing rate of each shift in each process based on the trend fitting table; calculating the cross-shift rate difference in the shift rate distribution table to identify processes whose differences exceed a preset tolerance and generating a rate difference process table; filtering the rate difference process table by continuous shift differences to exclude occasional differences and generating continuous difference markers; and generating shift difference markers based on the distribution density of the continuous difference markers.

[0047] Based on the trend fitting table, the deviation narrowing rate of each process and shift was statistically analyzed to generate a shift rate distribution table. The 69 records in the trend fitting table were grouped by process number and batch / shift identifier. The mixing temperature process had 23 fitting window records. For each fitting window, the deviation narrowing rate was calculated separately for the morning shift batch subset and the evening shift batch subset within the window. The deviation narrowing rate was represented by the absolute value of the linear fitting slope of the deviation quantity sequence within the subset. In the trend fitting table, within the 18th fitting window of the mixing temperature process, the fitting slope of the deviation quantity sequence of the 5 batches in the morning shift subset was -0.12, and the fitting slope of the deviation quantity sequence of the 3 batches in the evening shift subset was +0.28. The deviation narrowing rates of the two subsets were 0.12 and -0.28, respectively. The negative deviation narrowing rate of the evening shift subset indicates that the deviation in the evening mixing process not only did not narrow but continued to diverge within this fitting window. The evening shift operators relied on visual observation of the torque curve inflection point to determine the mixing endpoint, but the decreased attention during nighttime operations reduced the accuracy of endpoint determination. The shift rate distribution table is organized using a three-level index: process number, starting batch number of the fitting window, and shift identifier. Fitting windows with a goodness-of-fit R² below 0.3 in the trend fitting table are marked with low confidence in the corresponding positions in the shift rate distribution table. The deviation narrowing rate values ​​of the low confidence marks are weighted down in subsequent difference calculations, with a weighting coefficient of 0.5.

[0048] The shift rate distribution table is used to calculate the inter-shift rate difference to identify processes where the difference exceeds a preset tolerance, generating a rate difference process table. In the shift rate distribution table, the difference between the morning and evening narrowing rates within the same fitting window for the same process is defined as the inter-shift rate difference. A positive difference indicates the morning narrowing rate is higher than the evening rate, and a negative difference indicates the evening narrowing rate is higher than the morning rate. In the shift rate distribution table, the inter-shift rate difference for the 18th fitting window of the mixing temperature process is 0.40. The preset tolerance is set to the 75th percentile of the absolute value of the inter-shift rate difference for all fitting windows across all processes in the shift rate distribution table; the currently calculated preset tolerance is 0.22. Process fitting window combinations in the shift rate distribution table where the absolute value of the inter-shift rate difference exceeds the preset tolerance of 0.22 are identified as rate difference exceedance locations. The mixing temperature process has 6 fitting windows exceeding the tolerance, the extrusion pressure process has 1 fitting window exceeding the tolerance, and the vulcanization time process has no windows exceeding the tolerance after being weighted down due to a large number of low-confidence annotations. The rate difference process table summarizes out-of-specification cases for each process, linking them to a list of out-of-specification fitted window numbers, the rate difference across shifts for each window, and the proportion of out-of-specification windows to all valid fitted windows for that process. Fitted windows marked with low confidence in the shift rate distribution table have their cross-shift rate difference calculated using a weighting factor of 0.5. Only those exceeding the preset tolerance after weighting are included in the out-of-specification window list of the rate difference process table. Processes in the rate difference process table with an out-of-specification window proportion exceeding 30% are marked as systemic shift difference processes. The mixing temperature process, with 6 out-of-specification windows (26% of the 23 valid fitted windows, not exceeding 30%), is marked as a localized shift difference process.

[0049] The rate difference process table is filtered by continuous shift differences to exclude occasional differences and generate continuous difference markers. The six exceeding fitting windows for the mixing temperature process in the rate difference process table are numbered 18, 19, 20, 21, 24, and 25. Windows 18 to 21 are consecutive, and windows 24 to 25 are consecutive. The condition for consecutive arrangement is that the difference between the starting batch numbers of adjacent fitting windows is equal to 1. The screening condition for continuous differences is that three or more consecutive fitting windows for the same process in the rate difference process table exceed the standard. Four consecutive exceeding windows from batches 18 to 21 meet the condition, while two consecutive exceeding windows from batches 24 to 25 do not meet the minimum continuity requirement and are excluded as occasional differences. The continuous difference marker is associated with the process number, the starting and ending fitting window numbers for consecutive exceeding, the average rate difference across shifts within the segment, and the peak rate difference. In the rate difference process table, only one of the extrusion pressure process exceeded the standard and the fitting window did not meet the condition of three consecutive values, so it was not marked as a continuous difference. The exceeding of the standard in the extrusion pressure process was identified as an occasional difference and excluded. The continuous difference mark finally only included one record in the mixing temperature process, which corresponds to the continuous exceeding of the standard in the fitting window from batch 18 to batch 21. The average rate difference across shifts in the segment was 0.38, and the peak value was 0.47.

[0050] Shift difference markers are generated based on the distribution density of persistent difference markers. The continuous exceeding range of the mixing temperature process within the persistent difference markers covers the fitted window of batches 18 to 21, with a fitted window length of 8 batches. Window 18 covers batches 18 to 25, and window 21 covers batches 21 to 28. After merging, the corresponding production batch range is batches 18 to 28, and this range spans 11 batches in the batch time series. The distribution density of persistent difference markers is defined as the proportion of continuously exceeding batches to the total number of batches. The number of continuously exceeding batches in the mixing temperature process is 11, accounting for 0.37 of the total 30 batches, exceeding the high-density threshold of 0.30. Shift difference markers are assigned difference intensity levels based on the distribution density of the persistent difference markers. Process segments with a distribution density exceeding 0.30 are marked as high-density shift difference, those between 0.15 and 0.30 are marked as medium-density shift difference, and those below 0.15 are marked as low-density shift difference. The distribution density of 0.37 in the internal mixing temperature process record in the continuous difference marker corresponds to a high-density shift difference level. One high-density record is written in the shift difference marker, covering batches from 18 to 28. The shift difference marker is associated with the process number, difference intensity level label, covered batch range, distribution density value, and continuous difference marker source number. The occurrence of high-density shift differences indicates that the operation specifications of the internal mixing process in this section have deviated from the standard process requirements to the point that special rectification is required.

[0051] A quality control trend chart is generated by weighted integration of shift difference markers and trend fitting tables. The high-density difference segment in the mixing temperature process within the shift difference markers covers batches 18 to 28. The fitting slope sequence corresponding to this segment in the trend fitting table shows a consistently positive value in the late-shift batch subset. Weighted integration uses the difference intensity level of the shift difference markers as the weighting coefficient, with a weighting coefficient of 1.5 for the high-density level. The fitting slope values ​​of each process and fitting window in the trend fitting table are multiplied by the corresponding shift difference marker weighting coefficient when generating the quality control trend chart. The fitting window for processes not covered by the shift difference markers has a weighting coefficient of 1.0. The slope of the broken line in the quality control trend chart is amplified in the segments covered by the shift difference markers to highlight the degree of trend anomaly in those segments. The quality control trend chart uses batch time sequence as the horizontal axis and the weighted deviation narrowing rate sequence of each process as the vertical axis. Each process corresponds to a broken line, and each data point on the broken line corresponds to the weighted deviation narrowing rate of a fitting window in the trend fitting table. The affected sections of the shift difference markers are marked with a shaded background on the quality control trend chart. The fitting windows marked with low confidence in the trend fitting table are represented by dashed data points at their corresponding positions on the quality control trend chart. In the trend fitting table, the deviation narrowing rate of the mixing temperature process for batches 15 to 22 is negative. After weighting with shift difference markers, the absolute value of the slope of the line segment in the quality control trend chart for this section is larger, making the trend anomaly more visually apparent. The quality control trend chart contains three lines: mixing temperature, extrusion pressure, and vulcanization time. Each line contains 23 weighted data points.

[0052] In some embodiments, the step of using the quality control trend chart to detect abnormally rapid achievement of parameters to trigger a false acceptance warning and form a acceptance level includes: extracting the achievement rate of each process parameter from the quality control trend chart to generate an achievement rate distribution; identifying process segments with abnormally high achievement rates based on the achievement rate distribution to generate rapid achievement markers; tracing and identifying repeated rapid achievement processes under specific tooling based on the rapid achievement markers according to tooling number to generate tooling wear false acceptance markers; and classifying and quantifying the achievement rate distribution based on the tooling wear false acceptance markers to generate an acceptance level.

[0053] The pass rate distribution is generated by extracting the pass rate of each process parameter from the quality control trend chart. The three broken lines of the quality control trend chart store the weighted deviation narrowing rate value at each data point. The pass rate is defined as the reciprocal of the number of batches required for the measured deviation of the corresponding batch to decrease from the previous batch level to within the acceptable range; the fewer the batches, the higher the pass rate. In the quality control trend chart, the measured deviation of batch 24 in the extrusion pressure process is 0.2 MPa, which is within the acceptable range. The previous batch, batch 23, had a deviation of 1.8 MPa and was in a state requiring correction. From needing correction to being acceptable, only one batch was needed, resulting in a pass rate of 1.0, which is far higher than the historical average pass rate of 0.18 for this process. This abnormal behavior of the extrusion pressure suddenly dropping from needing correction to being acceptable within a single batch suggests that the acceptable status of this batch may not be due to a genuine improvement in process parameters, but rather more likely caused by abnormal values ​​in the testing process or measurement offsets caused by wear and tear on the tooling cavity dimensions. The compliance rate distribution only enters values ​​for batches in processes where compliance events actually occurred; batches that did not achieve compliance have blank values. The average normal compliance rate is stored as the benchmark value for rapid compliance identification in the process-level summary record of the compliance rate distribution. In the quality control trend chart, the mixing temperature process achieved compliance in batch 29, decreasing from a deviation of 5.2 degrees Celsius to 1.8 degrees Celsius, over two batches, with a compliance rate of 0.5, exceeding the historical average normal compliance rate of 0.11 by more than three times. For batches in the quality control trend chart with data verification pending confirmation status, the compliance rate corresponding to these batches is marked as pending verification in the compliance rate distribution. The compliance rate marked as pending verification is not included in the benchmark calculation of the average normal compliance rate.

[0054] Based on the achievement rate distribution, process segments with abnormally high achievement rates are identified and rapid achievement markers are generated. Batch positions in the achievement rate distribution where the single batch achievement rate exceeds three times the historical average of the corresponding process's normal achievement rate are defined as rapid achievement candidate points. For example, in the achievement rate distribution, the 24th batch of the extrusion pressure process, with an achievement rate of 1.0, exceeds the three-times threshold of 0.54 (0.18 average), and is confirmed as a rapid achievement candidate point. Further identification of rapid achievement markers requires that there are no known equipment alarm records within one batch before and after the rapid achievement candidate point, excluding normal rapid recovery after equipment failure repair. Rapid achievement candidate points with no equipment alarm records within the corresponding batch time window of the achievement rate distribution are confirmed and marked with a rapid achievement marker. Similarly, in the achievement rate distribution, the 29th batch of the mixing temperature process, with an achievement rate of 0.5 exceeding the threshold of 0.33, and with no equipment alarm records within the corresponding time window, is also marked with a rapid achievement marker. The rapid achievement marker is associated with the process number, rapid achievement batch number, achievement rate value, and rate multiple. In the distribution of compliance rates, if no compliance events occur in the vulcanization time process, the corresponding column will be empty and no rapid compliance mark record will be generated. In the distribution of compliance rates, even if the compliance rate value of the batch to be verified exceeds the threshold, the rapid compliance mark will not be written. The compliance of the batch to be verified can only be determined after data verification.

[0055] The rapid compliance markers are traced and identified by tooling number, identifying recurring rapid compliance processes under specific tooling and generating tooling wear pseudo-compliance markers. Each batch in the rapid compliance marker corresponds to a tooling number in the production record. Tooling numbers cover key tooling such as extrusion die tooling and internal mixer rotor assembly. Searching by batch number from the batch records of production parameter data, the 24th batch of the extrusion pressure process uses tooling number M-047 corresponding to the extrusion die tooling, and the 29th batch of the internal mixer temperature process uses tooling number M-032 corresponding to the internal mixer rotor assembly. After grouping the rapid compliance markers by tooling number, the historical occurrence count of rapid compliance events under the same tooling number is statistically analyzed. Tooling M-047 has 4 rapid compliance events in the historical batch records, and tooling M-032 has 1 rapid compliance event. The identification criteria for tooling wear false compliance marks are that the cumulative number of historical rapid compliance events under the same tooling number exceeds 3. Tooling M-047 meets the criteria with 4 cumulative occurrences. Due to long-term use, the inner wall of the die of tooling M-047 wears, causing the cavity cross-sectional area to gradually expand. After the cavity expands, the cross-sectional size of the extrudate is larger, but the extrusion pressure is lower due to reduced resistance. The lower extrusion pressure manifests as a sharp drop in deviation to within the acceptable range during quality inspection, creating the appearance of false compliance. In reality, the extrudate size has deviated from the standard specifications due to tooling wear. Tooling M-032 does not meet the criteria with 1 cumulative occurrence. The rapid compliance behavior recorded for tooling M-032 in the rapid compliance mark is classified as unexplained due to non-tooling reasons. The tooling wear false compliance mark is associated with the tooling number, the cumulative number of historical rapid compliance events, the list of involved process numbers, and the wear risk rating. The wear risk rating is graded according to the cumulative number of occurrences: 3 to 5 times is medium risk, and more than 5 times is high risk. Tooling M-047 is rated as medium risk with 4 cumulative occurrences.

[0056] Based on the tooling wear pseudo-compliance markers, the compliance rate distribution is categorized and quantified to generate compliance judgment levels. In the tooling wear pseudo-compliance markers, batch 24 of the extrusion pressure process corresponding to tooling M-047 is marked as a rapid compliance candidate point in the compliance rate distribution. The hit of the tooling wear pseudo-compliance marker confirms that the rapid compliance behavior of this batch is caused by tooling wear. The compliance judgment level for this batch is marked as tooling wear pseudo-compliance. The recommended remedial measures are to replace or repair tooling M-047 and perform a full re-inspection of the batch. The re-inspection items should focus on verifying whether the cross-sectional dimensions of the extrudate and the thickness of the finished liner are systematically excessive due to the enlargement of the tooling cavity. For batches in the compliance rate distribution that hit the rapid compliance marker but not the tooling wear pseudo-compliance marker, the compliance judgment level is marked as pseudo-compliance with unknown cause. This level requires a full re-inspection of process parameters for this batch. Batches in the compliance rate distribution without rapid compliance events are divided into two categories based on whether the measured deviation is within the acceptable range: normal compliance and normal non-compliance. Normal non-compliance batches correspond to the correction markings in the monitoring feedback table. The pass / fail rating covers all batches and process combinations that have experienced pass / fail events in the pass / fail rate distribution. Each associated pass / fail rating label, pass / fail rate value, tooling wear pseudo-pass mark hit status, and recommended handling measures are stored in the rating summary information as the overall pseudo-pass rate index of the pass / fail rating.

[0057] Step S150: Based on the qualification judgment level, adjust the detection frequency to generate process qualification certificates, perform multi-process synchronous critical batch screening on the process qualification certificates to form quality stability labels, and output rubber lining production quality control instructions based on the quality stability labels.

[0058] Specifically, the frequency of inspections is adjusted based on the pass / fail rating to generate process pass certificates. For the 24th batch of the extrusion pressure process corresponding to the pseudo-pass record for tooling wear in the pass / fail rating, the inspection frequency is adjusted to full-volume, piece-by-piece inspection. The pseudo-pass mark for tooling wear in the pass / fail rating is the direct basis for triggering full-volume inspection, and the corresponding certificate level label is marked as a full-volume certificate. For the 29th batch of the mixing temperature process corresponding to the pseudo-pass record for cause pending investigation in the pass / fail rating, the inspection frequency is adjusted to full-batch double sampling inspection. The cause pending investigation mark requires simultaneous manual verification of the process parameter records for this batch, and the corresponding certificate level label is marked as an enhanced certificate. The inspection frequency for normally qualified batches in the pass / fail rating is executed according to the regular frequency of the comprehensive control plan, and the corresponding certificate level label is marked as a regular certificate. In the conformity assessment level, the process conformity certificate level label for pseudo-conforming batches is marked as a pending confirmation certificate. A pending confirmation certificate becomes effective only after a manual verification conclusion is written in, at which point it can be converted to a formal certificate status. The pending confirmation record for process conformity certificates is stored in a suspended state in the quality control system. During the suspension period, the corresponding batch of products is prohibited from being released into the warehouse and is physically isolated in a dedicated inspection area until the certificate status is converted to formal. Process conformity certificates are generated batch by batch, process by process. Each certificate is associated with a conformity certificate level label, the actual frequency of testing, the conformity assessment level source label, and the number of samples sampled. All process batch combinations covered by the conformity assessment level are written into the process conformity certificate.

[0059] In some embodiments, the step of performing multi-process synchronous critical batch screening on the process qualification certificates to form quality stability labels includes: extracting the distance values ​​between each process parameter and the qualification boundary from the process qualification certificates to generate a boundary distance distribution table; identifying batches whose multi-process parameter boundary distances are synchronously lower than the safety margin using the boundary distance distribution table to generate synchronous critical batch markers; identifying sections where multiple consecutive batches are continuously in a critical state using the synchronous critical batch markers to generate system drift markers; and classifying and calibrating the system drift markers and synchronous critical batch markers to generate quality stability labels.

[0060] A boundary distance distribution table is generated by extracting the distance values ​​between each process parameter and the acceptable boundary from the process qualification certificate. Each record in the process qualification certificate stores the measured deviation of the corresponding batch and process. The upper and lower boundaries of the acceptable boundary are inherited from the allowable deviation range of the graded quality control benchmark. The boundary distance is calculated using the formula d=min(|xU|,|xL|), where x is the measured deviation, U is the acceptable upper boundary, and L is the acceptable lower boundary. The smaller the d value, the closer the parameter is to the failure boundary. In the process qualification certificate, the measured deviation of the mixing temperature process for batch 26 is 7.1 degrees Celsius, the acceptable upper boundary is 8.0 degrees Celsius, and the boundary distance d=0.9 degrees Celsius; the measured deviation of the extrusion pressure process for the same batch is 7.6 MPa, the acceptable upper boundary is 10.0 MPa, and the boundary distance d=2.4 MPa; the deviation of the vulcanization time process for the same batch is 4.2 minutes, the acceptable upper boundary is 5.0 minutes, and the boundary distance d=0.8 minutes. The boundary distance distribution table stores the boundary distance values ​​and directional indicators from the upper or lower bound at the intersection of each batch and process. For batch and process combinations in the process qualification certificate with pending verification status, an "unverified" label is added to the corresponding position in the boundary distance distribution table. The size of the boundary distance distribution table is consistent with the number of batch and process combinations covered by the process qualification certificate, totaling 90 boundary distance value units (approximately 30 batches multiplied by 3 key processes). Densely populated sections in the boundary distance distribution table below the safety margin threshold correspond to vulnerable quality periods in the production process.

[0061] The boundary distance distribution table identifies batches where the boundary distances of multiple process parameters are simultaneously below the safety margin, generating synchronous critical batch markers. The safety margin thresholds in the boundary distance distribution table are set to 20% of the allowable deviation half-width for each process. For example, the allowable deviation half-width for the mixing temperature process is 8.0 degrees Celsius, corresponding to a safety margin threshold of 1.6 degrees Celsius; for the extrusion pressure process, the allowable deviation half-width is 10.0 MPa, corresponding to a safety margin threshold of 2.0 MPa; and for the vulcanization time process, the allowable deviation half-width is 5.0 minutes, corresponding to a safety margin threshold of 1.0 minute. The boundary distance distribution table is scanned batch by batch to identify the boundary distance values ​​for each process. Batch 26 has a mixing temperature boundary distance of 0.9 degrees Celsius, below the threshold of 1.6 degrees Celsius, and a vulcanization time boundary distance of 0.8 minutes, below the threshold of 1.0 minute. Both processes are simultaneously below the safety margin, meeting the synchronous critical identification condition. The situation where multiple process parameters are simultaneously approaching the acceptable boundary means that the batch of liners is on the verge of failure in multiple key quality dimensions. Even a small disturbance in any single process could cause the batch to cross the acceptable boundary and become unacceptable. The identification condition for synchronous critical batch marking is that two or more processes within the same batch have boundary distances below their respective safety margin thresholds. Batches 25 to 28 in the boundary distance distribution table all meet this condition. Synchronous critical batch marking is associated with the batch number, the list of triggering process numbers, and the number of synchronous critical processes. In the boundary distance distribution table, batch 26 has two processes (mixing temperature and vulcanization time) that are simultaneously below the safety margin, resulting in a synchronous critical process count of 2, the same as batches 25 and 27. Batch 28 has three processes that are simultaneously below the safety margin, resulting in a count of 3, which is marked as the highest critical severity in the synchronous critical batch marking.

[0062] System drift markers are generated for segments where multiple consecutive batches are in a critical state, identified by synchronous critical batch markers. Batch numbers from batch 25 to 28 in the synchronous critical batch markers are consecutively arranged. Consecutive arrangement is determined by a difference of 1 between adjacent batch numbers; four consecutive batches meeting this condition constitute a continuous critical segment. System drift markers are identified when three or more consecutive batches in the synchronous critical batch markers are in a synchronous critical state; four consecutive batches from batch 25 to 28 meeting this condition trigger system drift marker generation. The sequence of synchronous critical process numbers for each batch within a consecutive segment in the synchronous critical batch markers is 2, 2, 2, 3. The system drift marker records the average number of processes in this segment (2.25) as a quantitative indicator of drift severity. The fact that four consecutive batches were in a state of simultaneous criticality across multiple processes indicates a systematic shift in the overall process parameters of the production line. Viscosity fluctuations caused by uneven carbon black dispersion in the mixing process are amplified by the extrusion process and further transmitted to the vulcanization process. The parameter deviations in these three processes synchronously approach their respective qualification boundaries due to the amplification effect of the process chain, forming a systematic drift pattern of multi-process linked criticality. The system drift is marked by the starting and ending batch numbers of the associated segment, the number of synchronously critical batches within the segment, the union of triggering process numbers, and the average drift severity.

[0063] Quality stability labels are generated based on the classification and labeling of system drift markers and synchronous critical batch markers. The 25th to 28th batch segments hit by system drift markers are labeled as system drift levels in the quality stability labels. The four records of this segment in the synchronous critical batch markers are merged into one segment record in the quality stability labels. If there are isolated synchronous critical batches not covered by system drift markers in the synchronous critical batch markers, they are labeled as critical warning levels as single batches in the quality stability labels. All four records of the current synchronous critical batch marker are covered by system drift markers, and there are no isolated critical warning records in the quality stability labels. The average severity of system drift markers is converted into a joint risk score in the quality stability labeling. The joint risk score is calculated using the formula Q=Nc×sqrt(n), where Nc is the average number of synchronous critical processes in each batch within the system drift marker segment, and n is the total number of consecutive batches within the segment. The current joint risk score for the segment, Q=2.25×sqrt(4)=4.5, exceeds the high-risk threshold of 4.0. The recommended measures are to suspend production, investigate all triggering processes, and perform sample re-inspection on all batches of products within the segment. The conditions for resuming production are that the parameter boundary distances of the three triggering processes have all recovered to above the safety margin threshold and there are no synchronous critical triggers in two consecutive batches. The distribution table of boundary distances in the synchronous critical batch markers shows that the batches to be verified are marked with additional data to be verified labels in the quality stability labeling.

[0064] The production quality control instructions for rubber linings are generated based on the quality stability labels. These instructions are driven by the records in the quality stability labels, and the instruction type is jointly determined by the stability level label and the data verification status. System drift level triggers a pause instruction, critical warning level triggers an enhanced monitoring instruction, stability level triggers a normal release instruction, and pending verification label triggers a suspension instruction. If the joint risk score of the system drift level segment in the quality stability labels exceeds the high-risk threshold of 4.0, the production quality control instructions for rubber linings generate a pause instruction. The pause covers the three key processes: mixing, extrusion, and vulcanization. The starting batch of the pause corresponds to the starting number of the system drift label segment. During the pause, all work-in-process within the segment is transferred to an isolation area for inspection. During the pause, the maintenance team performs specific inspections on the wear condition of the mixing mill rotor, the screw clearance of the extruder, and the thermocouple array of the vulcanizing mill. The resumption condition for the pause instruction is that the boundary distances of the parameters in all three triggering processes have recovered to above the safety margin threshold and there are no synchronous critical triggers in two consecutive batches. In the quality stability labeling, the batch with the stability level corresponds to a normal release instruction. The release instruction includes a process qualification certificate number as a quality traceability credential. The certificate number and batch number are bidirectionally linked to ensure traceability. For batches with labels pending verification in the quality stability labeling, a suspension instruction is generated. The condition for the suspension instruction to transition to a formal instruction status is that data verification is completed and the verification conclusion is written into the corresponding record of the process qualification certificate. The quality control instructions for rubber lining production are associated with the instruction type label, the source number of the quality stability labeling, the scope of the executed process, the scope of the effective batches, and the transition conditions. All instruction execution results are written into the production record archive, forming a one-to-one mapping with the corresponding record of the quality stability labeling, serving as a closed-loop credential archive for this quality control cycle.

[0065] To implement the quality control method for a rubber liner production process corresponding to the above method embodiments, in order to achieve the corresponding functions and technical effects. See also Figure 2 , Figure 2 This application provides a structural block diagram of a quality control system 200 for a rubber liner production process, comprising: The adaptation analysis unit 201 is used to acquire production parameter data and process configuration parameters, and to perform cross-process linkage analysis on the production parameter data and the process configuration parameters to form a process matching coefficient. The quality control scheme unit 202 is used to perform defect severity statistics and generate a defect heat map according to the production parameter data, extract the process hierarchy based on the process configuration parameters and generate a process control list, map the defect heat map to the process control list to locate suspicious processes with sudden defect changes and establish a step-by-step quality control scheme. Alarm control unit 203 is used to apply weighted adjustment to the step-by-step quality control scheme based on the process matching coefficient to form a comprehensive control scheme, generate detection anomaly alarms for process sections where deviations are amplified and completely absorbed by the comprehensive control scheme, and determine graded quality control benchmarks by extracting failure distribution rules according to the detection anomaly alarms. The deviation monitoring unit 204 is used to perform step-by-step correction on the production parameter data through the graded quality control benchmark to generate a monitoring feedback table, perform deviation monitoring on the monitoring feedback table to generate a quality control trend chart, and use the quality control trend chart to detect parameter abnormalities, quickly reach the standard, trigger a false pass warning, and form a pass judgment level. The result output unit 205 is used to adjust the detection frequency based on the qualified judgment level to generate process qualified certificates, perform multi-process synchronous critical batch screening on the process qualified certificates to form quality stability labels, and output rubber liner production quality control instructions based on the quality stability labels.

[0066] The quality control system 200 described above for a rubber liner production process can implement a quality control method for a rubber liner production process according to the above method embodiments. The options in the above method embodiments are also applicable to this embodiment, and will not be detailed here. The remaining contents of this application embodiment can be referred to the contents of the above method embodiments, and will not be repeated in this embodiment.

[0067] The above embodiments are not an exhaustive list based on the present invention, and there may be many other embodiments not listed. Any substitutions and improvements made without departing from the concept of the present invention are within the protection scope of the present invention.

Claims

1. A quality control method for the production process of rubber lining sheets, characterized in that, include: Acquire production parameter data and process configuration parameters, and conduct cross-process linkage analysis on the production parameter data and the process configuration parameters to form process matching coefficients; Based on the production parameter data, defect severity statistics are performed to generate a defect heat map. Based on the process configuration parameters, process hierarchy is extracted to generate a process control list. The defect heat map is mapped to the process control list to locate suspicious processes with sudden defect changes and establish a step-by-step quality control plan. Based on the process matching coefficient, a weighted adjustment is applied to the step-by-step quality control scheme to form a comprehensive control scheme. The comprehensive control scheme identifies process sections where deviations are amplified and completely absorbed, and generates detection anomaly alarms. The failure distribution law is extracted according to the detection anomaly alarms to determine the graded quality control benchmark. The production parameter data is corrected step by step using the graded quality control benchmark to generate a monitoring feedback table. Deviation monitoring is performed on the monitoring feedback table to generate a quality control trend chart. The quality control trend chart is used to detect parameter anomalies, quickly reach the standard, trigger a false pass warning, and form a pass judgment level. Based on the aforementioned qualification level, the frequency of testing is adjusted to generate process qualification certificates. The process qualification certificates are then subjected to multi-process synchronous critical batch screening to form quality stability labels. Based on the quality stability labels, rubber lining production quality control instructions are output.

2. The method according to claim 1, characterized in that, The step of performing cross-process linkage analysis on the production parameter data and the process configuration parameters to form a process matching coefficient includes: Based on the production parameter data, the fluctuation distribution of key parameters in each process is extracted to generate a fluctuation distribution table; The fluctuation distribution table and the process configuration parameters are used to verify the upstream and downstream process associations and generate a process association difference table. The process association difference table identifies process pairs where upstream processes significantly influence downstream processes, generating fragile linkage markers. The process matching coefficient is generated by weighted integration of the fragile linkage marker and the process association difference table.

3. The method according to claim 1, characterized in that, The step of mapping the defect heatmap to the process control list to locate processes with suspected sudden defect changes and establishing a hierarchical quality control plan includes: The defect heatmap is used to extract the defect density of each process according to the process time period to generate a process defect density table; The process defect density table is used to identify processes with a sudden increase in defect density before and after shift change nodes, and a shift change surge marker is generated. The process defect density table identifies random spikes in processes caused by equipment malfunctions and generates equipment spike markers. A hierarchical quality control scheme is established by mapping the shift change escalation markers and equipment escalation markers to the process control list.

4. The method according to claim 1, characterized in that, The process segment for identifying deviation amplification and complete absorption in the comprehensive control scheme generates anomaly alarms, including: The comprehensive control scheme extracts the deviation transmission ratio between adjacent processes to generate a transmission ratio distribution table; Based on the conduction ratio distribution table, conduction amplification markers are generated for process sections where the conduction ratio is consistently greater than the reference value. The conductivity ratio distribution table is used to identify process sections where the conductivity ratio remains zero, and complete absorption markers are generated thereafter. An alarm for detection anomalies is generated based on the combined use of the conduction amplification marker and the complete absorption marker.

5. The method according to claim 1, characterized in that, The step of performing deviation monitoring and generating a quality control trend chart on the monitoring feedback table includes: The historical deviation records of each key parameter in the monitoring feedback table are extracted according to the process to generate a deviation history table; The deviation history table is fitted with the deviation trends of mixing temperature, extrusion pressure and vulcanization time to generate a trend fitting table. The trend fitting table is used to identify sections with significant differences in narrowing rate between different shifts of the same process, and shift difference markers are generated. A quality control trend chart is generated by weighted integration of the shift difference markers and the trend fitting table.

6. The method according to claim 1, characterized in that, The method of using the quality control trend chart to detect abnormal parameters and quickly reach the target to trigger a false pass warning and form a pass / fail judgment level includes: Extract the achievement rate of each process parameter from the quality control trend chart to generate the achievement rate distribution; Based on the aforementioned compliance rate distribution, identify process sections with abnormally high compliance rates and generate rapid compliance markers; The rapid compliance mark is traced and identified by tooling number to identify repeated rapid compliance processes under specific tooling, and tooling wear pseudo-compliance mark is generated; Based on the pseudo-qualification markers of tooling wear, the distribution of the compliance rate is classified and quantified to generate a qualification level.

7. The method according to claim 1, characterized in that, The process of simultaneously screening critical batches across multiple processes to generate quality stability labels includes: Extract the distance values ​​between each process parameter and the qualification boundary from the qualification certificate of the process to generate a boundary distance distribution table; The boundary distance distribution table identifies batches whose multi-process parameter boundary distances are lower than the safety margin, generating synchronous critical batch markers. The synchronous critical batch marker identifies the drift marker of the segment generation system that continuously generates multiple batches in a critical state; Quality stability labels are generated based on the system drift markers and the synchronous critical batch markers.

8. The method according to claim 2, characterized in that, The process of identifying upstream processes that significantly influence downstream processes and generating vulnerable linkage markers based on the process correlation difference table includes: The process association difference table is used to extract the dispersion of each pairing difference according to the process pairing type to generate a pairing dispersion table; Based on the pairing dispersion table, identify processes with abnormally high dispersion and generate a suspected abnormal pairing table; For the suspected abnormal pairing table, newly emerging linkage abnormal pairs after the historical stability is identified, new abnormal tags are generated; Based on the newly added anomaly markers, the suspected anomaly pairing table is classified and labeled to generate fragile linkage markers.

9. The method according to claim 5, characterized in that, The step of identifying sections with significant differences in narrowing rates between different shifts of the same process using the trend fitting table and generating shift difference markers includes: Based on the trend fitting table, the deviation narrowing rate of each process and shift is statistically analyzed to generate a shift rate distribution table. The shift rate distribution table is used to calculate the rate difference across shifts to identify processes whose rate differences exceed a preset tolerance and generate a rate difference process table. The rate difference process table is filtered by continuous differences in consecutive shifts to exclude occasional differences and generate continuous difference markers; Shift difference markers are generated based on the distribution density of the persistent difference markers.

10. A quality control system for the production process of rubber lining sheets, characterized in that, include: An adaptation analysis unit is used to acquire production parameter data and process configuration parameters, and to perform cross-process linkage analysis on the production parameter data and the process configuration parameters to form a process matching coefficient. The quality control scheme unit is used to perform defect severity statistics and generate a defect heat map according to the production parameter data, extract the process hierarchy based on the process configuration parameters and generate a process control list, map the defect heat map to the process control list to locate suspicious processes with sudden defect changes and establish a step-by-step quality control scheme. The alarm control unit is used to apply weighted adjustment to the step-by-step quality control scheme based on the process matching coefficient to form a comprehensive control scheme, generate detection anomaly alarms for process sections where deviations are amplified and deviations are completely absorbed by the comprehensive control scheme, and determine the graded quality control benchmark by extracting the failure distribution law according to the detection anomaly alarms. The deviation monitoring unit is used to perform step-by-step correction on the production parameter data through the graded quality control benchmark to generate a monitoring feedback table, perform deviation monitoring on the monitoring feedback table to generate a quality control trend chart, and use the quality control trend chart to detect parameter abnormalities, quickly reach the standard, trigger a false pass warning, and form a pass judgment level. The result output unit is used to adjust the detection frequency based on the qualified judgment level to generate process qualified certificates, perform multi-process synchronous critical batch screening on the process qualified certificates to form quality stability labels, and output rubber liner production quality control instructions based on the quality stability labels.

Images