A multi-category silica gel product forming integrated production method

Abstract

The present application relates to the technical field of silica gel product forming process control, and discloses a multi-category silica gel product forming integrated production method; for solving the problems of lacking criterion capable of quantitatively characterizing the residual removal degree and on-line verification of the criterion in the traditional method; the method first receives a switching task and freezes parameters, mold states and threshold versions; then, the mixing head and pipeline are discharged in volume pulse by partition, the cold runner and mold cavity generate event marker curves by air cycle, and pressure, flow, displacement and temperature are collected; finally, multi-index confidence is calculated according to threshold, strategy version integrity gate, verification records are generated, and the records that do not pass are rechecked by segmented pulse, temperature control disturbance and mold cavity boundary scanning, the records that pass are confirmed by the first piece, and the parameter validity and threshold version are updated, forming a switching closed-loop control.

Description

Technical Field

[0001] This invention relates to the field of silicone product molding process control technology, specifically to an integrated production method for molding multiple types of silicone products. Background Technology

[0002] The molding and manufacturing of silicone and liquid silicone rubber products generally consists of processes such as metering and mixing, conveying, injection or compression molding, heating and curing, demolding and removal, and finishing and inspection. To further improve production efficiency, existing technologies link hard rubber injection molding and silicone molding. For example, the published invention patent application CN107498778B discloses a production line that links injection molding and liquid molding, achieving linked molding through linkage groups, carrier transfer, and robotic arm transfer of different products. Another example is the published invention patent application CN112959568B, which discloses an automatic compression molding and vulcanization system and method for rubber products, achieving automatic feeding, vulcanization, and automatic mold changing to adapt to different drawing number tasks through a logistics system, compression molding and vulcanization unit, and mold storage. The above solutions can improve the automation level of the line and the efficiency of process connection, but there are still industry pain points for the key links of integrated molding production of multi-category silicone products. Specifically, in the integrated molding production of multi-category silicone products, when product formulations, colors, or hardnesses need to be frequently changed, silicone raw materials inevitably leave residues and stagnation in the metering mixing head, delivery pipelines, cold runners, and mold cavities. Because these residues differ from the new formulation in component ratios, coloring systems, hardness grades, and reactivity, cross-contamination is easily generated in the initial molding stage after the change, causing deviations in the process window for filling viscosity, vulcanization rate, and demolding state. This leads to quality problems such as short shots, abnormal flash, surface defects, or fluctuations in mechanical properties in the first piece and initial batches. Current production typically relies on experience-based cleaning, manual observation, or trial production verification to determine whether material and color changes have been completed, lacking the ability to... The lack of quantifiable criteria for characterizing the degree of residue removal, coupled with the absence of means to verify these criteria online, makes it difficult to predict switchover time and establish standardized control procedures. This not only prolongs the first-piece compliance cycle, reduces yield, and worsens batch consistency, but also makes it difficult to quickly locate the source of residue and trace and isolate it after quality anomalies occur, thus hindering the stable operation of integrated production lines for multiple product categories. Therefore, it is necessary to propose an integrated production method for silicone product molding and processing to establish quantifiable cleaning completion criteria and achieve online verification for cross-contamination and process drift caused by residue retention. This would improve the controllability of the switchover process, shorten the first-piece compliance time, and enhance batch consistency and traceability. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides an integrated production method for molding multiple types of silicone products, which solves the problems of traditional methods lacking criteria for quantifying the degree of residue removal and for verifying such criteria online.

[0004] To achieve the above objectives, the present invention provides the following technical solution: A method for integrated molding production of multiple types of silicone products, comprising: S1: Receive the switching task, freeze the current parameter version and mold status, generate task identifier and timestamp, and initialize the data window and verification rounds; S2: Perform residual discharge, discharge the mixing head and pipeline according to the discharge volume and pulse rhythm, discharge the cold runner and mold cavity according to the empty circulation rhythm, and collect pressure, flow, displacement and temperature data. S3: Perform online verification, obtain verification conclusions and confidence levels based on the data window, and generate verification records; S4: If the verification fails, determine the cleaning path according to the verification record, and sequentially perform pipeline segment pulse discharge, cold runner temperature disturbance discharge and mold cavity boundary scanning, and then enter online verification; S5: After verification, execute the first confirmation sequence, record the switch completion information, and update the parameter version validity identifier and threshold configuration version number.

[0005] In a preferred embodiment, receiving a handover task includes: After receiving the task switch, generate and queue a structured task record and perform field validation. The task number is generated based on the date code, the shift incrementing serial number, and the check digit. It is written into the switch start allowed time window and the scrap counter bound to the task number is initialized. The verification strategy version number is locked and the threshold configuration version number reference field is fixed.

[0006] In a preferred embodiment, the current parameter version and mold status are frozen, a task identifier and timestamp are generated, and the data window and verification rounds are initialized, including: When freezing the current parameter version and mold status, a freeze snapshot is generated and written with the snapshot number, snapshot timestamp, equipment setting set and recent cycle feature summary; Based on the alarm events, part retrieval failure count, and pressure over-limit count in the observation window, the mold status is determined and the status reason code is written. A health check of the data acquisition link is performed and the conservative mode flag is recorded.

[0007] In a preferred embodiment, residual discharge is performed by discharging the mixing head and piping according to the discharge volume and pulse rhythm, and discharging the cold runner and mold cavity according to the empty circulation rhythm, including: When discharging the mixing head and pipeline according to the discharge volume and pulse rhythm, a discharge configuration record is generated and locked. The discharge configuration record is written with the area boundary identifier, sampling frequency and degradation strategy parameter set. The discharge volume is determined based on pipeline geometric parameters and historical correction coefficients. Pulses are executed progressively according to the rhythm template. When the pressure sampling value reaches the product of the set pressure upper limit and the preset pressure close to the threshold, or when the pressure rise exceeds the pressure change rate threshold within the preset time window, a degradation record is written and additional actions are executed.

[0008] In a preferred embodiment, pressure, flow rate, displacement, and temperature data are collected, including: When collecting pressure, flow, displacement and temperature data, periodic curve packages are generated based on event markers and written into the curve package index. An injection speed perturbation sequence is introduced, and a data integrity check is performed. When any periodic curve package is missing an injection start event marker or a pressure holding end event marker, or when the pressure sequence length or displacement sequence length is less than the preset minimum number of points, or when the time interval from injection start to pressure holding end is less than the preset lower limit or greater than the preset upper limit of the target period duration, an empty loop is added and the missing reason code and conservative mode flag are recorded, and the discharge stage end timestamp is written to the data window metadata.

[0009] In a preferred embodiment, online verification is performed, and verification conclusions and confidence levels are obtained based on the data window to generate verification records, including: When obtaining the verification conclusion and confidence level based on the data window, extract pipeline segments, injection curve segments and product-level segments based on the threshold configuration version number and verification strategy version number, and perform integrity gating. The satisfaction score of the mandatory indicators is determined according to the threshold rule, and the confidence score is obtained by weighting the scores according to the weights. Generate verification records and write them into the fragment index set, mismatch type, risk level, and recommended cleaning path number.

[0010] In a preferred embodiment, if the verification fails, the cleaning path is determined according to the verification record, and the following steps are performed sequentially: segmented pulse discharge of the pipeline, temperature disturbance discharge of the cold runner, and mold cavity boundary scanning, including: If the verification fails, a path execution order is generated based on the verification record and task constraints and then entered into the database. The path execution order includes the path number, action template number, threshold index, exit rules, degradation rules, and rollback rules. Pressure thresholds are determined based on the upper limit of the equipment's rated pressure and the upper limit of the pressure set in the frozen snapshot. When the upper limit of the equipment's rated pressure is less than the upper limit of the pressure set in the frozen snapshot, the pressure threshold is taken as the upper limit of the equipment's rated pressure. When the upper limit of the rated pressure of the equipment is greater than or equal to the upper limit setting value of the pressure in the frozen snapshot, the pressure threshold shall be taken as the upper limit setting value of the pressure in the frozen snapshot. The remaining thresholds are obtained through the frozen version number index and written to the path execution form.

[0011] In a preferred embodiment, online verification includes: When sequentially executing pipeline segment pulse discharge, cold runner temperature disturbance discharge, and mold cavity boundary scan, the segment-level execution single parameter of the pipeline segment and the disturbance table parameter of the mold cavity boundary scan are determined according to the mismatch type. Short-cycle acquisition is performed between adjacent actions to obtain pipeline segments and cycle curve packages, and the start and end timestamps of the pipeline segments and the list of cycle numbers are written into the segment index set; When the abort condition is triggered, a rollback record is generated and the process is returned to the online verification process. At the same time, a resource approach marker is recorded at the process entry point.

[0012] In a preferred embodiment, after verification is passed, a first-article confirmation sequence is executed, including: After verification, the first-piece confirmation sequence is executed, and the release decision is made based on the data integrity mark and the risk release upper limit threshold. The number of confirmed pieces, the frequency of testing, and the statistical window are determined according to the first piece confirmation rule number. The production mode is switched to trial production status, and the original values ​​of quality characteristics are collected for each piece and associated with the periodic curve package index. When a test item is missing, adjust the release threshold level according to the downgrade table mapping and record the manually entered mark; The decision is made based on continuity judgment, statistical judgment, and curve inheritance conditions. If the conditions are not met, the first failure record is recorded and the process proceeds to the cleaning path or review path.

[0013] In a preferred embodiment, the switching completion information is recorded, and the parameter version validity identifier and threshold configuration version number are updated, including: When recording the switchover completion information and updating the parameter version validity identifier and threshold configuration version number, a switchover completion record is generated and written with the task number, verification round, cleaning round, scrap count, strategy version number, threshold version number, parameter version number, and key curve index set. Update the validity identifier and write the reason code according to the ternary key composed of parameter version number, equipment identifier and mold identifier; Tighten the temporary threshold triggered by resource approach and record it in the verification record. Adjust the threshold that meets the solidification conditions to generate or reference the threshold version number, and write it into the degradation record when the observation window is triggered.

[0014] Compared with the prior art, the present invention provides an integrated production method for molding multiple types of silicone products, which has the following beneficial effects: 1. This invention structurally records and queues switching tasks, freezing parameter versions, mold states, and threshold versions, unifying the primary key and judgment criteria for the switching process traceability; during the residual discharge stage, it discharges volume, pulse rhythm, empty cycle beat, and event marker curve packages according to spatial partitions such as mixing head, delivery pipeline, cold runner, and mold cavity, setting gating and degradation records for data collection integrity and pressure risk, and solidifying verifiable process evidence; during the online verification stage, it extracts pipeline segments, injection curve segments, and product-level segments based on threshold versions and verification strategy versions, executes strong integrity constraints, and establishes pipeline stability, injection curve convergence, and filling... The system employs multiple indicators, including consistency, demolding characteristics, and appearance consistency, to determine satisfaction and confidence levels. If verification fails, a cleaning path is selected based on the mismatch type and risk level, and a rollback strategy is implemented. Upon successful verification, piece-by-piece data collection, downgrade threshold control, and curve inheritance verification are performed according to the first-piece confirmation rule. Parameter version validity identifiers and threshold versions are updated uniformly, and an observation window review is conducted. This forms a standardized switching closed-loop control system where criteria are quantifiable, verifiable online, anomalies are locatable, and the process is traceable. This solves the problem in traditional methods of lacking criteria that can quantify the degree of residue removal and the need for online verification of these criteria.

[0015] 2. This invention divides the switchover process into four version-controllable elements: discharge configuration, verification configuration, cleaning path, and first-piece confirmation rules. It also sets verification rounds, scrap counts, and time constraints at the task level, ensuring the switchover process follows rule-driven action templates and threshold tables, reducing operator experience and the need for ad-hoc parameter adjustments. Through event-marked periodic curve packages and segment index sets, process signals on the pipeline and mold cavity sides are uniformly aligned and traceably verified. In case of mismatch, the system uses the mismatch type, corresponding to the partition, segment level, and disturbance cycle to determine the basis for downgrading or reverting the cleaning intensity. Furthermore, the invention restricts subsequent calls to abnormal tasks through parameter version ternary key validity identification and threshold version solidification or referencing mechanisms, supporting cross-batch and cross-mold configuration reuse and consolidation. Ultimately, this ensures the switchover process meets standard, reproducible, and consistent quality and safety constraints. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the integrated production process for molding multiple types of silicone products according to the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1: Figure 1 A method for integrated molding production of multiple types of silicone products is presented, including: S1: Receive the switching task, freeze the current parameter version and mold status, generate a task identifier and timestamp, and initialize the data window and verification rounds. The specific implementation is as follows: When the integrated molding production line is in stable piecework production, a switching task is initiated at the production planning end or on-site human-machine interface, and the switching process entry is entered. In order to provide a unified traceability key and time base for subsequent residue removal, online verification and first piece confirmation, the switching task is solidified from a structured record into a single task when it enters the task queue, and field verification and version locking are completed when it is enqueued. Task records are divided into required fields and extended fields. Required fields include task number, target product family number, target mold number, target parameter version number, allowable start time window for switchover, allowable scrap limit, verification strategy version number, first piece confirmation rule number, and traceability index key. Extended fields include task priority, task source, operator identifier, task creation timestamp, task status, and status reason code. Task numbers are alphanumeric combinations of 16 to 24 characters, generated by adding a date code, an incrementing serial number for the current shift, and a check digit. The date code is accurate to the day, and the serial number increments from 1 for the current shift. The check digit is obtained by concatenating the date code and serial number, mapping the result to a base-36 integer, and then taking the modulo 10. Letters are mapped from 0 to 35, and numbers are mapped to their original values. If a duplicate task number is detected in the queue, the task is rejected and a conflict reason code is written. The allowed start time window for switching is expressed as an absolute timestamp, with values ​​from 0 to 1800 seconds after the current time constraining the line beat, batch boundaries, and production scheduling window. If the current time exceeds the upper limit of the allowed time window, the task is placed in a pending manual confirmation state and a reason code is written. The allowed scrap limit is expressed as an integer, such as 20 pieces. When a scrap item is enqueued, the initial scrap counter is set to 0 and bound to the task number. Subsequent empty loop scrap and cleaning scrap are counted in this counter, constraining switching consumption. Verification strategy version number and threshold configuration version number are managed separately; task records lock the verification strategy version number and save the threshold configuration version number reference field. The threshold configuration version number reference field is used to determine a unique threshold set and threshold set during task execution; when the task is running, the threshold configuration version number reference field is parsed into a specific version number and a frozen snapshot is saved. Subsequently, only the specific version number can be referenced and cannot be changed. The judgment criteria can be synchronously reviewed. After the task is enqueued, a switchover preparation process is initiated. This process first freezes the current production context, generating a frozen snapshot. A traceability index key is written into the frozen snapshot, ensuring consistency between the frozen snapshot, task records, and subsequent full-process records. The frozen snapshot contains the snapshot number, snapshot timestamp, current parameter version number, current mold number, number of cavities, set of key equipment settings, a summary of process characteristics for the last 50 cycles, current quality judgment statistics, current message summary, and current piece count number. The snapshot number is an incrementing integer, bound to the task number, and used for the most recent snapshot in anomaly rollback. The set of key equipment settings is limited to settings related to the molding process and subsequent discharge and verification. At least the following settings should be written: temperature control setting, injection speed setting, switching point location setting, holding pressure time setting, pressure upper limit setting, and mold opening and closing cycle time setting. Each setting should be saved as a key-value pair, and the unit and allowable range should be written. The process characteristics of the most recent 50 cycles should be collected and statistically analyzed according to the preset field set. The preset fields should at least include the pressure peak, filling end pressure, holding pressure end pressure, cycle time, part removal time or part removal current peak. The mean, standard deviation, minimum value and maximum value should be calculated for each of the 50 cycles to serve as the baseline before switching. The current quality judgment statistics should at least include the pass count, fail count and fail type count, and the range of the most recent N pieces covered by the statistics, for example, N is 100 pieces. The mold status is synchronously fixed during the preparation process. The mold status is represented by an enumeration value and written with a status reason code. The enumeration value includes at least normal, maintenance required, and restricted operation. The mold status is determined based on the events and counting rules in the preset observation window, which selects the most recent 120 minutes or the most recent 300 cycles. If there is an unresolved high-level alarm, it is judged as restricted operation. If the number of consecutive part retrieval failures is not less than 2, it is judged as restricted operation. If the restricted operation conditions are not met and the pressure over-limit count exceeds the maintenance threshold, it is judged as maintenance required. The maintenance threshold is, for example, 1 time. If the restricted operation conditions are not met and the part retrieval failure count is 0 and the pressure over-limit count does not exceed the maintenance threshold, it is judged as normal. A pressure over-limit event is defined as a pressure sample value exceeding the set pressure limit for not less than 0.2 seconds, which is counted as 1 time. Repeated over-limit intervals of less than 2 seconds are combined into 1 time. The mold status and status reason code are written to the frozen snapshot and task process record as constraint inputs for subsequent discharge cycle, cleaning action intensity, boundary scan amplitude, and abnormal handling strategies. After freezing the snapshot, the control terminal reads the task creation timestamp from the task record and generates a switch start reference timestamp and an initial value for the verification round. The task creation timestamp is used to distinguish the enqueue time; the switch start reference timestamp is used to align the data window and the next data segment capture, and is also used to switch the end time of the preparation process. The time precision is 1 millisecond and a monotonically increasing clock source is used. The initial value for the verification round is 1 and written to the task record, and then updated according to the round each time online verification is entered. Then, the data window is initialized. The data window is a sliding window structure and is bound to the task number. The data window is divided into a process curve window and a statistical feature window. The system includes window and product-level window. The process curve window is used to cache continuous signal segments such as pressure, flow rate, displacement, and temperature, with a window length of, for example, 1200 seconds. The statistical feature window is used to cache fluctuation coefficients and stability statistics at a 30-second granularity, with a window length of, for example, 30 seconds and maintained in a scrolling manner. The product-level window is used to cache discrete records such as weight, appearance judgment, and size judgment for the most recent 30 pieces, with a window length of, for example, the most recent 30 pieces. During window initialization, the cache of the previous task is cleared, the window start point is aligned with the switching start reference timestamp, and the window configuration version number and window validity flag are written to ensure that data segments are retrieved according to the same window caliber during subsequent online verification. To reduce the uncertainty caused by missing data collection during verification, a health check is performed on the data collection link during the switch preparation flow. The health check range threshold, change rate threshold, and duration threshold are obtained using the device rated parameter table version number and threshold configuration version number index. Thresholds related to device capabilities are obtained from the device rated parameter table and sensor calibration parameters. Thresholds related to product family consistency are obtained from the product family baseline statistics corresponding to the threshold configuration version number. The threshold configuration version number is the version number frozen for this task. The health check performs online status verification and rationality verification for pressure, displacement, temperature, and flow signals. The pressure signal must be within the range of 0 to the set pressure upper limit, and the change within 100 milliseconds must not exceed 20% of the rated pressure. Displacement... The signal must be within the travel range, and the proportion of sampling points with a change of less than 0.01 mm within one second must not exceed 80%; the temperature signal must be within the set upper and lower limits, and an abnormality is determined when the deviation from the set value exceeds 15℃ and lasts for more than 60 seconds; the flow signal must not be all zeros, and the proportion of times the absolute value of the difference between adjacent samples exceeds 15% of the rated flow within one second must not exceed 30%, and there must be no abrupt change clusters where the absolute value of the difference between five consecutive sampling points exceeds this threshold; if any critical acquisition link does not meet the above verification conditions, an abnormal code is recorded and the task mode flag is set to conservative mode; in conservative mode, the subsequent verification threshold is increased according to the tightening table corresponding to the verification strategy version number, and the manual review entry is enabled; At the end of the preparation process, the task status is updated from "never executed" to "running," and a state transition record is generated. The state transition record includes at least the previous state, target state, transition timestamp, trigger source, and operator identifier. The trigger source is saved using a planned or manually enumerated value, and the operator identifier is generated by the login account or device number. The state transition record shares a traceability index key with the frozen snapshot, mold status record, window configuration record, and mode flag record. Then, for the residual removal stage, the task number, version number, counter, and round frozen in this step are used as the association key and constraint parameter for subsequent processes.

[0019] S2: Perform residual discharge, discharging from the mixing head and pipeline according to the discharge volume and pulse rhythm, and discharging from the cold runner and mold cavity according to the empty circulation rhythm, while collecting pressure, flow rate, displacement, and temperature data. Specific implementation details are as follows: After the task switching is completed and the switching preparation is finished, the residual removal stage begins. A removal configuration record, pulse event sequence, and empty cycle curve package are generated within the task number, and the index is written into the task process record and data window metadata as the first round of input for online verification. The residual discharge is divided into four regions based on spatial topology: the mixing head region, the delivery pipeline region, the cold runner region, and the mold cavity region. Discharge parameters and sampling frequencies are set for each region. The parameters of each region are locked and written into the discharge configuration record within this task. The discharge configuration record includes at least the region number, region boundary identifier, discharge volume target, discharge rhythm parameters, sampling frequency configuration, event flag list, and degradation strategy parameter set. The region boundary identifier is expressed using physical interfaces or valve position numbers, such as the mixing head outlet valve, pipeline segment valve, cold runner inlet, and mold cavity injection port, to ensure that the effective boundary of the discharge action can be located. The degradation strategy parameter set includes at least the pressure approaching threshold ratio, pressure change rate threshold reference, peak value reduction range, percentage reduction magnitude, continuous triggering criterion, and additional action lower limit. The pressure approaching threshold ratio is used to trigger in advance before the pressure approaches the set pressure upper limit. Degradation is implemented to reserve a safety margin for control response lag and pressure overshoot. Its default value is 0.90, and it can be configured within the range of 0.85 to 0.95 according to the equipment's rated pressure margin, mold status, and risk level. The default value of 0.90 is based on the premise of reserving a margin of no less than 10% for the inertial segment of pressure rise without significantly reducing discharge efficiency, so as to reduce the probability of exceeding the limit. The pressure change rate threshold is defined as the upper limit of the pressure rise within a 0.2-second window, in megapascals per 0.2 seconds, which can be converted to megapascals per second. The specific threshold is obtained by indexing the version number of the equipment rated parameter table and is written into the degradation strategy parameter set when the task enters this stage. When peak degradation, cycle time extension, or additional empty loop occurs during execution, a degradation record is generated and associated with the area number and the corresponding pulse sequence number or cycle sequence number. The degradation record is written in an append manner and does not overwrite the discharge configuration record. The discharge from the mixing head area and the delivery pipeline area is controlled by two methods: discharge volume and pulse rhythm. The discharge volume is determined based on the pipeline's geometric parameters and historical correction coefficients. The pipeline's geometric parameters are provided by a pipeline length and inner diameter registration table. The registration table fields include at least the section number, length, inner diameter, estimated volume, number of valves, number of elbows, and section valve positions. The estimated volume is calculated from the length and inner diameter and recorded on the registration table. The number of valves and elbows is used to identify the retention reinforcement section. The historical correction factor is used to compensate for the effective retention volume deviation caused by the internal cavities of valve chambers, elbows, tees, and mixing heads. The value range is 0.8 to 1.2, with an initial value of 1.0. The historical correction factor is updated in a restricted incremental manner, with each update not exceeding 0.05, and updates are only allowed after subsequent online verification is passed and the first piece confirmation is completed. The update is based on the switching fingerprint, which consists of the statistical summary of the last round of pulse event sequence in this stage and the morphological summary of the empty cycle curve package. It must contain at least the average pressure peak value, the standard deviation of pressure fluctuation, the standard deviation of flow fluctuation, and the injection curve deviation summary, and is associated with the task number and written into the traceability database for subsequent task correction factor write-back calculation. The target discharge volume is set to 2 to 3 times the estimated volume, for example, 2.5 times; the pulse rhythm consists of alternating high flow and low flow phases, with a single pulse cycle of 4 to 12 seconds, and the high flow phase accounting for 30% to 70%; the high flow peak is set to 70% to 95% of the rated flow, the low flow trough is set to 20% to 50% of the rated flow, and the number of pulses is set to 6 to 20 times, for example, 12 times.

[0020] The pulse rhythm is executed progressively, and the progressive parameters are selected by the rhythm template number, which is consistent with the verification strategy version number. The progressive is divided into three stages: the first third uses a high peak ratio and low flow stage as the overall replacement; the middle third reduces the peak and increases the low flow stage as the mixing and diffusion of the stagnation zone and the mainstream zone; and the last third increases the peak as the residual boundary of the flushing and mixing, so that the discharge takes into account both the replacement rate and the release of the stagnation zone. During pulse execution, the transport flow rate and transport pressure are continuously collected at a sampling frequency of 20 Hz, and the sampled data is written to the process curve window of the data window. The control terminal records the start and end timestamps of each pulse to form a pulse event sequence. The pulse event sequence fields include at least the pulse number, peak value setting, valley value setting, percentage setting, execution start and end timestamps, maximum and minimum pressure values ​​during execution, and maximum and minimum flow rates during execution. To constrain the risk of pressure surges, when the pressure rise exceeds the pressure change rate threshold within 0.2 seconds, or when the pressure reaches the product of the set pressure upper limit and the pressure approaching the threshold, the control terminal performs degradation processing on the next pulse, reducing the peak value by 5% to 15% and the high flow rate percentage by 10%, while generating a degradation record. The degradation record includes at least the degradation reason code, degradation magnitude, and degradation duration pulse number. If three consecutive pulses trigger degradation processing, the discharge cycle of this region is extended according to the degradation strategy parameter set, and at least two additional pulses are added. In the cold runner and mold cavity areas, empty cycle execution is used to form a cycle curve package for online verification. Empty cycle refers to driving the injection action according to the injection time axis corresponding to the parameter version and producing test parts or scrap parts, without entering the normal part counting. The scrap parts generated by the empty cycle are counted in the allowable scrap counter. The empty cycle cycle is set to 5 to 12 cycles, for example, 8 cycles. During the empty cycle, the injection pressure, injection displacement, and injection speed are sampled at 200 Hz, and the mold temperature is sampled at 2 Hz. When the part picking end provides the part picking current or part picking time signal, it is collected synchronously and aligned according to the cycle number. To ensure that the curve can be aligned and can be reviewed, a fixed time is extracted for each empty cycle. The time segments are fragmented and packaged into a periodic curve package. The time segment boundaries are determined by event markers, and are uniformly taken as 0.2 seconds before the start of injection to 0.5 seconds after the end of holding pressure. The periodic curve package fields must include at least the period number, time axis, pressure sequence, displacement sequence, velocity sequence, temperature sequence, and event marker set. The event marker set must record at least the timestamps of mold closing completion, injection start, holding pressure start, holding pressure end, mold opening start, and part removal completion, and satisfy the constraint that the timestamps are incremented and that no key events are missing. The target period duration is obtained from the period duration field of the parameter version, or calculated from the time interval from the start of injection to the completion of part removal in the event markers, and is used for the time upper limit determination of integrity check. To enhance the observability of response differences due to end-stage residues, a fixed perturbation sequence is introduced into the empty cycle time while maintaining the core configuration of the parameter version unchanged. The core configuration includes at least the switching point location setting, holding time setting, and pressure upper limit setting remaining unchanged. The perturbation only makes small step changes to the injection speed setting, with a step amplitude of ±5% and a step cycle of switching every 2 empty cycles, forming a low-high-low or high-low-high sequence. The perturbation sequence number is written into the discharge configuration record, and the corresponding sequence site identifier is written into the event marker extension field of the cycle curve package, enabling subsequent online verification to be grouped and compared by sequence. During the discharge process, an online data acquisition integrity check is set up to ensure sufficient input for subsequent online verification. The integrity check requires that each cycle curve package has at least two event markers: injection start and pressure holding end, and the length of the pressure sequence and displacement sequence is no less than 300 points. At the same time, the time interval from injection start to pressure holding end is no less than 1 second and no more than 1.5 times the target cycle length. If a missing or abnormality is detected, the control terminal adds 1 to 3 empty cycles to supplement the data and records the missing reason code. The missing reason code should at least distinguish between sensor offline, sampling interruption, missing event marker, time axis abnormality, and insufficient data length. If the integrity rules are still not met after adding the missing data, the integrity failure mark is recorded and the task mode flag is maintained or set to conservative mode, so that subsequent online verification will execute the corresponding strategy according to the mode flag. After the residue is discharged, the discharge configuration record index, pulse event sequence index, empty cycle period curve package index, append empty cycle count and waste counter increment are written into the task process record. The discharge stage end timestamp is written into the data window metadata as the boundary for subsequent online verification and extraction of data segments.

[0021] S3: Perform online verification, obtain the verification conclusion and confidence level from the data window, and generate a verification record. The specific implementation is as follows: After the residual removal is completed, the online verification phase begins. Online verification occurs when the task is in the running state and the data window is valid. The removal end timestamp is used as the right boundary of the pipeline segment backtracking by reading the metadata of the data window. Online verification generates verification records, which are bound to the task number and verification round number. At least the original values ​​of the indicator set, threshold configuration version number, satisfaction score set, confidence level, verification conclusion, mismatch type, suggested cleaning path, risk level, and data integrity flag are written in the record. The fragment index set and resource usage flag are also input. At the start of online verification, the control terminal reads the verification configuration locked by the task using the threshold configuration version number and verification strategy version number as indexes. The function of the verification configuration is to provide the input segment caliber and judgment caliber for online verification, and should at least include the segment extraction range, mandatory indicator set, satisfaction mapping rules, confidence weight, conclusion threshold, mismatch type priority table, risk level mapping rules, and path output rules. Under the verification configuration, the control terminal extracts segments. Segment extraction is divided according to signal category. For pipeline segments, the right boundary backtracking continuous sampling data is taken, and the backtracking time is, for example, 30 seconds. For injection curve segments, the most recent empty cycle periodic curve package is taken, and the number of periodic curve packages is, for example, 8. For product-level segments, the most recent product discrete records are taken, and the number of discrete records is, for example, 10 pieces. After the segment extraction is completed, the input segment index set is written into the verification record. The index set should include the start and end timestamps of the pipeline segments, the cycle number list corresponding to the injection curve segments, and the piece number list corresponding to the product-level segments, and the segment extraction configuration number is written in, marking the segment caliber of this round of verification. Online verification uses data integrity gating to constrain the availability of verification inputs. The data integrity gating employs strong constraints, disallowing a pass conclusion if any critical segment is insufficient. Insufficient critical segments must meet at least one of the following conditions: the number of sampling points for the pipeline segment is less than the configured minimum number (e.g., 600 points); the number of periodic curve packages for the injection curve segment is less than the configured number, or any periodic curve package is missing an injection start event marker or a holding pressure end event marker; the number of product-level segments is less than the configured number, or product-level records are missing critical fields; the configuration of the number of sampling points for the pipeline segment... The minimum number of points is determined by the segment backtracking duration and sampling frequency. For example, 600 points correspond to a 30-second backtracking and a sampling frequency of 20 Hz. When a critical segment is found to be insufficient, the data integrity flag is set to false, and the verification conclusion is determined based on whether the missing data involves mandatory indicator input: if the missing data involves mandatory indicator input, the verification conclusion is set to failed; if the missing data only involves non-mandatory indicator input, the verification conclusion is set to require review. The missing data type, missing field list, and missing segment index are written into the verification record. The missing data types include at least sampling interruption, missing event flag, insufficient data length, missing fields, and timeline anomalies. After data integrity is marked as true, the indicator verification process begins. Crossover and drift caused by residual retention are described as consistency constraints of multi-source observable characteristics, judged based on threshold version numbers and verification strategy version numbers. The indicator set should include five indicators: pipeline stability, injection curve convergence, filling consistency, demolding characteristics, and appearance consistency. Pipeline stability is fluctuation on the mixing and conveying side; injection curve convergence is deviation in injection morphology; filling consistency is end-filling and product-level stability; demolding characteristics are part removal and demolding status; and appearance consistency is appearance deviation. All indicators are based on signal statistics and periodic events, without incorporating formulation, hardness, or color ratio parameters. Pipeline stability is expressed as the fluctuation level of flow and pressure, using fluctuation coefficient statistics. The calculation window and threshold of the fluctuation coefficient are indexed by the threshold configuration version number, for example, the flow fluctuation coefficient is no higher than 0.05 and the pressure fluctuation coefficient is no higher than 0.06. Injection curve convergence is expressed as the morphological deviation of the pressure and displacement time segments in the periodic curve package. The morphological deviation is aligned with the segment from the start of injection to the end of holding pressure, using event markers to align the time axis, and the deviation ratio is obtained after scaling by peak amplitude or segment integral amplitude, with a threshold of, for example, a deviation ratio no higher than 10%. Filling consistency is jointly expressed by cycle-end characteristics and product-level statistical consistency. Cycle-end characteristics include at least the deviation level between the filling end pressure and the holding pressure end pressure, and product-level statistical consistency includes at least the weight. The coefficient of variation, for example, is a weight coefficient of variation not exceeding 1.2%; demolding characteristics are expressed by the stability of the picking signal and the failure count, with a threshold of, for example, 0 picking failures and the fluctuation of picking current or picking time not exceeding the configured threshold; the fluctuation of picking current or picking time is expressed using standard deviation or coefficient of variation; appearance consistency is expressed by the standardized deviation value output by the vision device, with a threshold of, for example, a standardized deviation value not exceeding 1.5; the standardized deviation value is the normalized deviation score calculated by the vision device for the selected appearance feature vector, and the normalization benchmark is the mean and standard deviation of the product family baseline or the upper limit of the range recorded by the threshold configuration version number, in order to achieve dimensional consistency between different batches and different devices; all the above thresholds are obtained by the threshold configuration version number index, and the threshold configuration version number is the frozen version for this task; The threshold configuration version number corresponds to a threshold generated from the baseline sample of the stable piece-rate production stage of the product family, combined with inspection capability constraints. The baseline sample range is at least 500 pieces or data from the most recent 120 minutes. Continuous quantity thresholds are generated using the quantile method or the mean plus standard deviation method, such as the 95th quantile or the mean plus 2 times the standard deviation. The threshold configuration version generation record must at least include the threshold generation rule, baseline sample range, collection start and end timestamps, equipment identifier, and mold identifier. Discrete event thresholds are generated using the upper limit of the count, and the generation record includes the statistical window length and counting caliber. Threshold updates generate a new threshold configuration version number, including the generation timestamp and applicable scope. The threshold configuration version number referenced in this task is frozen during task execution, and the verification caliber is consistent within the same task. Online verification maps each indicator to a satisfaction score, ranging from 0 to 1. The relative relationship between the original indicator value and the threshold determines the satisfaction score. If the original indicator value is within the threshold's allowable range, the score is higher; as it approaches the threshold boundary, the score decreases. Once the original indicator value exceeds the threshold boundary, the score decreases towards 0. The satisfaction mapping uses a segmented rule and sets a buffer. The buffer width is determined according to the threshold configuration version number index, for example, 10% to 30% of the threshold. When the original indicator value reaches the buffer, the satisfaction score increases linearly or decreases linearly in segments along the normalized distance between the original value and the threshold boundary. The segment boundaries and buffer width are recorded in the verification record as the verification criteria. The confidence level is obtained by weighting the scores of each indicator. The weights are indexed by the verification strategy version number and locked within this task, such as 0.2 for pipeline, 0.35 for injection curve, 0.25 for filling consistency, 0.1 for demolding, and 0.1 for appearance. The confidence threshold is determined by the verification strategy version number, such as 0.9 for pipeline. The verification conclusion is determined by the superposition of two rules: mandatory indicator gating and confidence level gating. When the data integrity is marked as true, the mandatory indicators are met, and the confidence level is not lower than the threshold, it is judged as passing. The set of mandatory indicators is set according to the product family, and at least pipeline, injection curve, and filling consistency are set as mandatory indicators. If any indicator is not met, it is judged as failing and entering the mismatch type judgment. If all mandatory indicators are met and the confidence level is lower than the threshold, it is judged as needing to be reviewed. The need-to-review mark is written into the verification record. The conservative path mark or conservative combination path number is output in the suggested cleaning field to constrain the path selection and recording scope in subsequent stages. The mismatch type is determined using a priority table, which is fixed within the verification strategy version number. The priority rules are risk-oriented, with equipment safety-related events having the highest priority, followed by injection curve drift and uneven filling, then pipeline fluctuations, and appearance deviation having the lowest priority. A compound contradiction is triggered when mandatory indicators are simultaneously abnormal, or when indicator conclusions are inconsistent. The contradiction determination conditions must at least meet the following: pipeline stability is met, injection curve deviation is higher than the threshold, injection curve convergence is met, product-grade consistency is higher than the threshold, or other indicators are met when demolding fails. After a compound contradiction is triggered, the verification record is written to the contradiction pair list, the original values ​​of each indicator, and the set of input fragment indices that triggered the contradiction. The contradiction pair list is a combination of indicator pairs and abnormal directions. The recommended cleaning path is determined by the mismatch type, confidence level, and resource usage flag, and outputs using path numbers, which are related to the verification strategy version number. The recommended cleaning path should include at least a pipeline reinforcement path, a temperature control disturbance path, a cavity boundary scan path, and a conservative combination path. For mismatch types of pipeline fluctuation, output a pipeline reinforcement path; for mismatch types of injection curve drift, output a combined path of temperature control disturbance and cavity boundary scan; for mismatch types of inconsistent filling, output a cavity boundary scan path with additional pipeline reinforcement; for mismatch types of abnormal demolding, output a conservative combination path with a higher risk level; for mismatch types of appearance deviation and other mandatory indicators being met, output a lightweight path or an output requiring verification. Risk levels are expressed in three tiers (low, medium, high) or as integers from 0 to 3, determined by the mismatch type and safety event triggering conditions. When any of the following events occurs: pressure approaching the upper limit, temperature control anomaly, or demolding failure, the risk level is set high, and the recommended cleaning path is limited to a conservative combination path. The triggering event and its corresponding threshold are written into the verification record. After online verification is completed, the control terminal generates verification conclusions and a traceability database. The traceability database is stored using the task number and verification round number as a composite key and is not overwritten. The verification round number is incremented by 1, and the verification conclusions and suggested cleaning paths are sent to the process state machine as input for the next stage. When the verification result indicates that a review is required, it is sent to the process state machine through a manual review prompt, and a review requirement is marked in the task status reason code to constrain the execution and recording criteria for the next step.

[0022] S4: If the verification fails, determine the cleaning path according to the verification record, and sequentially perform pipeline segment pulse discharge, cold runner temperature disturbance discharge, and mold cavity boundary scanning, followed by online verification. The specific implementation is as follows: When the online verification output result is "fail" or requires re-verification, the cleaning path is executed. The control terminal reads the verification record based on the task number and verification round number. According to the frozen time limit, scrap limit, threshold configuration version number, verification strategy version number, abnormal strategy configuration number and mold status identifier in the verification record, a warehouse entry path execution order is generated. The path execution order is bound to the task number and verification round number, and at least the path number, action sequence template number, action parameter set, exit condition, degradation rule number and rollback strategy enumeration are written. The pressure limit threshold is the smaller value between the equipment rated pressure limit and the frozen snapshot pressure limit setting value, and is written to the segment-level execution order and the disturbance table safety limit field, respectively. Other thresholds are obtained according to the frozen version number index and written to the path execution order. The cleaning action sequence follows a chain, starting with segmented pulse discharge from the pipeline, temperature disturbance discharge from the cold runner, and mold cavity boundary scanning. The action parameters are mapped from mismatch type to action intensity, and risk level and resource occupancy marker are mapped to exit, downgrade, and rollback. Each round of cleaning generates an input segment that can be used for the next round of online verification, and the state machine returns to online verification. The pipeline segmented pulse discharge is based on the pipeline registration form as the basis for project implementation. Pulse parameters are discharged segment by segment according to segment number. During the switchover preparation stage, the pipeline registration form must freeze fields including at least the segment number, segment length, segment inner diameter, segment volume, number of valves, number of elbows, and segment boundary valve position identifier. For example, the number of segments is 6. After input, the control terminal generates segment-level execution orders by segment; the segment-level execution order must include at least the segment number, number of pulses, peak ratio, valley ratio, cycle duration, high flow rate ratio, target discharge volume, segment-level pressure upper limit threshold, and segment-level degradation threshold. The number of pulses per segment is 3 to 8, and the value is determined by the mismatch type and risk level. When the mismatch type is pipeline fluctuation, increase the flushing intensity, set the peak high flow rate to 90% of the rated flow rate, take the upper limit of the pulse number, and set the high flow rate ratio to 60% to 70%; when the mismatch type is injection curve drift, emphasize mixing and diffusion, set the peak high flow rate to 80% of the rated flow rate, and extend the low flow rate stage; when the mismatch type is filling inconsistency, use a medium peak value and increase the total discharge volume to 3 times the estimated volume. During each execution period, flow rate and pressure are sampled at 20 Hz, and pressure proximity is calculated in real time. When the pressure approaches 90% of the segment-level pressure upper limit threshold, peak degradation is triggered. To suppress transient spikes, a pressure slope threshold is set at the control end. When the pressure rises by more than 20% of the rated pressure within 0.2 seconds, it is determined that the slope exceeds the limit and peak degradation is triggered. The pressure slope threshold is obtained by indexing the version number of the equipment rated parameter table, and after being converted to a 0.2-second window according to the sampling period, it is written into the segment-level degradation threshold field of the segment-level execution order. After peak degradation is triggered, the degradation range is 5% to 15%, while the high traffic percentage is reduced by 10% and the pulse period is extended by 2 to 4 seconds. The degradation event is written to the segment-level event in a structured record. The segment-level event must at least include the trigger threshold, degradation range, number of degradation pulses, and recovery conditions. After each segment is executed, a segment-level summary is generated and written into the task process record. The segment-level summary includes at least the segment number, actual number of pulses, actual discharge volume, execution time, segment-level flow fluctuation summary, segment-level pressure fluctuation summary, and anomaly code. The task number and verification round number are used as index keys to achieve traceability of pipeline-side action results. After the segmented pulse discharge of the pipeline is completed, the cold runner temperature disturbance discharge stage begins. The cold runner temperature control setting within the allowable range of the equipment is subjected to small-cycle disturbances to amplify the response differences in the local stagnation area and release them. The allowable temperature control safety range of the equipment is obtained from the version number index of the equipment rated parameter table, and the upper and lower limits of the disturbance are determined based on the temperature control settings of the frozen snapshot of this task. The temperature disturbance parameters are configured using a tripartite structure of amplitude, period, and total duration. The disturbance amplitude is, for example, ±8 degrees Celsius, the disturbance period is, for example, 120 seconds, and the total disturbance duration is, for example, 240 to 600 seconds. When the control terminal enters this action, it reads the mold status flag. When the mold status is restricted operation, the disturbance amplitude is tightened to ±5 degrees Celsius, and the lower limit of the total disturbance duration is set to 360 seconds to reduce the risk of thermal shock. The conditions for enabling and skipping disturbances are jointly determined by the mismatch type and the temperature control status. Disturbances are enabled when the mismatch type is injection curve drift or filling inconsistency. Disturbances are allowed to be skipped when the mismatch type is pipeline fluctuation and the temperature control response is normal and the cold runner temperature is stable, but a skipping reason code must be written, along with the original value of the temperature control stability index and the threshold judgment result. The temperature control stability index includes at least the actual temperature fluctuation amplitude, the average set deviation, and the response lag time. The temperature control stability threshold is obtained according to the threshold source rule index, where the upper limit of the response lag is taken from the equipment capability threshold corresponding to the version number of the equipment rated parameter table, and the upper limit of the temperature fluctuation and the upper limit of the set deviation are taken from the product family consistency threshold corresponding to the threshold configuration version number. The threshold values ​​are then written into the path execution form. During the disturbance process, 2 to 4 empty cycles are executed simultaneously to collect periodic curve packets. The empty cycle beat and sampling frequency use the configuration frozen in the discharge stage, so that injection curve segments and temperature segments that can be used for the next round of online verification can be formed immediately after the disturbance. The control end sets temperature control response monitoring rules. When the actual temperature deviates from the set temperature by more than 5°C and lasts for more than 60 seconds, the temperature control abnormality code is recorded, the disturbance amplitude of the subsequent mold cavity boundary scan is tightened, the risk level is increased by 1 level, and a temperature control abnormality flag is added to the resource occupation flag. The process then proceeds to cavity boundary scanning. Cavity boundary scanning is used to characterize the effects of end-of-line residues as a consistency verification process for controlled micro-perturbations, ensuring that the adequacy of cleaning is reflected through the stability of process curves under perturbation conditions and product-level statistics. Boundary scanning performs sequential perturbations on injection speed, switching point displacement, and holding time. The perturbation amplitude is, for example, ±10% for injection speed, ±1.5 mm for switching point displacement, and ±10% for holding time. The perturbation sequence length is set to 3 to 9 cycles, for example, 5 cycles. The switching point displacement is represented by a uniform internal unit, converted to millimeters using a calibration coefficient. The calibration record is written to the calibration record corresponding to the version number of the equipment parameter table. The calibration record must include at least the calibration timestamp, calibration method number, conversion coefficient to millimeters, applicable equipment identification, and expiration date to ensure the perturbation amplitude is reproducible. The perturbation order is determined based on the mismatch type. For mismatch types such as injection curve drift (small to large), the order is... For inconsistencies in filling, the process proceeds from largest to smallest. For verification conclusions requiring review or mismatches that are complex and contradictory, a forced "small to large" approach is adopted. After each disturbance cycle, injection curve segments and product-level statistics are recorded as alignment inputs for subsequent online verification. To enhance reproducibility, a disturbance table is generated and written to the task process record upon entering the boundary scan action. The disturbance table fields must include at least the cycle number, speed offset ratio, switching point offset, holding time offset ratio, expected cycle duration, and safety upper limit. Subsequent online verification is based on the aligned segments in the disturbance table, using the disturbance table index to locate the corresponding curve segments and product-level records. There are mandatory termination conditions during boundary scans. If demolding fails, pressure is too high, or a marker is missing, the scan will enter a rollback branch and a manual review branch, generating a rollback record. The rollback record must correspond to the task number and verification round number, and include a rollback strategy enumeration, trigger reason code, and post-rollback mode flag value to ensure traceability and explainability of anomalies. During the cleaning path execution, short-cycle data acquisition is performed after each type of action, and a data acquisition record is generated. Short-cycle acquisition captures at least 30 seconds of tubing segments and two cycles of injection curve packages, and writes the corresponding segment index set into the path execution sheet. The segment index set must at least contain the start and end timestamps of the tubing segments and a list of cycle numbers. After the cleaning path execution is completed, the control terminal writes the actual cleaning action sequence, action parameters, execution duration, waste count increment, exception code set, and segment index set into the task process record, and switches the process status to the online verification entry. The control terminal continuously monitors the cumulative cleaning time percentage and the waste count percentage. For example, when the cumulative time consumption reaches 80% of the allowed switching time limit or the scrap count reaches 80% of the allowed scrap limit, the risk level will be increased, and the subsequent path output will be limited to a conservative combination path. At the same time, a resource approach flag will be carried when writing back to the online verification entry. The resource approach flag must include at least the time consumption percentage and the scrap percentage. After reading the resource approach flag, the online verification will adjust the verification judgment conditions according to the preset tightening rules. The adjustment method is to raise the confidence threshold or lower the buffer limit. For example, the confidence threshold will be raised by 0.02 to 0.05 or the buffer limit will be lowered by 10%. The adjusted judgment conditions will be written into the verification record.

[0023] S5: After verification, execute the first confirmation sequence, record the switch completion information, and update the parameter version validity identifier and threshold configuration version number. The specific implementation is as follows: When the online verification record concludes as passed and the data integrity is marked as true, the process enters the first-item confirmation sequence. If the risk level of the verification record exceeds the risk release upper limit threshold corresponding to the verification strategy version number, or if the prohibition event corresponding to the abnormal strategy configuration number is triggered, the process does not enter the first-item confirmation sequence but is transferred to the review-required branch or the restricted production branch. The prohibition reason code is written into the task status reason code, and the trigger event type and trigger threshold are recorded. The risk release upper limit threshold is expressed as an enumeration of risk levels; for example, the highest risk level allowed for first-item confirmation is medium risk, and high risk prohibits entry into first-item confirmation. The release stop event and its threshold, duration, and counting method are determined by the abnormal strategy configuration number index, and include at least pressure over-limit events, temperature control abnormality continuous events, demolding failure continuous events, and key detection item missing events; among them, the threshold for pressure over-limit events is the smaller value between the upper limit of the equipment's rated pressure and the upper limit of the frozen snapshot pressure; the threshold for temperature control abnormality continuous events is, for example, a deviation of more than 5°C between the actual temperature and the set temperature that lasts for more than 60 seconds; the threshold for demolding failure continuous events is, for example, two consecutive demolding failures; and the threshold for key detection item missing events is, for example, two or more missing items. All related configurations are written to the task record and are traceable. The first-piece confirmation sequence is indexed by the first-piece confirmation rule number in the task record. The first-piece confirmation rule number is prefixed with the product family number and is frozen and written to the task record during the switchover preparation phase. The first-piece confirmation rule number cannot be changed during the first-piece confirmation process. The first-piece confirmation rule must define at least the number of confirmed pieces, the detection frequency, the quality feature judgment set, the statistical window length, the anomaly handling strategy, the release threshold level, and the degradation table mapping. An example of the number of confirmed pieces is 6 pieces, the detection frequency is every detection, and the statistical window length is the most recent 6 pieces. The anomaly handling strategy must define at least the return to the cleaning path after failure, the entry into manual review, and the entry into restricted production, and define the trigger priority. The trigger priority is determined sequentially based on risk orientation. When a safety event occurs, or there are consecutive failures or missing detections, manual review or restricted production is given priority. The degradation table mapping must define at least the relationship between the number of missing detection items and the tightening level. When one item is missing, the number of confirmed pieces is increased by 2, and the statistical window length is increased by 4. When two or more items are missing, the process is directly transferred to the branch requiring review. The calculation result is written to the release threshold level field and bound to the first-piece confirmation rule number and version number to ensure that the release criteria are consistent and can be reviewed when the detection capability decreases. Before the first article confirmation begins, the control unit switches the production mode to trial production. Output in trial production is not included in the formal production volume, but it needs to be written to the traceability database piece by piece, corresponding to the task number. The control unit operates production based on the number of confirmed pieces determined by the first article confirmation rule number. For each piece produced, a product-level data acquisition is triggered and written to the first article confirmation cache. The first article confirmation cache consists of the task number and piece number. The cache needs to record the piece number, acquisition timestamp, set of original quality characteristic values, associated cycle curve package index, associated verification round number, and acquisition source marker. The set of original quality characteristic values ​​is written according to the settings corresponding to the first article confirmation rule number, and must include at least weight and appearance criteria. The code selectively writes important size judgment codes and color deviation values ​​according to the product family configuration; the data source mark is used to identify automatic data collection and manual input; when the detection equipment is unavailable, the detection data is missing, or the detection data is abnormal, the control terminal switches the quality-affecting characteristics to manual input according to the first article confirmation rules, and forcibly writes the input account and input timestamp; the control terminal calculates the tightening result of the release threshold level according to the degradation table mapping, and writes the tightening result to the first article confirmation cache and the first article confirmation process record; when the tightening result meets the review trigger condition in the rule, the control terminal terminates the first article confirmation sequence and switches to the review branch to avoid the release conclusion that cannot be reviewed due to insufficient detection capability; First article confirmation employs both continuity and statistical criteria in parallel. The criteria are based solely on measurable features at the product level and process curve indices. Continuity confirmation requires all parts within the corresponding part number to meet the quality characteristic judgment set. This set is determined by the target product family configuration and is frozen within the first article confirmation rule number, including at least weight and appearance defects. For size-sensitive product families, critical dimension judgment is strengthened; for appearance consistency-sensitive product families, color deviation judgment is strengthened. Statistical judgment uses continuous quantities such as weight and dimensions, or coefficient of variation, or deviation thresholds. Appearance defects are judged using defect counts, and color deviations are judged using deviation thresholds. The statistical window length is determined by the first article confirmation rule number. The thresholds are obtained and frozen during the first article confirmation and cannot be temporarily changed; each judgment threshold is obtained by indexing the threshold configuration version number, and the threshold configuration version number is fixed within this task; the threshold corresponding to the threshold configuration version number is generated by the statistical distribution of the baseline during the stable production period of the target product family, with the detection resolution and inspection cycle constraints set, and the threshold generation rules and sample windows are fixed and traceable by the verification strategy version number; an example of a threshold is that the weight variation coefficient is not higher than 1.2%, the number of appearance defects in the confirmation window is 0, and the color deviation is not higher than 1.5; the critical dimension deviation threshold is given by the product family configuration and recorded in the configuration item corresponding to the threshold configuration version number. The first article confirmation and online verification use the same threshold configuration version number to ensure consistency of judgment criteria and review; The first confirmation is linked to the previous verification. The first confirmation sets verification inheritance conditions and solidifies them as judgment constraints. The verification inheritance conditions require that each record in the first confirmation cache be associated with its corresponding periodic curve package index, and the last verification record input segment index set be associated with the baseline periodic curve package index. The baseline periodic curve package index is stored in the baseline reference field of the first confirmation cache for the next alignment review. Inheritance consistency should be checked at the level of event marker sequence, ensuring at least the duration from injection to the end of holding pressure, the duration from mold opening to part removal, etc. The duration deviation threshold is obtained through the threshold configuration version number index, for example, not less than ±15% of the baseline, and the value is based on the statistical distribution of cycle time fluctuations in the stable production stage of the target product family and the equipment control resolution constraints. If any item fails to achieve inheritance consistency, it is considered a curve inheritance failure and enters the failure branch handling, to avoid the distortion of the first piece data due to cycle time fluctuations, abnormal operations, or accidental triggering after verification, which would affect the auditability of the switchover completion judgment. When continuity judgment fails, statistical judgment fails, or curve inheritance fails, the control terminal immediately stops the first piece confirmation and saves the first piece failure record. The first piece failure record is bound to the task number and written to the traceability database. The fields of the first piece failure record shall include at least the failure reason code, failure piece number, trigger threshold, measurement value, release threshold level, current resource consumption, associated cycle curve package index, suggested return path, and forced action flag. The failure reason code shall include at least the following: weight exceeding limit, appearance defect, size exceeding limit, color deviation, curve inheritance failure, detection missing, and manual input abnormality. It is recommended that the return path be output as a path number and can carry parameter increments. Parameter increments should include at least the boundary scan disturbance amplitude increment and the pipeline pulse count increment. The path number is bound to the verification strategy version number. Parameter increments are simultaneously constrained by an increment limit and a total amplitude limit. The constraint table is obtained from the verification strategy version number index. For example, the boundary scan speed disturbance amplitude increment should not exceed ±5% and the total disturbance amplitude should not exceed ±10%; the switching point displacement disturbance increment should not exceed ±0.5 mm and the total disturbance amplitude should not exceed ±1.5 mm; the holding time disturbance increment should not exceed ±5% and the total disturbance amplitude should not exceed ±10%; and the pipeline pulse count increment should not exceed 2 times per segment and the maximum increment per segment should not exceed 8 times. If the limits are exceeded or convergence is still not achieved, the process should proceed to manual review or branch rollback. It is recommended that the return path selection rule be determined by the failure... The failure reason code is determined together with the current risk level. If the weight or size exceeds the limit continuously, the combination path of pipeline reinforcement and boundary scanning is selected first, and the boundary scanning amplitude is increased. If the appearance deviation occurs alone and the weight and size meet the requirements, it enters restricted production and the detection frequency is increased. If the curve inheritance fails, the detection is missing, or the manual input is abnormal, it enters manual review first. The forced action mark uses an enumerated value, which includes at least enabling the conservative combination path, prohibiting entry into restricted production, forcing manual review, and forcing rollback to the previous parameter version. When two consecutive demolding abnormalities or appearance defects are clustered, the conservative combination path is forcibly enabled and entry into restricted production is prohibited. After the first piece fails, the control end switches the process status to the cleaning path or manual review according to the abnormal handling strategy, and writes the index key of the first piece failure record into the task process record to form an auditable failure branch evidence chain entry. Once the first piece is confirmed, the control unit generates a switchover completion record and writes it to the traceability database. The switchover completion record is a combination of a structured object and a task number. The fields include at least the task number, completion timestamp, total time, number of verification rounds, number of cleaning rounds, total number of scraps, final verification strategy version number, final threshold configuration version number, parameter version number, mold number, first piece confirmation rule number, key process curve index set, risk level, resource usage flag, and release threshold level. The completion timestamp is the moment the first piece is confirmed, and the total time is calculated from the switchover start timestamp to the completion timestamp. The key process curve index set includes at least the baseline periodic curve package index of the last verification and the index of all periodic curve packages in the first piece confirmation sequence, so that subsequent switchover completion determination and anomaly source review can be performed by curve package replay. After the switchover completion record is generated, a consistent update action is performed to update the parameter version validity identifier and the threshold configuration version number. The parameter version validity identifier uses a ternary key consisting of the parameter version number, device identifier, and mold identifier. The identifier value is either verified, requires verification, or is prohibited from automatic application. The scope of application of the ternary key is limited to automatic invocation of the same parameter version number under the condition of the same device identifier and the same mold identifier. The update rules are as follows: when the switchover is completed and the first piece confirmation is passed, the current ternary key identifier is updated to verified, and the verification timestamp, verification task number, and release threshold level are written. When manual input, detection downgrade, verification requirement conditions are triggered, or restricted release occurs during the first piece confirmation process, the current ternary key identifier is updated to require verification, and a verification reason code is written. When a prohibited release event is triggered, a forced rollback to the previous parameter version is executed, or manual verification confirmation fails, the current ternary key identifier is updated to prohibit automatic application. The prohibition reason code is written; the update rules for the threshold configuration version number distinguish between temporary threshold adjustments and configuration fixation adjustments within the task: threshold tightening based on resource approach markers is only written as a temporary adjustment in the verification record and does not generate a new threshold configuration version number; a new threshold configuration version number is generated and bound to the product family and device mold combination only when the fixation conditions of the abnormal policy configuration number index are met, and after manual or rule confirmation, and the effective range and effective start timestamp are written; the minimum condition for rule confirmation is that K consecutive product family switches are confirmed for the first time and no abnormal clustering occurs in the corresponding observation window, K is, for example, 3 times; when no fixation adjustment occurs, the threshold configuration version number remains unchanged, but a reference record is written. The reference record fields must at least include the old version number, reference timestamp, reference task number, and reference range to form a traceable reference chain and support reuse of similar switches; After recording and updating, the control terminal is updated to "completed," and production changes from trial production to formal piecework status, which is written to the shift report index. The control terminal then begins monitoring via an observation window. The length of the observation window is determined by the exception policy configuration number index, for example, the next 50 pieces. The observation window checks whether the triggering conditions of the exception policy configuration number index are triggered. The triggering conditions must include at least a threshold for the number of consecutive non-conforming pieces and a threshold for the percentage of non-conforming pieces, such as 3 consecutive non-conforming pieces or a non-conforming percentage exceeding 5% within 50 pieces. When the triggering conditions are met, the parameter version validity flag is downgraded to "requires review" and written to the downgrade record. The downgrade record must include at least a downgrade timestamp, triggering conditions, trigger statistics, related piece number range, suggested handling path, and is associated with the task number.

[0024] In this embodiment, when the integrated molding production line is in stable piece-rate production, the control terminal receives a switching task from the production planning terminal or the on-site human-machine interface. The task is queued according to fixed fields, and the current parameter version, mold status, and threshold configuration version number are frozen. A task identifier and a millisecond-level timestamp are generated. The process curve window, statistical feature window, and product-level window are initialized, and the verification rounds are set. Subsequently, residual discharge is performed according to a partitioning strategy. The mixing head and conveying pipeline are replaced according to the estimated volume and pulse rhythm. The cold runner and mold cavity form a periodic curve package according to the empty cycle beat. Pressure, flow rate, displacement, and temperature are collected simultaneously, and data integrity checks are performed. After discharge, online verification is performed, generating pipeline stability, injection curve convergence, and filling parameters based on the data window. Verification indicators such as consistency, demolding characteristics, and appearance consistency are filled in. Based on the freezing threshold, the satisfaction and confidence levels are formed, and the verification conclusion, mismatch type, risk level, and suggested cleaning path are output. When the verification fails or needs to be re-verified, the cleaning path is driven according to the verification record. The pipeline segment pulse discharge, cold runner temperature disturbance discharge, and mold cavity boundary scanning are executed in sequence. After short-cycle acquisition, the process returns to online verification until the verification passes or a rollback and manual review are triggered. When the verification passes and the risk meets the release conditions, the process enters the first piece confirmation sequence. In the trial production state, several pieces are tested and statistically judged according to the rules. A switchover completion record is generated, and the parameter version validity identifier and threshold configuration version number are updated consistently, so that the switchover process forms a traceable, replayable, and reusable closed-loop evidence chain.

[0025] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0026] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission; wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission includes infrared, microwave, etc. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center containing one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0027] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0028] In the embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0029] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0030] In addition, the functional modules in the embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0031] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0032] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0033] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for integrated production of multiple types of silicone products, characterized in that, include: S1: Receive the switching task, freeze the current parameter version and mold status, generate task identifier and timestamp, and initialize the data window and verification rounds; S2: Perform residual discharge, discharge the mixing head and pipeline according to the discharge volume and pulse rhythm, discharge the cold runner and mold cavity according to the empty circulation rhythm, and collect pressure, flow, displacement and temperature data. S3: Perform online verification, obtain verification conclusions and confidence levels based on the data window, and generate verification records; S4: If the verification fails, determine the cleaning path according to the verification record, and sequentially perform pipeline segment pulse discharge, cold runner temperature disturbance discharge and mold cavity boundary scanning, and then enter online verification; S5: After verification, execute the first confirmation sequence, record the switch completion information, and update the parameter version validity identifier and threshold configuration version number.

2. The integrated production method for molding multiple types of silicone products according to claim 1, characterized in that, Receive the switchover task, including: After receiving the task switch, generate and queue a structured task record and perform field validation. The task number is generated based on the date code, the shift incrementing serial number, and the check digit. It is written into the switch start allowed time window and the scrap counter bound to the task number is initialized. The verification strategy version number is locked and the threshold configuration version number reference field is fixed.

3. The integrated production method for molding multiple types of silicone products according to claim 1, characterized in that, Freeze the current parameter version and mold status, generate task identifier and timestamp, initialize data window and verification rounds, including: When freezing the current parameter version and mold status, a freeze snapshot is generated and written with the snapshot number, snapshot timestamp, equipment setting set and recent cycle feature summary. Based on the alarm events, part retrieval failure count, and pressure over-limit count in the observation window, the mold status is determined and the status reason code is written. A health check of the data acquisition link is performed and the conservative mode flag is recorded.

4. The integrated production method for molding multiple types of silicone products according to claim 1, characterized in that, Residual discharge is performed, discharging from the mixing head and piping according to the discharge volume and pulse rhythm, and from the cold runner and mold cavity according to the empty circulation rhythm, including: When discharging the mixing head and pipeline according to the discharge volume and pulse rhythm, a discharge configuration record is generated and locked. The discharge configuration record is written with the area boundary identifier, sampling frequency and degradation strategy parameter set. The discharge volume is determined based on pipeline geometric parameters and historical correction coefficients. Pulses are executed progressively according to the rhythm template. When the pressure sampling value reaches the product of the set pressure upper limit and the preset pressure close to the threshold, or when the pressure rise exceeds the pressure change rate threshold within the preset time window, a degradation record is written and additional actions are executed.

5. The integrated production method for molding multiple types of silicone products according to claim 1, characterized in that, Collect pressure, flow, displacement, and temperature data, including: When collecting pressure, flow, displacement and temperature data, periodic curve packages are generated based on event markers and written into the curve package index. An injection speed perturbation sequence is introduced, and a data integrity check is performed. When any periodic curve package is missing an injection start event marker or a pressure holding end event marker, or when the pressure sequence length or displacement sequence length is less than the preset minimum number of points, or when the time interval from injection start to pressure holding end is less than the preset lower limit or greater than the preset upper limit of the target period duration, an empty loop is added and the missing reason code and conservative mode flag are recorded, and the discharge stage end timestamp is written to the data window metadata.

6. The integrated production method for molding multiple types of silicone products according to claim 1, characterized in that, Perform online verification, obtain the verification conclusion and confidence level from the data window, and generate a verification record, including: When obtaining the verification conclusion and confidence level based on the data window, extract pipeline segments, injection curve segments and product-level segments based on the threshold configuration version number and verification strategy version number, and perform integrity gating. The satisfaction score of the mandatory indicators is determined according to the threshold rule, and the confidence score is obtained by weighting the scores according to the weights. Generate verification records and write them into the fragment index set, mismatch type, risk level, and recommended cleaning path number.

7. The integrated production method for molding multiple types of silicone products according to claim 1, characterized in that, If the verification fails, the cleaning path is determined according to the verification record, and the following steps are performed sequentially: segmented pulse discharge of the pipeline, temperature disturbance discharge of the cold runner, and mold cavity boundary scanning, including: If the verification fails, a path execution order is generated based on the verification record and task constraints and then entered into the database. The path execution order includes the path number, action template number, threshold index, exit rules, degradation rules, and rollback rules. Pressure thresholds are determined based on the upper limit of the equipment's rated pressure and the upper limit of the pressure set in the frozen snapshot. When the upper limit of the equipment's rated pressure is less than the upper limit of the pressure set in the frozen snapshot, the pressure threshold is taken as the upper limit of the equipment's rated pressure. When the upper limit of the rated pressure of the equipment is greater than or equal to the upper limit setting value of the pressure in the frozen snapshot, the pressure threshold shall be taken as the upper limit setting value of the pressure in the frozen snapshot. The remaining thresholds are obtained through the frozen version number index and written to the path execution form.

8. The integrated production method for molding multiple types of silicone products according to claim 1, characterized in that, Accessing online verification includes: When sequentially executing pipeline segment pulse discharge, cold runner temperature disturbance discharge, and mold cavity boundary scan, the segment-level execution single parameter of the pipeline segment and the disturbance table parameter of the mold cavity boundary scan are determined according to the mismatch type. Short-cycle acquisition is performed between adjacent actions to obtain pipeline segments and cycle curve packages, and the start and end timestamps of the pipeline segments and the list of cycle numbers are written into the segment index set; When the abort condition is triggered, a rollback record is generated and the process is returned to the online verification process. At the same time, a resource approach marker is recorded at the process entry point.

9. The integrated production method for molding multiple types of silicone products according to claim 1, characterized in that, After verification, the first-item confirmation sequence is executed, including: After verification, the first-piece confirmation sequence is executed, and the release decision is made based on the data integrity mark and the risk release upper limit threshold. The number of confirmed pieces, the frequency of testing, and the statistical window are determined according to the first piece confirmation rule number. The production mode is switched to trial production status, and the original values ​​of quality characteristics are collected for each piece and associated with the periodic curve package index. When a test item is missing, adjust the release threshold level according to the downgrade table mapping and record the manually entered mark; The decision is made based on continuity judgment, statistical judgment, and curve inheritance conditions. If the conditions are not met, the first failure record is recorded and the process proceeds to the cleaning path or review path.

10. The integrated production method for molding multiple types of silicone products according to claim 1, characterized in that, Record the switchover completion information, update the parameter version validity identifier and threshold configuration version number, including: When recording the switchover completion information and updating the parameter version validity identifier and threshold configuration version number, a switchover completion record is generated and written with the task number, verification round, cleaning round, scrap count, strategy version number, threshold version number, parameter version number, and key curve index set. Update the validity identifier and write the reason code according to the ternary key composed of parameter version number, equipment identifier and mold identifier; Tighten the temporary threshold triggered by resource approach and record it in the verification record. Adjust the threshold that meets the solidification conditions to generate or reference the threshold version number, and write it into the degradation record when the observation window is triggered.

Images