Method for temperature control optimization of a sealing element baking process facing an injection molding machine

Abstract

This disclosure provides a method for optimizing temperature control during the baking process of seals in injection molding machines, including: generating batch identifiers based on batch information output by the injection molding machine and establishing a baking process archive; setting target material state and consistency judgment thresholds; collecting oven cavity temperature data, seal surface temperature data, and oven cavity gas characteristic data to form a process data stream after cleaning and calculating a set of state features; inputting the set of state features into a material state estimation model to obtain material state indicators and state compliance, and calculating consistency deviation to generate a set of control feedback quantities; under the constraints of maximum allowable temperature, maximum heating rate, zone temperature difference, and anti-over-baking, using a wolf pack algorithm to optimize and solve the temperature control execution quantities and perform closed-loop iterative updates; when the state compliance meets the target material state, the consistency deviation does not exceed the consistency judgment threshold, and the anti-over-baking constraint is not triggered, executing termination control and archiving the temperature control strategy sequence.

Description

Technical Field

[0001] This disclosure relates to the field of industrial process temperature control technology, and in particular to a method for optimizing temperature control during the baking process of seals for injection molding machines. Background Technology

[0002] Seals used in injection molding machines typically require baking after molding to remove volatile components, release internal stress, and stabilize the material. Current production methods generally employ a "setting an oven temperature profile + timed heat preservation" approach for baking, primarily controlling the oven temperature or the temperature at a few measuring points.

[0003] However, in mass production scenarios, the baking process of seals is often simultaneously affected by factors such as carrier stacking and obstruction, differences in thermal inertia between zones, fluctuations in circulating air volume, changes in air intake and exhaust opening, and disturbances caused by opening doors to load and unload materials. This results in inconsistent heat exposure conditions at different locations within the same batch. The direct consequence is not the conventional problem of "temperature not reached," but rather the batch-specific phenomenon of "furnace temperature reaching the target but inconsistent material condition." Some seals have reached a usable assembly state, while others remain substandard, and some locations even pose a risk of excessive heat exposure, leading to rework, material mixing risks, and potential post-assembly failure. Therefore, this invention proposes a method for optimizing temperature control during the seal baking process in injection molding machines.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] This disclosure provides a method for optimizing temperature control during the baking process of seals for injection molding machines, in order to at least solve the above-mentioned technical problems existing in the prior art.

[0006] According to the present disclosure, a method for optimizing temperature control during the baking process of seals in injection molding machines is provided, comprising the following steps: S1: Generate batch identifiers based on batch information output by the injection molding machine, establish baking process files corresponding to the batch identifiers, write the target material status and consistency judgment thresholds, and generate the binding relationship between monitoring and execution objects; S2: Collect furnace cavity temperature data, sealing component surface temperature data and furnace cavity gas characteristic data according to the binding relationship, form a process data stream, and obtain the cleaning process data stream after time synchronization and anomaly rejection, calculate and file the status characteristic set; S3: Input the set of state features into the material state estimation model to obtain the material state index and state compliance, and calculate the consistency deviation based on the material state index at each location to generate a set of control feedback variables for filing. S4: Read the set of control feedback quantities to construct a temperature control optimization problem. Under the constraints of maximum allowable temperature, maximum heating rate, upper limit of temperature difference and anti-over-baking, use the wolf pack algorithm to solve the temperature control execution quantity, apply it to the baking device and write it into the temperature control strategy sequence, and return to S2 to S4 for closed-loop iterative update. S5: When the state meets the target material state and the consistency deviation does not exceed the consistency judgment threshold and the over-baking constraint is not triggered, output the baking completion command and terminate the update, and archive the temperature control strategy sequence and baking process file for traceability.

[0007] Compared with existing technologies, the temperature control optimization method for the baking process of seals in injection molding machines disclosed herein has the following beneficial effects: This invention uses a closed-loop feedback mechanism of material condition indicators, condition compliance, and consistency deviation to ensure that the baking control target directly corresponds to the "target material condition that can be assembled and used," and simultaneously constrains batch consistency, thus avoiding situations where the oven temperature meets the standard but the condition of the sealing components is inconsistent.

[0008] This invention introduces over-heating prevention constraints (combining suspected over-heating risk markers with gas release rate threshold penalties) into temperature control optimization, which can limit excessive heat exposure while ensuring compliance, reducing the probability of performance degradation and minimizing ineffective heat preservation time. Through event marker bits, abnormal channel markers, and degradation marker mechanisms, state estimation and strategy solving can still be completed even in the event of door opening thermal shock, sensor malfunctions, or partial data loss, improving the stability and feasibility of continuous production.

[0009] This invention unifies the archiving of post-cleaning process data streams, state feature sets, control feedback sets, and temperature control strategy sequences, enabling complete reproduction of the input-calculation-output chain for each cycle, facilitating quality traceability and process auditing. The temperature control execution quantities and statistical results of the trigger cycle for non-abnormal batches are written back for target setting and wolf pack algorithm initialization in subsequent batches of the same material type, accelerating convergence, reducing inter-batch fluctuations, and improving consistency compliance rates.

[0010] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0011] The above and other objects, features, and advantages of this disclosure will become readily apparent from the following detailed description of exemplary embodiments, taken in conjunction with the accompanying drawings. Several embodiments of this disclosure are illustrated in the drawings by way of example and not limitation, in which: In the accompanying drawings, the same or corresponding reference numerals indicate the same or corresponding parts.

[0012] Figure 1A schematic diagram illustrating the implementation flow of the temperature control optimization method for the baking process of seals in injection molding machines according to an embodiment of this disclosure is shown. Detailed Implementation

[0013] To make the objectives, features, and advantages of this disclosure more apparent and understandable, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without creative effort are within the scope of protection of this disclosure.

[0014] Example 1: As Figure 1 As shown, this embodiment provides a method for optimizing temperature control during the baking process of seals for injection molding machines, including the following steps: S1: Generate batch identifiers based on batch information output by the injection molding machine, establish baking process files corresponding to the batch identifiers, write the target material status and consistency judgment thresholds, and generate the binding relationship between monitoring and execution objects.

[0015] In one possible implementation, S1 specifically includes sub-steps S101 to S104: S101: Obtain the batch information corresponding to the seal to be baked from the injection molding machine and generate a unique batch identifier. Establish a baking process file that corresponds one-to-one with the batch identifier. The baking process file shall include at least the batch identifier, batch information, target material status, consistency judgment threshold, binding relationship table, and index field.

[0016] Specifically, after each batch of seals is injection molded, the injection molding machine outputs the batch information of that batch of seals to the baking control unit. The batch information includes at least: material type (e.g., material formula number M2025-01), injection completion time window (e.g., March 1, 2025, 10:05:00–10:12:00), quantity (e.g., 480 pieces), carrier number (e.g., carrier TR-03), and process number (used to point to a process record in the process database, e.g., process number BK-07).

[0017] After receiving batch information, the baking control unit generates a unique batch identifier according to the following rules: The equipment identifier, timestamp, batch number, and checksum are concatenated in a fixed order. The equipment identifier is the unique number of the injection molding machine or production line (e.g., IM-02), the timestamp is the start time of the injection molding completion time window (e.g., 20250301100500), the batch number is the incrementing number of the equipment for that day (e.g., batch 015, recorded as 015), and the checksum is calculated from the above three characters according to a preset hash rule, taking the last two digits (e.g., using modulo 97 and taking the remainder to get 23, recorded as 23). Therefore, the batch identifier "IM-02_20250301100500_015_23" is obtained.

[0018] The baking control unit writes the batch identifier into the batch identifier table and performs a uniqueness check: it searches the batch identifier table to see if there is a duplicate batch identifier. When a duplicate is found, the batch number is incremented by 1, the checksum is recalculated, and the batch identifier is generated again. This process continues until no duplicates are found, thus ensuring the uniqueness of the batch identifier in the batch identifier table.

[0019] After the batch identifier uniqueness verification is passed, the baking control unit establishes a baking process file corresponding one-to-one with the batch identifier. The baking process file is stored in a structured record format, containing at least the following data items, with an index key for each data item: batch identifier, batch information, target material state, consistency judgment threshold, binding relationship table, process data flow index, state feature set index, material state estimation result index, temperature control strategy sequence index, and termination judgment record. Specifically, the process data flow index is used to locate the furnace cavity temperature data, seal surface temperature data, and furnace cavity gas characteristic data collected in subsequent steps; the temperature control strategy sequence index is used to locate the strategy sequence formed over time by the temperature control execution quantities output in subsequent steps; and the termination judgment record is used to locate the baking completion judgment result and baking completion instruction output in subsequent steps.

[0020] S102: Determine the target material state and consistency judgment threshold based on the material type and process number, and write the target material state and consistency judgment threshold into the baking process file.

[0021] Specifically, the baking control unit reads the target material state and consistency judgment threshold that match the material type from the process database based on the material type and process number written in the baking process file in step S101.

[0022] The target material state is represented by a "target range of material state indices." These material state indices are numerical indicators output by the material state estimation model in subsequent steps, used to characterize the state evolution of the seal during the baking process. The target range consists of a lower limit and an upper limit. When the material state indices fall within this range, it is determined that the material has met the requirements for assembly and use.

[0023] To ensure that the target material state can be directly implemented, the process database stores a mapping table of "material type - target range". For example, for material type M2025-01, the process database gives a target range of [0.82, 0.90], indicating that the seal of this material type meets the target material state when the material state index reaches between 0.82 and 0.90 at the end of baking.

[0024] The consistency judgment threshold is used to constrain the degree of consistency of material condition of seals in the same batch. Specifically, the consistency judgment threshold is represented by the "maximum-minimum difference threshold", that is, the difference between the maximum and minimum values ​​of the material condition index of different locations or different parts within the same batch is used as the consistency deviation. When the consistency deviation does not exceed the consistency judgment threshold, the batch is judged to be consistent.

[0025] To ensure that the consistency judgment threshold can be directly applied, a mapping table of "material type - consistency judgment threshold" is stored in the process database. For example, for material type M2025-01, the consistency judgment threshold is set to 0.05, which means that the maximum and minimum difference of the material state index at any location within the batch does not exceed 0.05 to meet the batch consistency requirement.

[0026] When the process database does not contain a target range or consistency threshold corresponding to the material type, the baking control unit executes the default generation rule: It filters qualified batches of the same material type from the baking process files of historical qualified batches, extracts the final value set of material state indicators for each qualified batch, and uses the 5th percentile of the final value set as the lower limit of the target range and the 95th percentile as the upper limit. Simultaneously, it extracts the final value set of consistency deviations for each qualified batch and uses the 95th percentile of the final value set of consistency deviations as the consistency threshold. The generated target range and consistency threshold are then written back to the process database and recorded in the baking process file for this batch.

[0027] For example, if the set of final values ​​for the material condition indicators of qualified batches of the same material type is {0.83, 0.84, 0.85, 0.86, 0.88, 0.89}, then the target range can be [0.83, 0.89]. If the set of final values ​​for consistency deviation is {0.03, 0.04, 0.04, 0.05, 0.05, 0.06}, then the consistency judgment threshold can be 0.06.

[0028] The baking control unit writes the target material state and consistency judgment threshold into the baking process file established in step S101, and creates an index key for it so that it can be directly called as the control target and judgment basis when constructing subsequent temperature control optimization problems.

[0029] S103: Generate a binding relationship table for monitoring and execution objects based on the zoning configuration of the baking device and write it into the baking process file. Then, perform verification, abnormal channel marking, and degradation marking on the binding relationship table and file them. Specifically, the baking control unit generates a binding relationship table based on the carrier number and the partition configuration of the baking device in step S101. The binding relationship table is used to establish a one-to-one correspondence between "batch identifier - sensor channel - spatial location - sampling parameters" and serves as the input source for subsequent steps such as synchronous preprocessing, material state estimation and temperature control strategy solution.

[0030] The binding relationship table must contain at least the following fields: batch identifier, sensor number, sensor type, partition number or installation location identifier, data channel number, sampling period, unit identifier, and measurement object identifier. The sensor type must include at least furnace cavity temperature, seal surface temperature, and furnace cavity gas characteristics. The measurement object identifier indicates the corresponding "furnace cavity location" or "carrier location," with the carrier location encoded using "layer number-column number-row number."

[0031] For example, if carrier TR-03 is a three-layer carrier with 10 columns and 16 rows per layer, the surface temperature measurement position can be coded as L2-C05-R08, representing the 5th column and 8th row of the 2nd layer. The baking control unit binds the infrared temperature measurement channel IR-01 to L2-C05-R08, binds the oven cavity temperature sensor T-03 to zone Z2, and binds the gas characteristic sensor G-01 to the return air duct position RF1. The above binding relationships are written into the binding relationship table in the form of record rows and then into the baking process file.

[0032] After binding is complete, the baking control unit performs binding verification to confirm the consistency between the data channel and the physical object: during the heating phase, oven cavity temperature data and seal surface temperature data are collected, determining that the oven cavity temperature shows an upward trend over multiple consecutive sampling periods, and that the seal surface temperature lags behind the oven cavity temperature and remains within a preset difference range. Simultaneously, oven cavity gas characteristic data is collected, determining that the gas characteristics show a response change consistent with the temperature change direction during the heating phase. If any determination fails, the corresponding data channel is marked as an abnormal channel and written into the baking process file, so that subsequent steps can remove, downgrade, or replace the abnormal channel.

[0033] To ensure feasibility on-site, when sensors are missing or there are abnormal channels, the baking control unit generates a degradation flag and writes it into the baking process file: When the surface temperature data of the seal is unavailable, the degradation flag indicates that the surface temperature characteristics should be estimated in subsequent steps using "furnace cavity temperature data + carrier position code". When furnace cavity gas characteristic data is unavailable, the degradation flag indicates that the gas characteristic change correlation characteristics should be replaced with "temperature change rate correlation characteristics and insulation stability correlation characteristics" in subsequent steps. The degradation flag also instructs the material state estimation model to select the corresponding model version to match the available input in subsequent steps.

[0034] The baking control unit writes the binding relationship table, abnormal channel markers, and degradation markers into the baking process file and establishes an index key, so that subsequent steps can directly read the binding relationship table to determine the acquisition object and channel mapping, perform data removal or downweighting based on the abnormal channel markers, and select the material state estimation model version and feature construction method based on the degradation markers, thereby ensuring that the output of steps S101 to S103 is completely consumed in subsequent steps and forms a closed loop.

[0035] In this embodiment of the invention, the post-cleaning process data stream, state feature set, control feedback quantity set, and temperature control strategy sequence are uniformly archived, which can completely reproduce the input-calculation-output link of each cycle, facilitating quality traceability and process auditing.

[0036] S2: Collect furnace cavity temperature data, seal surface temperature data and furnace cavity gas characteristic data according to the binding relationship, form a process data stream, and obtain the cleaning process data stream after time synchronization and anomaly rejection, calculate and file the status characteristic set.

[0037] In one possible implementation, S2 specifically includes sub-steps S201 and S203: S201: Acquire furnace cavity temperature data, seal surface temperature data, and furnace cavity gas characteristic data according to the binding relationship and adaptive sampling period, generate a raw process data stream containing timestamps and event flags, and write it to the baking process archive. Specifically, the baking control unit initiates data acquisition under the constraints of the binding relationship table in step S103, and generates a raw process data stream arranged in chronological order for this batch identifier.

[0038] The raw process data stream consists of several records, each containing fixed fields: timestamp, furnace temperature data vector, seal surface temperature data vector, furnace gas characteristic data vector, event flag, and channel validity flag. The timestamp is in seconds, the event flag indicates whether the timestamp falls within the door-opening thermal shock phase, and the channel validity flag indicates whether each data channel in the record is available.

[0039] To ensure controllable acquisition resolution and data volume, an adaptive sampling period is used for acquisition. Among these, The sampling period is The timestamp interval (in seconds) between two adjacent records, and The value ranges from 0.5 seconds to 5 seconds.

[0040] The baking control unit calculates the rate of temperature change in the oven cavity in real time during the heating phase and sets accordingly. When the rate of temperature change in the furnace cavity exceeds 2 degrees Celsius per minute, Set to 1 second. When the furnace temperature change rate is less than 0.5 degrees Celsius per minute, [the following will occur]. Set to 3 seconds. Otherwise, Set to 2 seconds.

[0041] The furnace cavity temperature data vector is composed of temperature values ​​from multiple measuring points marked as furnace cavity temperature type in the binding relationship table at the same time stamp, arranged in order of measuring point number. The temperature unit is degrees Celsius. The seal surface temperature data vector is composed of temperature values ​​from multiple temperature measuring locations marked as surface temperature type in the binding relationship table at the same time stamp, arranged in order of location code. The temperature unit is degrees Celsius. The furnace cavity gas characteristic data vector is composed of values ​​from multiple measuring points marked as gas characteristic type in the binding relationship table at the same time stamp, arranged in order of measuring point number. The gas characteristic includes at least one or more of humidity, dew point, or volatile component concentration, and their units are recorded in the binding relationship table.

[0042] The event flag is generated by the door sensor or door lock switch signal: when the door sensor signal changes from off to on, the event flag is set to 1 and remains so until the door sensor signal returns to off. When the door sensor signal is off, the event flag is set to 0.

[0043] For example, if the oven cavity temperature is detected to rise from 120 degrees Celsius to 123 degrees Celsius within 60 seconds during the heating phase, the temperature change rate is 3 degrees Celsius per minute, and the baking control unit will... Set to 1 second and generate one record per second. If the oven cavity temperature fluctuates by only 0.6 degrees Celsius within 300 seconds during the heat preservation phase, the temperature change rate is less than 0.5 degrees Celsius per minute, and the baking control unit will... Set the interval to 3 seconds and generate a record every 3 seconds.

[0044] The baking control unit writes the storage location of the original process data stream into the process data stream index field of the baking process file, so that step S202 can read the data stream by index as the only input.

[0045] S202: Perform time synchronization and interpolation resampling on the original process data stream, execute physical rationality verification and jump detection to form a synchronized and cleaned process data stream, and write abnormal channel markers and thermal shock section annotations into the baking process file: Specifically, the baking control unit reads the original process data stream from the process data stream index written in step S201 and converts it into a synchronously cleaned process data stream on a unified time axis. The synchronously cleaned process data stream serves as the sole data source for feature extraction in step S203.

[0046] The baking control unit constructs a unified timeline based on the timestamps in the original process data stream and sets the maximum allowable time error between channels to 0.1 seconds. When the sampling time of any channel deviates from the unified timeline by no more than 0.1 seconds, the data of that channel is directly mapped to the corresponding timestamp. When the deviation exceeds 0.1 seconds, linear interpolation resampling is performed on that channel to generate data on the unified timeline.

[0047] To make interpolation directly implementable, linear interpolation is performed according to the following rules: if a channel is at a timestamp With timestamp There is valid data at this point, and there are data within the unified timeline that fall between... and target timestamps between Then the target timestamp The data at the location is taken as Data + ( Data processing (Data from the data source) × ( ) / ( ).in, To unify the target timestamp (seconds) on the timeline. and The timestamp (in seconds) of the two adjacent valid original records of this channel.

[0048] After synchronization is complete, the baking control unit performs a physical validity check on each record and updates the channel validity flag. The physical validity check includes the following executable decisions: If the temperature at any measuring point in the furnace cavity temperature data vector is less than 0 degrees Celsius or greater than 300 degrees Celsius, the data at that measuring point is deemed invalid and the corresponding channel is marked as invalid.

[0049] If any temperature in the surface temperature data vector of the seal is less than 0 degrees Celsius or greater than 300 degrees Celsius, the data at that location is determined to be invalid and the corresponding channel is marked as invalid.

[0050] To constrain the relative relationship between surface temperature and furnace cavity temperature, a difference check is performed on the surface temperature and furnace cavity temperature at the same timestamp: when the surface temperature is greater than "maximum furnace cavity temperature + 10 degrees Celsius", the surface temperature data is deemed invalid, and the corresponding channel is marked as invalid. The maximum furnace cavity temperature is taken as the largest element in the furnace cavity temperature data vector.

[0051] If any value in the furnace cavity gas characteristic data vector is negative, the gas channel is determined to be invalid and the corresponding channel's valid mark position is set to invalid.

[0052] To eliminate obvious measurement jumps, the baking control unit performs jump detection on the oven cavity temperature and surface temperature: it calculates the difference between two adjacent temperature records in the same channel, and if the absolute value of the difference is greater than 15 degrees Celsius per second, it is determined to be an abnormal jump, and the channel's valid mark at that timestamp is marked as invalid. When three consecutive abnormal jumps occur, the channel is marked as an abnormal channel, and the abnormal channel mark is written into the baking process file so that subsequent steps can remove or demote the abnormal channel.

[0053] The event segment annotation strictly uses the event flag bit from step S201: when the event flag bit is 1, the record at that timestamp is marked as a thermal shock segment record. Thermal shock segment records are not included in the statistics when calculating the "thermal insulation stability related characteristics" in step S203, but are used as disturbance event inputs in the constraint penalty calculation when constructing the temperature control optimization problem in step S401, thus ensuring that the output of the event annotation is consumed subsequently and not becomes useless output.

[0054] The baking control unit writes the synchronized cleaning process data stream to the storage medium and writes its storage location to the "cleaning data stream index" subfield in the process data stream index field of the baking process file, so that step S203 only needs to read the index to obtain the unique input data stream.

[0055] S203: Based on a preset time window, calculate the heating rate-related characteristics, heat preservation stability-related characteristics, temperature uniformity-related characteristics, and gas characteristic change-related characteristics from the process data stream after synchronous cleaning, normalize them, form a set of state characteristics, and write them into the baking process file: Specifically, the baking control unit reads the process data stream after synchronous cleaning from the cleaning data stream index written in step S202, and calculates the state feature set for material state estimation without using invalid channel data. Then, the state feature set is written into the baking process file and a state feature set index is established so that step S301 can directly read it as input according to the index.

[0056] The set of state features is calculated using time windows. The Chinese definition of a time window is "a continuous recording interval used to calculate the same set of features." In this embodiment, the heating window length is set to 30 seconds, and the holding window length is set to 300 seconds. The heating window is used to calculate features related to the heating rate, and the holding window is used to calculate features related to holding stability and temperature uniformity.

[0057] The heating rate characteristic is calculated as follows: Within any heating window, the difference between the average furnace cavity temperature at the start and end of the window is taken, and then divided by the duration of the window to obtain the heating rate. The average furnace cavity temperature is the arithmetic mean of all valid measurement points in the furnace cavity temperature data vector. If the number of valid measurement points within the window is less than 70% of the total number of furnace cavity measurement points in the binding relationship table, the heating rate characteristic for that window is marked as invalid, and an invalidity reason flag is attached when the file is entered.

[0058] The thermal insulation stability characteristics are calculated as follows: Within any thermal insulation window, the standard deviation of the average furnace cavity temperature sequence is taken as the stability index. The standard deviation calculation only uses records with an event flag of 0 and only valid measurement points. If the proportion of records with an event flag of 1 within the thermal insulation window exceeds 20%, the stability index for that window is marked as invalid and an invalidity reason is attached when it is entered into the file.

[0059] Temperature uniformity-related characteristics are calculated as follows: Within any insulation window, the "maximum effective temperature value" of the furnace cavity temperature data vector is calculated for each record. The uniformity index is obtained by averaging the difference series of minimum effective temperatures. A smaller uniformity index indicates a more uniform temperature distribution. If the number of effective measurement points for a record is less than 70% of the total number of measurement points, that record is not included in the uniformity index statistics.

[0060] The gas characteristic change-related features are calculated in the following executable manner: For each dimension of the gas feature in the furnace cavity gas feature data vector, calculate the difference between two adjacent records and divide by the corresponding value. The rate of change is obtained, and the average rate of change is taken within a 30-second heating window as the release rate index of the gas feature. When the gas feature channel is marked as invalid in step S202, the gas feature change-related features are set to empty and a missing flag is attached when the file is entered, so that step S301 can select the corresponding model version according to the downgrade flag.

[0061] To ensure that features with different dimensions can be used by the same material state estimation model, the baking control unit performs 0-to-1 normalization on each feature. Normalization maps feature values ​​to the range of 0 to 1, and is performed as follows: For a given feature, the minimum value of that feature in historical qualified batches for that material type is taken as the lower bound, and the maximum value as the upper bound. The current feature value is then normalized according to the formula "(current value)". (Lower Boundary) / (Upper Boundary) The normalized result is obtained by calculating the lower bound. When the upper bound equals the lower bound, the normalized result is set to 0.5 and the mark indicating that the upper bound equals the lower bound is recorded.

[0062] For example, if the minimum temperature rise rate characteristic of a certain material type in historical qualified batches is 0.6 degrees Celsius per minute and the maximum is 3.0 degrees Celsius per minute, and the current window calculates a temperature rise rate of 1.8 degrees Celsius per minute, then the normalized result is (1.8...). 0.6) / (3.0) 0.6) = 0.5. If the average difference between the maximum and minimum temperatures in a certain window is 6 degrees Celsius, the historical lower bound is 3 degrees Celsius, and the upper bound is 9 degrees Celsius, then the uniformity normalization result is (6) 3) / (9) 3) = 0.5.

[0063] The state feature set is archived in a key-value pair structure. Each feature record includes at least: time window number, window start and end timestamps, normalized heating rate result, normalized insulation stability result, normalized temperature uniformity result, normalized gas release rate result, event percentage, effective channel percentage, and invalid reason marker. The event percentage and effective channel percentage are used in step S301 to select the model version and adjust the material state estimation confidence level, and in step S401 to assign weights to the constraint penalty terms in the temperature control optimization problem, thereby ensuring that this output is actually used in subsequent steps.

[0064] The baking control unit writes the storage location of the state feature set into the state feature set index field of the baking process file, serving as the direct input source for material state estimation in step S301, and simultaneously records the state feature set used in this batch during the writing process. Adaptive rule parameters and anomaly removal threshold parameters are used to reproduce the feature calculation process during subsequent tracing.

[0065] In this embodiment of the invention, by using event flags, abnormal channel flags, and degradation flags, state estimation and strategy solving can still be completed even when there is thermal shock from door opening, sensor malfunction, or partial data loss, thereby improving the stability and feasibility of continuous production.

[0066] S3: Input the set of state features into the material state estimation model to obtain the material state index and state compliance, and calculate the consistency deviation based on the material state index at each location to generate a set of control feedback variables for filing.

[0067] In one possible implementation, S3 specifically includes sub-steps S301 to S304: S301: Read the state feature set, degradation markers, and abnormal channel markers from the baking process archive; select the material state estimation model version according to the degradation marker; input the state feature set into the material state estimation model to obtain material state indices; and perform mechanistic constraint trimming on them. Specifically, the baking control unit reads the set of state features written in step S203 from the baking process file, and reads the degradation flag and abnormal channel flag written in step S103, in order to determine the input dimension and model version required for material state estimation.

[0068] In this embodiment, the material condition index is defined as a normalized state quantity with a value ranging from 0 to 1. A larger material condition index value indicates that the seal is closer to the target material state for assembly and use, where 0 represents a substandard state and 1 represents a fully compliant state. The material condition index is calculated solely from the set of state features and does not incorporate any external inputs not recorded in the baking process file.

[0069] The baking control unit selects the material state estimation model version according to the following rules: When the state feature set contains a missing gas release rate marker, or a downgrade marker indicating "gas feature unavailable", select the model version that does not include the gas release rate input dimension, and denote it as model version. .

[0070] When there is no missing gas release rate marker in the state feature set and the degradation marker does not indicate "gas feature unavailable", select the model version that includes the gas release rate input dimension, and denote it as the model version. .

[0071] When the proportion of effective channels is less than 0.70, the material state estimation result of this window is marked as low confidence, and the confidence mark is written into the subsequent feedback quantity set for S401 penalty weighting and S501 decision constraint.

[0072] Specifically, the material state estimation model adopts a fusion structure of "data-driven mapping + mechanism constraint trimming": data-driven mapping is used to calculate the initial values ​​of material state indices from the set of state features. Mechanism constraint trimming is used to restrict the initial values ​​to between 0 and 1 and to penalize suspected over-baking or thermal shock disturbances to avoid state jumps in the output that violate process common sense.

[0073] To enable direct implementation of online inference, the data-driven mapping is implemented using a linear weighted summation method. The baking control unit reads the following normalized features from the state feature set for each time window: normalized heating rate, normalized heat preservation stability, normalized temperature uniformity, normalized gas release rate (if available), event proportion, and effective channel proportion, and constructs a feature vector.

[0074] In model version In this process, the initial values ​​of the material condition indicators are calculated as follows: The "normalized result of heating rate" is denoted as The "normalized result of thermal insulation stability" is denoted as The "temperature uniformity normalization result" is denoted as The "normalized result of gas release rate" is denoted as .in, , , The larger the value, the more conducive the process is to achieving the target material state. A larger value indicates a higher degree of temperature non-uniformity and is detrimental to achieving the target material state. Therefore, the initial value for the material state index is set as follows: .

[0075] In model version Since the gas release rate is unavailable, the initial values ​​for the material condition parameters are taken as follows: .

[0076] The aforementioned weighting coefficients are obtained from historical qualified batch archives: The baking control unit stores a mapping table of "material type - weighting coefficient" in the process database. During calibration, historical qualified batches of the same material type are used to make the target value of the material state index at the end of the qualified batch window close to 1. The weighting coefficients are then fitted using the least squares method and written into the process database. When the process database lacks a weighting coefficient for that material type, the default coefficient is used and this estimate is marked as "default coefficient used" for subsequent updates.

[0077] Mechanism-constrained clipping is performed according to the following rules: When the initial value is less than 0, the initial value of the material condition index is clipped to 0. When the initial value is greater than 1, the initial value of the material condition index is clipped to 1. The material condition index of the window is obtained after clipping.

[0078] When the event percentage is greater than 0.20, multiply the material condition index of this window by (1). The event percentage is used as a disturbance penalty to reflect the uncertainty in state estimation caused by the thermal shock phase of the door opening.

[0079] When the percentage of effective channels is less than 0.70, the material state index of the window is multiplied by the percentage of effective channels as a confidence penalty, and a low confidence mark is written into the feedback set.

[0080] The baking control unit calculates the degree of compliance based on the target material state target range written into the baking process file in step S102. The target range includes a lower limit value. With upper limit ,and and All values ​​are between 0 and 1.

[0081] The status compliance rate is calculated according to the following executable rules: When the material condition index is at When the interval is within the specified range, the state attainment rate is set to 1.

[0082] When the material condition index is less than At that time, the degree of compliance with the material condition is taken as the material condition index divided by . .

[0083] When the material condition index is greater than At that time, the compliance rate is set to 1, and the "suspected over-drying risk flag" is set to 1 for subsequent over-drying constraint penalties in S401. The purpose of the suspected over-drying risk flag is to prevent the controller from pursuing compliance by infinitely increasing the material condition index while ignoring the risk of over-drying.

[0084] The baking control unit writes the material state indicators, state compliance, suspected over-baking risk markers, and the corresponding event proportions and effective channel proportions for each time window into the material state estimation result index field of the baking process file, which is used for subsequent consistency deviation calculation and feedback quantity encapsulation.

[0085] For example, if the target interval is [0.82, 0.90], the features of a certain window are: , , , Model version adopted The initial value is 0.535. Since the initial value is between 0 and 1, no trimming is needed, and the material condition index is 0.535. Because the material condition index is less than 0.82, the condition compliance rate is approximately 0.535 / 0.82 ≈ 0.652.

[0086] S302: Calculate the compliance rate of the target material state based on the target range, and generate a suspected over-baking risk marker, which is then written into the baking process file. Determine the consistency sample set according to the binding relationship table. If the preset minimum number of valid samples is met, calculate the consistency deviation based on the difference between the maximum and minimum values ​​of the material state indicators in the consistency sample set, and write it into the baking process file; otherwise, generate an insufficient sample marker. Specifically, the baking control unit determines the set of consistency samples for consistency assessment based on the binding relationship table written into the baking process file in step S103. The set of consistency samples consists of material state indicators corresponding to multiple surface temperature measurement positions, and each surface temperature measurement position corresponds one-to-one with the carrier position code.

[0087] To ensure the reproducibility of the consistency assessment, this embodiment selects at least four surface temperature measurement locations on each layer of the vehicle as a fixed set of locations for the consistency sample set: the four corners of the left front, right front, left rear, and right rear of each layer. When the vehicle has three layers, the total number of samples in the consistency sample set is 12.

[0088] The baking control unit extracts the material state index corresponding to the above fixed position set as sample value at the end of the same time window, and judges the validity of the sample: when the effective channel ratio of a certain position in the window is less than 0.70 or the position is associated with an abnormal channel, the sample at that position is marked as invalid and does not participate in the consistency deviation calculation.

[0089] To avoid instability caused by insufficient samples, the baking control unit sets the minimum number of valid samples for consistency deviation calculation to 4. When the number of valid samples is less than 4, the consistency deviation is marked as invalid and an "insufficient sample flag" is written to the baking process file so that S401 can apply a penalty to this window and continue iterative collection.

[0090] When the number of valid samples is not less than 4, the baking control unit calculates the consistency deviation as follows: The maximum and minimum values ​​of the material condition index are taken from the valid samples at the end of the window; the difference between the two is the consistency deviation, denoted as [missing value]. ,and The value range is from 0 to 1.

[0091] The baking control unit will monitor the consistency deviation of this window. The number of valid samples, insufficient sample markers, and invalid sample location list are written into the material state estimation result index field of the baking process file, serving as a direct input source for subsequent S401 temperature control optimization targets and S501 compliance determination.

[0092] For example, if there are 10 valid samples out of the last 12 positions in a window, and the set of valid samples for the material condition index is {0.84, 0.83, 0.85, 0.82, 0.86, 0.84, 0.83, 0.85, 0.82, 0.84}, then the maximum value is 0.86 and the minimum value is 0.82, indicating a consistency deviation. If step S102 gives a consistency threshold of 0.05, then the window meets the intra-batch consistency requirement.

[0093] S303: Encapsulate material condition indicators, condition compliance, consistency deviation, suspected over-baking risk markers, and insufficient sample markers into a set of control feedback quantities and write them into the baking process file: Specifically, the baking control unit encapsulates the material state indicators and state compliance obtained in step S301, the consistency deviation obtained in step S302, and their effective sample information into a set of control feedback quantities, and writes them into the baking process file for direct reading by subsequent steps.

[0094] The control feedback set adopts a fixed field structure, with each time window corresponding to one feedback record. The feedback record includes at least: batch identifier, window number, window start and end timestamps, material status index, status compliance, consistency deviation, number of valid samples, event percentage, percentage of valid channels, suspected over-drying risk marker, insufficient sample marker, and default coefficient usage marker.

[0095] Among them, material state indicators, state compliance, and consistency deviation serve as optimization target inputs when constructing the temperature control optimization problem in subsequent S401. Event proportion and effective channel proportion are used to assign weights to constraint penalty terms in subsequent S401. Suspected over-baking risk markers are used for anti-over-baking constraint penalties in subsequent S401. State compliance and consistency deviation are used for baking completion determination in subsequent S501. Insufficient sample markers are used by subsequent S401 to suppress premature termination and drive continued iterative acquisition when there are insufficient consistent samples.

[0096] The baking control unit writes the storage location of the control feedback set into the "control feedback set index" field of the baking process file, and uses this index as the only entry point for subsequent steps to read the feedback set, thereby ensuring that the output of steps S301 to S303 is completely consumed and forms a closed loop in subsequent steps.

[0097] In this embodiment of the invention, through closed-loop feedback of material state indicators, state compliance and consistency deviation, the baking control target directly corresponds to the "target material state that can be assembled and used", and simultaneously constrains the consistency within the batch, so as to avoid the situation where the furnace temperature meets the standard but the state of the sealing parts is discrete.

[0098] S4: Read the set of control feedback quantities to construct a temperature control optimization problem. Under the constraints of maximum allowable temperature, maximum heating rate, upper limit of temperature difference and anti-over-baking, use the wolf pack algorithm to solve the temperature control execution quantity, apply it to the baking device and write it into the temperature control strategy sequence, and return to S2 to S4 for closed-loop iterative update.

[0099] In one possible implementation, S4 specifically includes sub-steps S401 to S403: S401: Read the set of control feedback quantities, the target material state target range, and the consistency judgment threshold in control cycle units, and construct a temperature control optimization problem that includes state error terms, consistency error terms, energy consumption terms, and constraint penalty terms. The constraint penalty terms include at least the maximum allowable temperature penalty, the maximum heating rate penalty, the zone temperature difference penalty, and the anti-over-drying penalty. Specifically, the baking control unit performs closed-loop optimization control with a fixed control cycle. In this embodiment, 10 seconds are used. The time interval (in seconds) between two consecutive temperature control updates.

[0100] At the start of each control cycle, the baking control unit reads the set of control feedback quantities corresponding to the previous control cycle from the baking process file, and reads the target material state target range and consistency judgment threshold. The target range includes a lower limit. and upper limit ,and , All values ​​are between 0 and 1, and the consistency threshold is between 0 and 1.

[0101] The baking control unit in the first The temperature control optimization problem is constructed and solved in the first control cycle. Temperature control execution vector for each control cycle . In the first The set of control quantities issued to the actuator of the baking device within a control cycle.

[0102] Specifically, the baking apparatus is divided into: Each heating zone, and . Includes the following components and is expressed as a percentage or in engineering units: , , : Heating power of the three heating zones, expressed as a percentage.

[0103] : Circulating air volume, in percentage.

[0104] : Air intake and exhaust opening, in percentage.

[0105] To ensure feasibility, the baking control unit sets boundary ranges for each component: The value ranges from 0% to 100%. The value ranges from 20% to 100%. The value ranges from 0% to 100%.

[0106] The baking control unit constructs a computable optimization objective function in the kth control cycle. , This is used to evaluate the performance of candidate temperature control actuators. The formula is obtained by weighted summation of the state error term, consistency error term, energy consumption term, and constraint penalty term, as follows:

[0107] in, , , , These are the weighting coefficients, which are read from the process database according to material type. This is the state error term. For consistency error terms, For energy consumption items, This is a constraint penalty term. In this embodiment, when the process database is missing, [the following is used]: , , , And write the "default weight use tag" to the baking process file.

[0108] Calculated according to the following rules: The baking control unit reads the material state indicators from the control feedback set. . This refers to the material state index at the end of the window corresponding to this cycle. Less than the lower limit of the target interval hour, Take as .when Greater than or equal to hour, Set to 0 to ensure that the state error term only penalizes the part that does not meet the standard.

[0109] Calculate the consistency deviation in the control feedback set read by the baking control unit according to the following rules: . This represents the maximum and minimum difference in the material state index at the end of the corresponding window for that period. When the sample size is insufficient, it is marked as 1. The value is set to 1 to enforce a penalty and push the system to continue collecting samples until the minimum number of valid samples is met. When the number of samples is insufficient, it is marked as 0. Take as - Consistency determination threshold), to only penalize the portion exceeding the threshold.

[0110] Calculate the candidate temperature control execution vector according to the following rules: The approximate total power is obtained by averaging the power of the three partitions. and order This normalizes the energy consumption item to the range of 0 to 1.

[0111] It consists of four types of constraint penalties: maximum allowable temperature penalty, maximum heating rate penalty, zone temperature difference penalty, and over-drying prevention penalty. Each penalty is calculated based on the set of control feedback variables and the post-cleaning process data stream index.

[0112] To make the penalty calculation possible, the baking control unit reads the post-cleaning process data stream index from the baking process archive and extracts the most recent... Data from temperature measurement points at each furnace cavity within the time range were collected and calculated. Maximum permissible temperature penalty: When the maximum value of any furnace cavity temperature measuring point within this range exceeds [a certain threshold], the penalty is applied when [the temperature exceeds a certain threshold]. When, the penalty value is taken as Otherwise, the penalty value is 0. The maximum permissible temperature (degrees Celsius) is used in this embodiment. The temperature is recorded in degrees Celsius and written into the process database.

[0113] Maximum heating rate penalty: Calculates the rate of change of the average furnace temperature within this range. ,when Greater than The time penalty value is taken as Otherwise, take 0. To determine the maximum permissible heating rate (degrees Celsius per minute), this embodiment takes... Celsius per minute.

[0114] Zoned Temperature Difference Penalty: The average temperature of each zone is obtained from the temperature measuring points in the polymerization furnace cavity according to the zone number. The difference between the maximum and minimum average temperatures of the three zones is recorded as . ,when Greater than The time penalty value is taken as ( Otherwise, take 0. To determine the upper limit of the permissible temperature difference between zones (in degrees Celsius), this embodiment takes... Celsius.

[0115] Over-drying penalty: When the suspected over-drying risk flag in the control feedback set is 1, the penalty value is 1. When the suspected over-drying risk flag is 0, the normalized gas release rate result in the state feature set is further read. ,like If the value is greater than 0.85, the penalty value is set to... Otherwise, the penalty value is set to 0. 0.85 is the excessively fast gas release threshold, which is written into the process database and can be configured according to material type.

[0116] Baking control unit command This is equal to the sum of the four types of penalty values ​​mentioned above, thus allowing candidate executions that violate any constraint to incur a higher cost in the objective function.

[0117] For example, if a certain cycle reads , , If the consistency threshold is 0.05, then... =0.04, =0.01. If candidate make , , ,but , If recently The highest internal temperature is 225 degrees Celsius and The maximum temperature penalty is (225 degrees Celsius). 220) / 220≈0.0227, and the remaining penalties are 0. If the default weights are used , , , ,but .

[0118] S402: Encode the candidate temperature control execution vectors as candidate individuals in the wolf pack algorithm, initialize them according to the preset wolf pack size and step size, calculate the fitness based on the temperature control optimization problem, iterate the search through scouting updates, summoning updates, and siege updates, and output the optimal candidate temperature control execution quantity when the stopping condition is met: Specifically, the baking control unit uses a wolf pack algorithm to iteratively solve the temperature control optimization problem in each control cycle, outputting a value that satisfies the boundary range and makes... Minimum temperature control execution vector .

[0119] To make the wolf pack algorithm feasible, the baking control unit defines a "wolf" as a candidate temperature control execution vector. The candidate vector has a dimension of 5, corresponding to... , and Furthermore, each dimension is constrained by the boundary range given in step S401.

[0120] Baking control unit sets wolf pack size Set the maximum number of iterations to 20. It is 30. Among them, This represents the number of candidate solutions in each solution iteration. The maximum number of cycles for updating the wolf pack.

[0121] The baking control unit initializes the wolf pack in the kth control cycle: using the execution amount from the previous cycle that is currently being executed. As the central vector, random perturbations are superimposed on each dimension to generate... Sekiro. The perturbation amplitude uses component step size. , This represents the maximum amount of change allowed for a candidate solution in a single update. Specifically, The step size is 10%. The step size is 10%. The step size is set to 15%.

[0122] During initialization, boundary truncation is performed on each dimension of each wolf: if a dimension exceeds the upper boundary limit, it is set to the upper limit; if it is below the lower boundary limit, it is set to the lower limit, to ensure that candidate solutions can always be sent down for execution.

[0123] The baking control unit calculates the fitness for each wolf, where fitness is defined as follows: ,in The objective function is calculated from step S401. A higher fitness indicates a better candidate solution.

[0124] In each iteration, the baking control unit selects the wolf with the highest fitness as the alpha wolf, and the candidate vector corresponding to the alpha wolf is denoted as . .

[0125] The wolf pack algorithm updates in the order of "scouting-summoning-siege": Reconnaissance Update: For each wolf except the alpha wolf, randomly select a dimension. And update the dimension value as follows: .in To take values ​​in Uniformly random numbers between 1 and 1 This represents the step size for that dimension. Boundary truncation is performed on this dimension after the update. The reconnaissance update is used to expand the search range to avoid premature convergence.

[0126] Summon Update: For each wolf except the alpha wolf, send it to... The update rule is as follows: .in, This is the candidate execution vector for any wolf other than the alpha wolf. As the summoning coefficient, this embodiment takes... The updated function performs boundary truncation on each dimension. The update is then used to accelerate convergence to a better region.

[0127] Siege Update: Perform a detailed local search around the alpha wolf and generate... One siege candidate To determine the number of attack candidates, this embodiment takes... Each siege candidate is generated according to the following rules: .in, For the generated siege candidate vectors, The siege coefficient is taken in this embodiment. , Let the vector of random numbers follow a normal distribution with a mean of 0 and a variance of 1. These are the step size vectors for each dimension. After generation, boundary truncation is performed on each dimension. Each siege candidate is calculated separately. Replace with the smallest one. .

[0128] To ensure that the output execution value does not fluctuate drastically and cause control oscillation, the baking control unit performs smoothing and rate-of-change limiting on the output of u_best. .

[0129] The smoothing rule is: .in This is the smoothed candidate execution vector.

[0130] The rule for limiting the rate of change is: calculate for each dimension. ,when Greater than the step size of this dimension At that time, Truncation is defined as having the same sign and an absolute value equal to 1. Thus, the final output is obtained. , This represents the change in the candidate execution quantity for this dimension.

[0131] The stopping condition is executed according to the following rules: when the number of iterations reaches... Stop when... or when... in 5 consecutive iterations. corresponding The process stops when all improvements are less than 0.001. 0.001 is the convergence threshold, which is written to the process database and can be configured according to material type.

[0132] The baking control unit will eventually and its corresponding The value and stop condition trigger type are written into the temperature control strategy sequence index field of the baking process file, serving as the basis for subsequent traceability and strategy reproduction.

[0133] For example, if for In a certain iteration, we obtained for After smoothing The percentages are [54.5%, 51.5%, 56.0%, 66.0%, 24.5%]. Dimension If the value is less than 10% of the step size, then no truncation is performed. Dimension If the value is less than 15% of the step size, then no truncation is performed. Finally... The values ​​are set to [54.5%, 51.5%, 56.0%, 66.0%, 24.5%] and then issued for execution.

[0134] S403: Smooth and limit the rate of change of the optimal candidate temperature control execution value to obtain the temperature control execution value and send it to the baking device. Append it to the temperature control strategy sequence and return to the acquisition and estimation steps to form a closed-loop iterative update: Specifically, the baking control unit in the first Each control cycle will Issued to the baking device actuator and kept running. Seconds. Among them, zone heating power The power control and circulating air volume are respectively sent to the heater power control interface of the corresponding zone. The control parameters for air intake and exhaust openings are sent to the variable frequency control interface of the circulating fan. Send to the damper actuator control interface.

[0135] Baking control unit in The time of distribution will include "distribution timestamp, Each component value, corresponding Values, percentage of effective channels, percentage of events, suspected over-baking risk markers, and insufficient sample markers are added to the baking process file to form a time-incrementing temperature control strategy sequence. The temperature control strategy sequence is a vector sequence of temperature control execution quantities recorded according to the control cycle.

[0136] when Upon arrival, the baking control unit enters the next control cycle. The process loops in the following fixed order: read the latest post-cleaning process data stream and state feature set, call steps S301–S303 to update the control feedback variable set, and then execute steps S401–S402 to construct the objective function and solve it using the wolf pack algorithm. Finally, step S403 is executed and maintained, thus forming a closed-loop iteration of "acquisition-estimation-solution-execution".

[0137] The baking control unit records the index read and write results of each loop in the baking process file. The index read and write results include at least: the cleaned data stream index, the state feature set index, the control feedback quantity set index, and the temperature control strategy sequence index, so as to ensure that the input, calculation and output of each cycle can be reproduced according to the index in the future.

[0138] In this embodiment of the invention, the introduction of anti-over-baking constraints (combining suspected over-baking risk markers with gas release rate threshold penalties) in temperature control optimization can limit excessive heat exposure while ensuring compliance, reduce the probability of performance degradation and reduce ineffective heat preservation time.

[0139] S5: When the state meets the target material state and the consistency deviation does not exceed the consistency judgment threshold and the over-baking constraint is not triggered, output the baking completion command and terminate the update, and archive the temperature control strategy sequence and baking process file for traceability.

[0140] In one possible implementation, S5 specifically includes sub-steps S501 to S504: S501: Reads only the state compliance, consistency deviation, suspected over-baking risk marker, and insufficient sample marker from the control feedback quantity set index on a control cycle basis. A single-cycle compliance is determined if the state compliance meets the target material state, the insufficient sample marker is not present, the consistency deviation does not exceed the consistency judgment threshold, and the suspected over-baking risk marker is not present. If a preset number of consecutive control cycles achieve single-cycle compliance, a baking completion judgment result is generated, and a baking completion command is output to stop updating the temperature control execution quantity. Specifically, the baking control unit performs a compliance determination at the end of each control cycle. The control cycle is consistent with step S401, and the control cycle length is... It lasts for 10 seconds.

[0141] The baking control unit reads only the latest feedback record corresponding to the control feedback quantity set index from the baking process file, and reads the lower limit of the target material state target range written in step S102 from the baking process file. and upper limit In addition, a consistency judgment threshold is set to ensure that the data source for the compliance judgment is consistent with the data source for closed-loop optimization control.

[0142] The baking control unit extracts the following fields from the latest feedback record: material status index, status compliance, consistency deviation, suspected over-baking risk marker, insufficient sample marker, event percentage, and effective channel percentage, and then performs a single-cycle compliance determination. In this embodiment, the single-cycle compliance determination is valid only if the following logic is met simultaneously: The state compliance rate is 1.

[0143] Insufficient samples are marked as 0.

[0144] The consistency deviation is less than or equal to the consistency judgment threshold.

[0145] The risk of suspected over-baking is marked as 0.

[0146] If any of the above four conditions are not met, the single-cycle compliance determination is invalid, and the baking control unit writes the non-compliance reason code for that cycle into the termination determination record field of the baking process file. The non-compliance reason code includes at least one or a combination of four categories: "Status Not Met", "Insufficient Samples", "Consistency Exceeds Threshold", and "Over-baking Risk".

[0147] To avoid accidental termination due to momentary fluctuations, the baking control unit employs continuous judgment-triggered termination. Specifically, To ensure the number of control cycles required to continuously meet the single-cycle compliance criteria, this embodiment takes... .

[0148] The baking control unit maintains a continuous compliance counter: when a single-cycle compliance determination for a certain control cycle is successful, the continuous compliance counter increments by 1. When a single-cycle compliance determination for a certain control cycle is unsuccessful, the continuous compliance counter is reset to zero.

[0149] When the continuous compliance counter reaches At that time, the baking control unit generates a baking completion determination result and writes it into the termination determination record field. The baking completion determination result includes at least the trigger cycle number, the corresponding material state index, the state compliance rate, the consistency deviation, the suspected over-baking risk marker, and the consistency determination threshold.

[0150] To ensure consistent conflict handling priorities, the baking control unit sets a rule that "over-baking prevention has a higher priority than termination": when the suspected over-baking risk flag is equal to 1, even if the status compliance is equal to 1 and the consistency deviation does not exceed the consistency judgment threshold, no baking completion judgment result is generated, and the failure reason code is written to "over-baking risk". This rule ensures that the termination judgment will not conflict with the over-baking prevention constraint.

[0151] For example, if the feedback records for three consecutive control cycles are as follows: No. Period: Status compliance = 1, consistency deviation = 0.04, threshold = 0.05, suspected over-baking risk marker = 0, insufficient sample marker = 0.

[0152] No. Period: Status compliance = 1, consistency deviation = 0.05, threshold = 0.05, suspected over-baking risk marker = 0, insufficient sample marker = 0.

[0153] No. Period: Status compliance = 1, consistency deviation = 0.03, threshold = 0.05, suspected over-baking risk marker = 0, insufficient sample marker = 0.

[0154] The consecutive successive success counter will then be 1, 2, 3, and so on, until the target is reached. The baking completion determination result is generated and the process proceeds to step S502.

[0155] S502: Execute the termination control procedure, reduce the zone heating power to zero within a preset linear power reduction time, and maintain the circulating air volume and air intake / exhaust opening within a preset safe heat dissipation holding time to enter a safe state. Add the termination control tail section to the temperature control strategy sequence: Specifically, after generating the baking completion determination result, the baking control unit immediately outputs a baking completion command to stop the wolf pack algorithm solution and temperature control execution quantity update in subsequent control cycles, and enters the termination control process.

[0156] The termination control flow uses the execution quantity components defined in steps S401–S403. , , F and A provide reproducible power reduction and heat dissipation control for the actuators, avoiding temperature rebound or local overcooling caused by the thermal inertia of the furnace cavity due to sudden heat interruption.

[0157] Baking control unit sets linear power reduction time With safe heat dissipation time . To determine the time (in seconds) for the zone heating power to linearly decrease from its current value to 0, this embodiment takes... Second. This refers to the time (in seconds) during which the circulating air volume and intake / exhaust air openings are maintained for safe heat dissipation after the zone power is reduced to 0. In this embodiment, it is taken as... Second.

[0158] The baking control unit reads the last execution amount corresponding to the baking completion determination trigger cycle. ,in Include , , , , and with Used as the initial value for termination control.

[0159] exist Within the specified time, the baking control unit operates according to the following rules every Partition power is updated every second: for each partition z, from The linear function decreases to 0, and the linear function satisfies the condition that "at the start of the termination control..." After the termination control begins "The time is 0 seconds." Meanwhile, the circulating air volume... Keep as Air intake and exhaust opening Maintain at 30% to ensure controlled heat dissipation.

[0160] when After completion, the baking control unit will The airflow is fixed at 0 and enters the safe heat dissipation maintenance phase. During the safe heat dissipation maintenance phase, the circulating airflow... Fixed at 40%, air intake and exhaust opening Fixed at 30%, and continued Second.

[0161] To define the executable safety state, the baking control unit will " , , , , The condition is defined as "the maximum furnace temperature is below 150 degrees Celsius". The safe condition is an operating condition that allows opening the door to retrieve materials without posing a thermal risk to personnel or equipment.

[0162] During the safe heat dissipation maintenance phase, the baking control unit continuously reads oven cavity temperature measurement data from the post-cleaning process data stream index and calculates the maximum oven cavity temperature. When the maximum oven cavity temperature is below 150 degrees Celsius and the safe heat dissipation maintenance time has been reached... When the baking control unit outputs a "allow opening the door to retrieve materials" signal, it also writes a timestamp indicating that the safety status has been achieved into the termination determination record field.

[0163] The baking control unit appends the sequence of execution changes during the termination control process as the tail segment of the temperature control strategy sequence to the temperature control strategy sequence index field. This tail segment must include at least: the termination start timestamp, and the timestamps for each... seconds , , , , value, Parameter values The parameter values ​​and safety status are timestamped, making the termination control action traceable and reproducible.

[0164] For example, if the trigger is terminated for: , , , , ], then in The power is updated every 10 seconds. The percentages were 54.5%, 45.4%, 36.3%, 27.3%, 18.2%, 9.1%, and 0, respectively. It remains at 66.0% (because it is greater than 40%). Fixed at 30%. Then proceed to... Second phase, , , The system will continue to output a signal allowing the door to be opened and materials to be retrieved until the maximum furnace temperature is below 150 degrees Celsius.

[0165] S503: Archive the baking process file corresponding to the batch identifier using fixed fields, and under non-abnormal batch conditions, write back the temperature control execution quantity and statistical results corresponding to the baking completion judgment trigger cycle for the initialization of subsequent batches of the same material type: Specifically, after the baking control unit achieves a safe state, it solidifies the baking process file corresponding to this batch identifier into the final version and performs fixed field archiving. The archived content is used for assembly quality traceability and initialization input for subsequent batches of the same material type.

[0166] To ensure reproducibility of the archive, the baking control unit writes the following fields into the archiving area of ​​the baking process file according to a fixed structure: batch identifier, batch information, and target material status target range. Consistency judgment threshold, binding relationship table, abnormal channel marker, degradation marker, post-cleaning process data flow index, state feature set index, material state estimation result index, control feedback quantity set index, temperature control strategy sequence index, baking completion judgment result, termination control tail segment record, safety state achievement timestamp, total baking time, and default weight usage marker and default coefficient usage marker.

[0167] The total baking time is obtained by the difference between the "timestamp of the first original record of the batch entering baking" and the "timestamp of the baking completion determination trigger", and is written to the archive area in seconds for subsequent process comparison.

[0168] To prevent abnormal data from contaminating subsequent initializations, the baking control unit performs an abnormal batch determination on this batch and decides whether to allow it to participate in the initialization of the next batch. An abnormal batch determination is valid if any one of the following conditions is met: 1) Throughout the entire process of this batch, the number of times the insufficient sample marker appeared exceeded 20% of the total number of control cycles.

[0169] 2) Throughout the entire process of this batch, the number of times the suspected over-baking risk marker appeared exceeded 3.

[0170] 3) The number of control cycles with an effective channel ratio of less than 0.70 exceeds 30%. When an abnormal batch is determined, the baking control unit marks this batch as an abnormal batch and adds the batch identifier to the "initialization mask list" in the process database, so that it will not participate in the subsequent updates of the target range and weight coefficient of the same material type.

[0171] If the abnormal batch determination fails, the baking control unit executes the initialization write-back rules for the next batch: 1) For subsequent batches of the same material type, when initializing the wolf pack in step S402, the "baking completion determination trigger cycle" in the temperature control strategy sequence of this batch will be set to the appropriate value. "As The initial center vector is used to shorten the convergence time of the next batch of searches.

[0172] 2) For subsequent batches of the same material type, when generating the target range and consistency judgment threshold in step S102, the final value of the material state index and the final value of the consistency deviation of the batch baking completion judgment trigger cycle are added to the historical qualified batch set to update the statistical values ​​of the 5th percentile, 95th percentile and 95th percentile consistency threshold.

[0173] The baking control unit writes the execution result of the above write-back action into the write-back record field of the baking process archive. The write-back record field includes at least the write-back timestamp, the type of material to be written back, the name of the data item to be written back, and the write-back target (process database table name or index key), so that the closed-loop link of "archiving-write-next batch initialization" is traceable and reproducible.

[0174] In this embodiment of the invention, the temperature control execution quantity and statistical results of the write-back trigger cycle for non-abnormal batches are used for target setting and wolf pack algorithm initialization of subsequent batches of the same material type, which can accelerate convergence, reduce inter-batch fluctuations and improve the consistency compliance rate.

[0175] It should be noted that the above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0176] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0177] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this disclosure, "a plurality of" means two or more, unless otherwise explicitly specified.

[0178] The above are merely specific embodiments of this disclosure, but the scope of protection of this disclosure 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 disclosure should be included within the scope of protection of this disclosure. Therefore, the scope of protection of this disclosure should be determined by the scope of the claims.

Claims

1. A method for optimizing temperature control during the baking process of seals for injection molding machines, characterized in that, The method includes: S1: Generate a batch identifier based on the batch information output by the injection molding machine, establish a baking process file corresponding to the batch identifier, write the target material status and consistency judgment threshold, and generate a binding relationship between the monitoring and execution objects; S2: Collect furnace cavity temperature data, sealing component surface temperature data and furnace cavity gas characteristic data according to the binding relationship, form a process data stream, and obtain a cleaning process data stream after time synchronization and anomaly removal, calculate and file the state characteristic set; S3: Input the set of state features into the material state estimation model to obtain the material state index and state compliance, and calculate the consistency deviation based on the material state index at each location to generate a set of control feedback quantities for filing; S4: Read the set of control feedback quantities to construct a temperature control optimization problem. Under the constraints of maximum allowable temperature, maximum heating rate, upper limit of temperature difference and anti-over-baking, use the wolf pack algorithm to solve the temperature control execution quantity, apply it to the baking device and write it into the temperature control strategy sequence, and return to the closed loop iteration update from S2 to S4.

2. The method for optimizing temperature control during the baking process of seals for injection molding machines according to claim 1, characterized in that, Also includes: S5: When the state compliance meets the target material state and the consistency deviation does not exceed the consistency judgment threshold and the over-baking constraint is not triggered, output the baking completion command and terminate the update, and archive the temperature control strategy sequence and the baking process file for traceability.

3. The method for optimizing temperature control during the baking process of seals for injection molding machines according to claim 1, characterized in that, S1 specifically includes: S101: Obtain the batch information corresponding to the seal to be baked output by the injection molding machine and generate a unique batch identifier; establish a baking process file corresponding one-to-one with the batch identifier, wherein the baking process file includes at least the batch identifier, the batch information, the target material status, the consistency judgment threshold, the binding relationship table and the index field; S102: Determine the target material state and the consistency judgment threshold based on the material type and process number, and write the target material state and the consistency judgment threshold into the baking process file; S103: Generate a binding relationship table of monitoring and execution objects based on the partition configuration of the baking device and write it into the baking process file, and perform verification, abnormal channel marking and downgrade marking on the binding relationship table and file it.

4. The method for optimizing temperature control during the baking process of seals for injection molding machines according to claim 3, characterized in that, The binding relationship table includes at least the sensor number, sensor type, partition number or location identifier, data channel number, and sampling period.

5. The method for optimizing temperature control during the baking process of seals for injection molding machines according to claim 1, characterized in that, S2 specifically includes: S201: Collect the furnace cavity temperature data, the seal surface temperature data, and the furnace cavity gas characteristic data according to the binding relationship with an adaptive sampling period, generate a raw process data stream containing timestamps and event flag bits, and write it into the baking process file; S202: Perform time synchronization and interpolation resampling on the original process data stream, perform physical rationality verification and jump detection to form a synchronized and cleaned process data stream, and write the abnormal channel mark and thermal shock section mark into the baking process file.

6. The method for optimizing temperature control during the baking process of seals for injection molding machines according to claim 5, characterized in that, Also includes: S203: Based on a preset time window, calculate the heating rate related features, heat preservation stability related features, temperature uniformity related features, and gas characteristic change related features from the process data stream after synchronous cleaning, and normalize them to form the state feature set and write it into the baking process file.

7. The method for optimizing temperature control during the baking process of seals for injection molding machines according to claim 1, characterized in that, S3 specifically includes: S301: Read the state feature set, degradation flag and abnormal channel flag from the baking process file, select the material state estimation model version according to the degradation flag, input the state feature set into the material state estimation model to obtain the material state index and perform mechanistic constraint trimming on it; S302: Calculate the state compliance rate based on the target material state target range, and generate a suspected over-baking risk marker and write it into the baking process file; determine the consistency sample set according to the binding relationship table, and if the preset minimum number of valid samples is met, calculate the consistency deviation by the difference between the maximum and minimum values ​​of the material state index in the consistency sample set and write it into the baking process file; otherwise, generate an insufficient sample marker. S303: Encapsulate the material state index, the state compliance, the consistency deviation, the suspected over-baking risk marker, and the insufficient sample marker into the control feedback quantity set and write it into the baking process file.

8. The method for optimizing temperature control during the baking process of seals for injection molding machines according to claim 1, characterized in that, S4 specifically includes: S401: Read the set of control feedback quantities, the target material state target range, and the consistency judgment threshold in units of control cycles, and construct a temperature control optimization problem including state error terms, consistency error terms, energy consumption terms, and constraint penalty terms. The constraint penalty terms include at least the maximum allowable temperature penalty, the maximum heating rate penalty, the zone temperature difference penalty, and the anti-over-drying penalty. S402: Encode the candidate temperature control execution vector into a candidate individual in the wolf pack algorithm, initialize it according to the preset wolf pack size and step size, calculate the fitness based on the temperature control optimization problem, iterate the search through scouting update, summoning update and besieging update, and output the optimal candidate temperature control execution when the stopping condition is met. S403: The optimal candidate temperature control execution quantity is smoothed and the rate of change is limited to obtain the temperature control execution quantity and sent to the baking device. The temperature control strategy sequence is appended and written back to the acquisition and estimation steps to form a closed-loop iterative update.

9. The method for optimizing temperature control during the baking process of seals for injection molding machines according to claim 2, characterized in that, S5 specifically includes: S501: Read the state compliance, consistency deviation, suspected over-baking risk marker, and insufficient sample marker from the control feedback quantity set index only in units of control cycle, and determine that a single cycle is compliant if the state compliance meets the target material state, the insufficient sample marker is not present, the consistency deviation does not exceed the consistency judgment threshold, and the suspected over-baking risk marker is not present; if the single cycle is compliant for a consecutive preset number of control cycles, generate a baking completion judgment result and output the baking completion command to stop the updating of the temperature control execution quantity. S502: Execute the termination control process, reduce the zone heating power to zero within a preset linear power reduction time, and maintain the circulating air volume and air intake / exhaust opening within a preset safe heat dissipation time to enter a safe state, and add the termination control tail segment to the temperature control strategy sequence.

10. The method for optimizing temperature control during the baking process of seals for injection molding machines according to claim 9, characterized in that, Also includes: S503: Archive the baking process file corresponding to the batch identifier with fixed fields, and under non-abnormal batch conditions, write back the temperature control execution quantity and statistical results corresponding to the baking completion determination trigger cycle for the initialization of subsequent batches of the same material type.

Images