A self-optimization system and method for process parameters in precision injection molding production
By separating the immediate response component and the residual amount across cycles, the quality-related cycles are identified, solving the problem of inaccurate attribution of quality anomalies in precision injection molding. This enables precise optimization of process parameters and mold condition management, thereby improving the stability of product quality.
Patent Information
- Application Number
- CN202611181601.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-05
- Publication Date
- 2026-11-10
AI Technical Summary
In existing precision injection molding technology, it is difficult to accurately attribute the causes of quality abnormalities, leading to incorrect parameter correction directions, repeated oscillations, or the formation of pseudo-optimal process windows. Furthermore, the gradual degradation of the mold cannot be identified in a timely manner, affecting product quality.
By collecting process setting data and process response data, the instantaneous response component and cross-cycle residual amount are separated, a cross-cycle state sequence is established, the quality-related cycle range is identified, the quality deviation is allocated, candidate parameter adjustment data is generated, and process disturbances and mold degradation are distinguished for optimization adjustment.
It enables precise adjustment of process parameters, reduces the possibility of incorrect attribution of delayed quality results, distinguishes between compensable disturbances and mold degradation, avoids abnormal samples from affecting subsequent optimization, and improves the stability of product quality.
Smart Images

Figure CN123067375A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of precision injection molding control technology, and in particular to a process parameter self-optimization system and method for precision injection molding production. Background Technology
[0002] Precision injection molding is widely used in the production of automotive parts such as connector housings, sensor brackets, and functional clips. The quality of these products is influenced by a combination of parameters, including melt temperature, mold temperature, injection speed, injection pressure, holding pressure, holding time, and cooling time. Current technologies typically establish a correlation between process parameters and product quality by collecting injection molding machine operating data, cavity pressure, temperature profiles, or product inspection results. Predictive models or optimization algorithms are then used to adjust parameters for subsequent injection molding cycles.
[0003] However, injection molds, hot runners, cooling circuits, and resin within the barrel all exhibit continuous thermal and rheological states across cycles. Quality results such as dimensions, warpage, and assembly performance are often only obtained after the product is demolded and further cooled, or after offline testing. Existing methods tend to directly attribute delayed quality anomalies to the most recent parameter adjustment, ignoring the cumulative effects of previous cycles. This can lead to incorrect parameter correction directions, repeated oscillations, or the formation of pseudo-optimal process windows.
[0004] Furthermore, existing self-optimization methods typically treat material fluctuations, environmental changes, and mold condition variations as disturbances that can be compensated for through parameter adjustments. When venting channels become contaminated, cooling water channels become blocked, gates wear, or hot runner performance deteriorates, the system may temporarily maintain product quality by increasing pressure, extending holding time, or cooling time, thereby masking progressive mold degradation and causing process parameters to gradually approach safety boundaries. Therefore, existing technologies still suffer from the problem of accurately attributing delayed quality results to causal periods and the difficulty in distinguishing between compensable process disturbances and mold degradation that is no longer suitable for compensation. Summary of the Invention
[0005] To address the above problems, embodiments of the present invention provide a method for self-optimization of process parameters for precision injection molding production, the method comprising:
[0006] Collect process setting data, process response data and product identification data according to the injection molding cycle, and determine quality benchmark data from historical qualified cycles;
[0007] Extract the periodic state characteristics of the process response data, and establish a setting response reference based on the relationship between the changes in process setting data and the changes in periodic state characteristics of historical qualified cycles;
[0008] The periodic state characteristics of adjacent injection molding cycles are differentially analyzed, and the instantaneous response components are separated according to the set response reference. The remaining state changes form the cross-cycle residual amount, and the components are connected in time sequence to form a cross-cycle state sequence.
[0009] Obtain product quality data including detection time stamps and product identifiers, determine the production cycle based on the product identifiers, and determine the quality-related cycle range based on the detection time stamps and cross-cycle state sequences;
[0010] Based on the transmission direction of the residual amount across cycles and the sequential relationship of cycles within the aforementioned quality correlation cycle, the quality deviation is allocated to obtain the cycle attribution coefficient;
[0011] Based on the aforementioned cycle attribution coefficients, process setting data and quality deviations for each associated cycle are combined to form attribution association samples, and candidate parameter adjustment data are generated.
[0012] Compensation trend data is generated based on the immediate response components after adjusting the candidate parameters, the residual amount across cycles, and the quality data of subsequent products; a process disturbance identifier is generated when the process disturbance judgment condition is met, and a mold degradation identifier is generated when the mold degradation judgment condition is met.
[0013] Update subsequent cycle process settings data based on the process disturbance flag; stop updating and generate mold maintenance data based on the mold degradation flag.
[0014] Furthermore, the process setting data includes injection stage setting data, holding pressure stage setting data, and cooling stage setting data; the process response data includes at least two of cavity pressure timing, screw displacement timing, and mold temperature timing; based on the stage switching signal of the injection molding machine, the process response data is divided into filling stage data segments, holding pressure stage data segments, and cooling stage data segments, and the timing of each data segment is aligned, and the change direction, change rate, and stage connection state of each data segment are combined into the periodic state characteristics.
[0015] Furthermore, a setting response reference is established based on the correspondence between changes in process setting data and changes in cycle state characteristics during the historical qualified cycles; the cycle state characteristics of adjacent injection molding cycles are differentially processed to obtain the state change amount; according to the setting response reference, the instantaneous response component corresponding to the changes in process setting data is separated from the state change amount to obtain the cross-cycle residual amount; the cross-cycle state sequence is established based on the directional consistency and temporal continuity between the cross-cycle residual amounts of consecutive injection molding cycles.
[0016] Furthermore, a cycle timescale is configured for each injection molding cycle. Based on the interval between the detection timescale and the cycle timescale of the production cycle, the temporal correspondence between the product quality data and the production cycle is verified. Starting from the production cycle, injection molding cycles in which the cross-cycle residual amount maintains directional continuity are selected sequentially forward along the cross-cycle state sequence to form an initial association range. The position where the directional continuity of the cross-cycle residual amount is interrupted and the position where the process setting data is switched discontinuously are determined respectively, and the position closer to the production cycle is determined as the boundary of the quality association cycle range. When only one of the positions exists, the boundary is determined by that position. When neither of the two positions exists, the boundary is determined by the starting position of the initial association range.
[0017] Further, within the quality correlation cycle range, according to the arrangement of process response data type, forming stage, and cycle state characteristics, the corresponding cross-cycle residual amounts in consecutive correlation cycles are connected sequentially to form a residual transfer chain; taking the quality deviation direction corresponding to the production cycle as the end constraint, cross-cycle residual amounts corresponding to the quality deviation direction are selected forward along the residual transfer chain; based on the continuation position of the cross-cycle residual amounts in the residual transfer chain of each correlation cycle, and the changing relationship of cross-cycle residual amounts between adjacent correlation cycles, the quality deviation allocation amount corresponding to each correlation cycle is determined; based on the corresponding ratio between each quality deviation allocation amount and the quality deviation, the cycle attribution coefficient is obtained.
[0018] Furthermore, the associated cycle with the highest cycle attribution coefficient and a corresponding relationship between the change direction of the cross-cycle residual amount and the quality deviation is determined as the main attribution cycle; the process setting data, cycle state characteristics, and quality deviation assigned to the main attribution cycle are combined into the attribution association sample; using the cycle state characteristics of the main attribution cycle as the matching condition, a reference cycle with corresponding cycle state characteristics and different process setting data is selected from the historical qualified cycles; the differences in process setting data and quality state between the main attribution cycle and the reference cycle are compared to determine the candidate adjustment dimensions and candidate adjustment directions of the process setting data, forming the candidate parameter adjustment data.
[0019] Further, the expected response direction corresponding to the candidate parameter adjustment data is determined according to the set response reference; the instantaneous response component obtained after using the candidate parameter adjustment data is compared with the expected response direction, and the continuation direction of the cross-cycle residual amount and the quality deviation change of the subsequent product quality data are recorded in the order of the injection molding cycle to form the compensation trend data; when the instantaneous response component corresponds to the expected response direction, the cross-cycle residual amount does not continue continuously in the direction opposite to the expected response direction, and the absolute value of the quality deviation decreases, the process disturbance indicator is generated; when the instantaneous response component corresponds to the expected response direction, the cross-cycle residual amount continues continuously in the direction opposite to the expected response direction, and after continuously executing the candidate parameter adjustment data, the decrease in the absolute value of the quality deviation relative to the previous injection molding cycle gradually decreases, the mold degradation indicator is generated; when the cycle residual amount still continues continuously in the direction opposite to the expected response direction, and after continuously executing the candidate parameter adjustment data, the decrease in the absolute value of the quality deviation relative to the previous injection molding cycle gradually decreases, the compensation trend data is determined to meet the mold degradation judgment condition, and the state judgment indicator is set as the mold degradation indicator.
[0020] Further, after generating the mold degradation identifier, the process response data type, molding stage, and sensor acquisition location that formed the cross-cycle residual amount are read; the cross-cycle residual amounts that continue continuously in the same direction are aggregated according to the process response data type, molding stage, and sensor acquisition location to determine the corresponding abnormal molding stage and abnormal acquisition location; mold maintenance data containing the abnormal molding stage, abnormal acquisition location, process response data type, and quality correlation cycle range is generated; the attribution correlation samples corresponding to the quality correlation cycle range are marked as mold degradation samples, and the use of the mold degradation samples for the generation of subsequent candidate parameter adjustment data is stopped.
[0021] Furthermore, a method for self-optimizing process parameters for precision injection molding production further includes: performing consistency verification on the cross-cycle residual amount based on the stage transmission relationship of different process response data in historical qualified cycles; determining the cross-cycle residual amount that supports the stage transmission relationship and whose quality deviation direction is consistent as a reliable residual amount, and using only the reliable residual amount to form a cross-cycle state sequence; marking the cross-cycle residual amount that does not have a corresponding supporting cross-cycle residual amount and whose corresponding product quality data has not changed accordingly as abnormal residual amount.
[0022] A process parameter self-optimization system for precision injection molding production, the system comprising:
[0023] Cycle data acquisition module: Collects process setting data, process response data and product identification according to the injection molding cycle, and determines quality benchmark data from historical qualified cycles;
[0024] State feature construction module: Extracts the periodic state features of the process response data, and establishes a set response reference based on the relationship between the changes in process setting data and the changes in periodic state features in historical qualified cycles;
[0025] Cross-cycle state module: Differentiate the cycle state characteristics of adjacent injection cycles, separate the instantaneous response components according to the set response reference, form cross-cycle residual quantity with the remaining state changes, and connect them in time sequence to form a cross-cycle state sequence.
[0026] Quality cycle association module: acquires product quality data including detection time stamps and product identifiers, determines the production cycle based on the product identifiers, and determines the quality association cycle range based on the detection time stamps and cross-cycle state sequences;
[0027] Quality attribution calculation module: Based on the transmission direction and cycle sequence relationship of the residual amount across cycles within the quality correlation cycle range, the quality deviation is allocated to obtain the cycle attribution coefficient;
[0028] Parameter adjustment generation module: Combines the process setting data and quality deviation of each associated cycle according to the cycle attribution coefficient to form attribution association samples and generate candidate parameter adjustment data;
[0029] Compensation trend discrimination module: Based on the immediate response components after candidate parameter adjustment, cross-cycle residual amount and subsequent product quality data, compensation trend data is generated; when the process disturbance judgment condition is met, a process disturbance identifier is generated; when the mold degradation judgment condition is met, a mold degradation identifier is generated.
[0030] The optimization and maintenance execution module updates the subsequent cycle process setting data based on the process disturbance flag; it stops updating and generates mold maintenance data based on the mold degradation flag.
[0031] This invention has at least the following technical effects or advantages:
[0032] This invention achieves synergistic optimization of process parameter adjustment and mold condition management through cross-cycle state identification, quality attribution, and compensation trend discrimination. By separating the immediate response component caused by changes in process settings and extracting cross-cycle residual amounts, this invention can identify the state continuity relationship between consecutive injection molding cycles, reducing the possibility of delayed quality results being incorrectly attributed to the most recent cycle. Quality deviations are allocated based on the transmission direction of cross-cycle residual amounts and the cycle sequence, forming cycle attribution coefficients and attribution correlation samples, giving candidate parameter adjustment data a clear cycle source and state basis. By comparing the expected response direction, immediate response component, cross-cycle residual amounts, and subsequent quality deviations, it can distinguish between process disturbances that can be eliminated through parameter adjustment and persistent mold degradation states. After identifying mold degradation, parameter compensation is stopped, and mold maintenance data containing abnormal molding stages and abnormal acquisition locations is generated. Simultaneously, mold degradation samples are isolated to prevent abnormal samples from continuing to affect subsequent parameter optimization. Attached Figure Description
[0033] Figure 1 This is a flowchart of a process parameter self-optimization method for precision injection molding production, as shown in Example 1.
[0034] Figure 2 This is a flowchart of the cross-cycle residual consistency verification method in Example 2;
[0035] Figure 3 This is a schematic diagram of the connection of a process parameter self-optimization system for precision injection molding production in Example 3. Detailed Implementation
[0036] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0037] Example 1:
[0038] Please see Figure 1 As shown, embodiments of the present invention provide a method for self-optimization of process parameters for precision injection molding production, the method comprising:
[0039] S1. Collect process setting data, process response data and product identification according to the injection molding cycle, and determine the quality benchmark data from the historical qualified cycle.
[0040] S2. Extract the periodic state characteristics of the process response data, and establish a setting response reference based on the relationship between the changes in process setting data and the changes in periodic state characteristics of historical qualified cycles.
[0041] S3. Differentiate the periodic state characteristics of adjacent injection cycles, separate the instantaneous response components according to the set response reference, form the cross-cycle residual amount with the remaining state changes, and connect them in time sequence to form a cross-cycle state sequence.
[0042] S4. Obtain product quality data including detection time stamps and product identification, determine the production cycle based on the product identification, and determine the quality-related cycle range based on the detection time stamps and cross-cycle status sequences;
[0043] S5. Based on the transmission direction of the residual amount across cycles within the quality correlation cycle range and the relationship between cycles, the quality deviation is allocated to obtain the cycle attribution coefficient.
[0044] S6. Combine the process setting data and quality deviation of each related cycle according to the cycle attribution coefficient to form an attribution correlation sample and generate candidate parameter adjustment data.
[0045] S7. Based on the immediate response components after adjusting the candidate parameters, the residual amount across cycles, and the subsequent product quality data, compensation trend data is formed; when the process disturbance judgment conditions are met, a process disturbance identifier is generated; when the mold degradation judgment conditions are met, a mold degradation identifier is generated.
[0046] S8. Update subsequent cycle process setting data based on process disturbance flags; stop updating and generate mold maintenance data based on mold degradation flags.
[0047] In this embodiment, each time the injection molding machine completes the process of mold closing, filling, holding pressure, cooling and mold opening, it is recorded as an injection cycle. The controller establishes cycle data records in units of injection cycles and writes the process setting data and process response data within the same injection cycle into the corresponding cycle data records.
[0048] Process setting data refers to the data issued by the injection molding machine controller before or during molding to control the operating status of the corresponding molding stage. According to the process action stage, the process setting data is divided into injection stage setting data, holding pressure stage setting data, and cooling stage setting data. Injection stage setting data corresponds to the control process of molten material entering the mold cavity. Holding pressure stage setting data corresponds to the control process of maintaining the pressure state of the material in the cavity after the cavity is filled. Cooling stage setting data corresponds to the control process of cooling and solidifying the product in the mold. All types of process setting data retain their respective injection cycle, molding stage, and data writing time.
[0049] Process response data refers to the actual time-series detection data generated by the injection molding equipment and injection mold after executing the process setting data. It includes at least two of the following: cavity pressure time-series, screw displacement time-series, and mold temperature time-series. The cavity pressure time-series is continuously collected by a pressure detection unit located in or connected to the mold cavity to record the pressure change over time during the process of material entering the cavity and being pressurized within the cavity. The screw displacement time-series is read by the displacement detection unit or controller of the injection molding machine to record the axial position change of the screw during injection and holding pressure. The mold temperature time-series is obtained by a temperature detection unit located at a predetermined collection position on the mold to record the temperature change of the mold during filling, holding pressure, and cooling. All process response data use the same injection cycle identifier and record the corresponding collection time.
[0050] When the molding control mode changes, the injection molding machine controller outputs a stage switching signal. The process response data between the injection start signal and the injection to holding pressure switching signal is defined as the filling stage data segment, the process response data between the injection to holding pressure switching signal and the holding pressure end signal is defined as the holding pressure stage data segment, and the process response data between the holding pressure end signal and the cooling end signal is defined as the cooling stage data segment. For data that does not change continuously in a certain stage, the acquisition results within that stage are still retained to characterize the maintenance state of the data within the stage.
[0051] Since the duration of each stage in different injection molding cycles may vary, before extracting cycle state features, time-series alignment is performed on data segments of the same stage in the response data of the same process. Specifically, the time corresponding to the switching signal of each stage is used as the boundary of the data segment, and the start time of the data segment is used as the time sequence start point of the data segment. According to the sequential position of the sampling points in the data segment, the data segments of different injection molding cycles are mapped to a unified stage time sequence position. When the number of sampling points contained in different data segments is inconsistent, the corresponding time sequence position is supplemented between adjacent sampling points according to the data change relationship, so that data of the same data type and the same molding stage can be compared by position.
[0052] After timing alignment is completed, the direction of change, rate of change, and stage connection status of each data segment are determined. The direction of change is determined based on the data increase or decrease relationship between adjacent timing positions and is recorded as rising, falling, or remaining constant. The rate of change is determined based on the amount of data change at adjacent timing positions and the corresponding time interval and is arranged according to the timing order within the data segment. The stage connection status is determined based on the end state of the previous stage data segment and the beginning state of the next stage data segment, including whether the direction of data change between the two stages continues, whether the data change changes in a turning point, and whether the data between the preceding and following stages remains continuous.
[0053] Following the order of filling stage, holding pressure stage, and cooling stage, the change direction, change rate, and stage connection state of the corresponding data segments of cavity pressure timing, screw displacement timing, and mold temperature timing are arranged sequentially to form the periodic state characteristics of the current injection cycle. Each injection cycle adopts the same data type order, stage order, and characteristic arrangement method to make the periodic state characteristics of consecutive injection cycles have a corresponding relationship, and are used to form the cross-cycle state sequence of claim 1.
[0054] For example: In the continuous injection molding process of automotive connector housing, if the cavity pressure of a certain injection cycle continues to rise during the filling stage and then slowly decreases after entering the holding stage, the screw displacement continues to move forward during the filling stage and maintains a small change during the holding stage, and the mold temperature rises during the filling stage and falls during the cooling stage, then the direction and rate of change of the above data in the corresponding stages are recorded respectively, and the connection state between the filling stage and the holding stage, and between the holding stage and the cooling stage are recorded, thereby forming the periodic state characteristics of the injection cycle.
[0055] In this embodiment, after obtaining the cycle status characteristics of each injection molding cycle in the manner described above, the injection molding cycle in which the product quality data meets the preset quality inspection requirements is first selected from the historical production records as the historical qualified cycle; the quality benchmark data is determined based on the product quality data of the historical qualified cycle, the historical qualified cycles are arranged according to the production time, and the corresponding process setting data and cycle status characteristics are retained.
[0056] The set response reference is used to represent the corresponding changes in process response data within the same injection cycle when process setting data changes. When establishing the set response reference, adjacent historical qualified cycles are selected, and the process setting data of the two historical qualified cycles are compared to determine the changed injection stage setting data, holding pressure stage setting data, or cooling stage setting data. At the same time, the cycle state features with the same position in the two historical qualified cycles are compared to determine the corresponding feature change direction and degree. The change stage, change direction, and change degree of the process setting data are associated and recorded with the change direction and change degree of the corresponding cycle state feature to form the set response reference.
[0057] The cycle state characteristics of the current injection cycle and the previous injection cycle are differentially processed. The differential processing is carried out item by item according to the same data type, the same molding stage and the same feature arrangement position. The difference between the cycle state characteristics of the current injection cycle and the corresponding cycle state characteristics of the previous injection cycle constitutes the state change quantity. The state change quantity represents the change of the state corresponding to the cavity pressure timing, screw displacement timing and mold temperature timing between adjacent injection cycles.
[0058] Subsequently, the process setting data of adjacent injection cycles are compared. When the process setting data changes, the corresponding cycle state characteristic change is selected from the setting response reference according to the molding stage, direction and degree of change. The cycle state characteristic change is determined as the instantaneous response component. The instantaneous response component refers to the state change part directly caused by the change in the current process setting data and reflected in the process response data of the current injection cycle. The instantaneous response component is subtracted from the corresponding characteristic position of the state change amount, and the remaining part is used as the cross-cycle residual amount. When the process setting data of adjacent injection cycles does not change, the corresponding instantaneous response component is zero, and the state change amount is used as the cross-cycle residual amount.
[0059] According to the chronological order of the injection molding cycle, the cross-cycle residual amounts of consecutive injection molding cycles are connected. For the same data type, the same molding stage, and the same feature arrangement position, when the cross-cycle residual amounts of multiple consecutive injection molding cycles all show the same upward direction, downward direction, or maintenance direction, they are considered to have directional consistency. When the corresponding injection molding cycles are continuous in production time and there are no missing cycles in between, they are considered to have temporal continuity. The cross-cycle residual amounts that simultaneously satisfy directional consistency and temporal continuity are sequentially written into the same state chain segment. If they do not satisfy the conditions, the current state chain segment is terminated, and a new state chain segment is re-established from the subsequent injection molding cycle. The state chain segments and their corresponding injection molding cycles together constitute the cross-cycle state sequence.
[0060] For example: When the process setting data remains unchanged during a continuous injection molding cycle, but the cavity pressure corresponding to the holding pressure stage changes continuously in the same direction, this change cannot be attributed to the instantaneous response component generated by the change in process setting data, and is therefore written into the cross-cycle state sequence as a cross-cycle residual amount.
[0061] In this embodiment, a cycle time stamp is configured for each injection molding cycle. The cycle time stamp is used to indicate the position of the corresponding injection molding cycle in the production sequence. It can be the start time of the injection molding cycle, the end time of the cycle, or a sequence identifier generated by the production controller. Each cycle time stamp is associated with the corresponding cycle data record so as to call the process setting data, cycle status characteristics, and cross-cycle residual amount in the order of production.
[0062] After the product completes quality inspection, the inspection time stamp and product identifier carried by the product quality data are obtained. The inspection time stamp indicates the moment when the product quality data is formed or written into the production system. The corresponding cycle data record is found according to the product identifier. The injection molding cycle to which the cycle data record belongs is determined as the production cycle. The inspection time stamp is compared with the cycle time stamp of the production cycle to obtain the time interval from product molding to the formation of quality data.
[0063] The detection sequence of the product quality data is checked against the production sequence of the corresponding production cycle based on the time interval. When the detection timescale is later than the cycle timescale of the production cycle and the product identification is consistent, the time sequence correspondence between the product quality data and the production cycle is confirmed to be valid. The time interval is used to verify the correspondence between the product quality data and the production cycle, and is not used to limit the duration of the state before the production cycle.
[0064] Starting from the production cycle, the process searches forward cycle by cycle along the cross-cycle state sequence in the opposite direction to the production timeline. Injection cycles with continuous cross-cycle residual amounts are included in the initial association range. For each injection cycle selected, the cross-cycle residual amount between the current injection cycle and the next injection cycle is read, and the changing direction of the two in the same data type, the same molding stage, and the same feature arrangement position is compared. When the cross-cycle residual amounts of consecutive injection cycles all maintain the same direction of increase, decrease, or maintenance, the corresponding injection cycle is retained as the association cycle.
[0065] During the forward search, if the direction of change of the residual amount across cycles changes, it indicates that the continuity of the state between the previous and next cycles is interrupted. In this case, the next injection molding cycle after the change is taken as the starting boundary of the quality-related cycle range. If the process setting data changes discontinuously, the forward selection will also stop. Discontinuous switching means that the process setting data does not change continuously along the original adjustment direction, but rather the process scheme is replaced, production is restarted, or there is a switch between different molding conditions, so that the cycle states before and after the switch no longer have a basis for continuous comparison.
[0066] Finally, the injection cycles continuously selected from the starting boundary to the production cycle are determined as the quality-related cycle range, and each injection cycle within the range is used as the related cycle for subsequent calculation of cycle attribution coefficients. If there is no interruption of directional continuity or discontinuous switching of process setting data within the initial related range, the starting cycle of the initial related range is used as the starting boundary of the quality-related cycle range.
[0067] In this embodiment, after determining the quality-related cycle range, the controller performs source matching on the cross-cycle residual amount of each related cycle within the range. Source matching means that only cross-cycle residual amounts that originate from the same process response data type, the same molding stage, and the same arrangement position in the cycle state characteristics are matched. For example, the feature representing the increase in cavity pressure timing in the filling stage only corresponds to the feature at the same position in other related cycles, and is not mixed with the cavity pressure feature or screw displacement feature in the holding stage.
[0068] According to the production sequence of the associated cycles, the cross-cycle residual amounts that are of the same origin are connected in sequence to form a residual transfer chain. The residual transfer chain is used to record the continuation of the same process state change in the continuous injection molding cycle. If the corresponding cross-cycle residual amount is missing in the adjacent associated cycle, or the change direction of the corresponding cross-cycle residual amount is interrupted, the current residual transfer chain is terminated at that position.
[0069] The direction of quality deviation is determined based on the relationship between the product quality data and the quality benchmark data during the production cycle. When the product quality data is higher than the quality benchmark data, it is recorded as the first deviation direction; when it is lower than the quality benchmark data, it is recorded as the second deviation direction. For different quality items included in the product quality data, the direction of quality deviation is determined separately and then allocated accordingly.
[0070] Using the production cycle as the end of the residue transfer chain, the system searches forward cycle by cycle along the residue transfer chain from the production cycle. Based on the correspondence between the changes in the cycle state characteristics of the same position in the historical qualified cycles and the changes in product quality data, it is determined whether there is a correspondence between the cross-cycle residue amount and the direction of the quality deviation. The correspondence includes the relationship that the cross-cycle residue amount and the quality deviation change in the same direction, and the relationship that the cross-cycle residue amount and the quality deviation change in opposite directions. Only the cross-cycle residue amount that meets the correspondence and exists continuously in the residue transfer chain is retained.
[0071] For each retained correlation period, first determine the attribution contribution value of each correlation period to the current quality deviation. For the first correlation period in the product quality data... The first quality item, the The attribution contribution value for each associated period is determined according to the following formula:
[0072] ;
[0073] The attribution contribution values for each related period are processed uniformly and allocated to the [number]th [period]. The quality deviation for each associated period is determined by the following formula:
[0074] ;
[0075] Based on the ratio between the quality deviation allocation and the corresponding quality deviation in the production cycle, the first... The periodic attribution coefficient corresponding to each associated period is determined according to the following formula:
[0076] ;
[0077] When the sum of the attribution contribution values of each associated cycle is zero, or when the quality deviation corresponding to the production cycle is zero, no cycle attribution coefficient is formed for that quality item.
[0078] in, Indicates the first The first quality item corresponding to the Attribution contribution value for each associated period; Indicates the first The periodic attribution coefficient of each associated period; Indicates the sequential position of the correlation cycle within the range of quality correlation cycles; This indicates the quality items in the product quality data; Indicates the relationship with the first A set of related cycles with corresponding relationships among quality items; Indicates the sequential position of the production cycle within the associated cycle set;
[0079] Indicates the first In the related period, with the first The cross-cycle residual amount corresponding to each quality item is determined based on the arrangement of process response data type, forming stage, cycle state characteristics, and data change relationship in historical qualified cycles. This indicates the magnitude of the residual amount across cycles; the sum of the residual amounts across all associated cycles is used to form the residual percentage.
[0080] Indicates the direction corresponding to the marker, when the first The direction of cross-cycle residual change in the first associated cycle is related to the first cycle. When the direction of quality deviation for a quality item conforms to the corresponding relationship in the historical qualified cycle. Take the valid value; if it does not conform to the correspondence, Take an invalid value so that the corresponding associated period does not participate in the quality deviation allocation of this quality item;
[0081] The multiplication term in the formula represents the residual amount across periods, derived from the first period. The degree to which related cycles are continuously transmitted to the production cycle; Indicates the sequential position of adjacent associated cycles in the residual transmission chain; and These represent the cross-cycle residue levels of two adjacent related cycles, respectively; the ratio of the smaller residue level to the larger residue level is used to indicate the degree of residue retention between adjacent cycles. This represents a preset positive number used to avoid a denominator of zero, when the first... When the associated cycle is a production cycle, the product term takes a unit value;
[0082] This indicates the cyclical positional relationship between the associated cycle and the production cycle; the closer the associated cycle is to the production cycle, the larger this item is; by uniformly processing each attribution contribution value, the sum of the attribution coefficients for each cycle corresponding to the same quality item is made into a unit value.
[0083] Indicates the first step in the production cycle Quality deviation of product quality data for each quality item relative to the quality benchmark data; Indicates allocation to the first The sum of the quality deviations for each associated period and the corresponding quality deviation allocation for each associated period. Maintain consistency.
[0084] After the calculation is completed, the arrangement and position of the periodic attribution coefficient, quality deviation allocation, associated period, process response data type, forming stage and periodic state characteristics are associated and saved, which are used for determining the main attribution period and forming attribution association samples in step S6.
[0085] For example: If the product size deviates during the production cycle, and the cross-cycle residual amount corresponding to the cavity pressure during the holding pressure stage maintains the same direction in multiple consecutive related cycles and continues into the production cycle, then the size deviation is allocated forward along the residual transfer chain; related cycles before the directional interruption participate in the allocation, and injection molding cycles earlier than the directional interruption are no longer included in the current cycle attribution.
[0086] In step S6, based on the periodic attribution coefficient obtained in step S5, the main attribution period is determined from the quality-related period range, and candidate parameter adjustment data is generated based on the main attribution period.
[0087] Specifically, the related cycles are sorted according to the cycle attribution coefficients corresponding to each related cycle. The related cycle with the highest cycle attribution coefficient is selected, and the correspondence between the cross-cycle residual amount and the quality deviation of the related cycle is checked. The correspondence means that the change direction of the cross-cycle residual amount can correspond to the formation direction of the current quality deviation according to the change relationship between the process response data and the product quality data in the historical qualified cycles. The related cycle that satisfies the correspondence is determined as the main attribution cycle. If the related cycle with the highest cycle attribution coefficient does not satisfy the correspondence, the subsequent related cycles are checked according to the order of the cycle attribution coefficients.
[0088] The process setting data, cycle status characteristics, and quality deviations assigned to the main attribution cycle in step S5 are associated with the same cycle identifier to form an attribution association sample. The quality deviations assigned to the main attribution cycle are used to represent the corresponding part of the current product quality deviation in that cycle, avoiding the direct attribution of complete quality deviations to a single injection molding cycle.
[0089] Subsequently, using the periodic state characteristics of the main attribution cycle as the matching condition, a reference cycle is selected from the historical qualified cycles. During matching, the process response data type, forming stage, and periodic state characteristics are compared item by item. Historical qualified cycles with the same direction of change, corresponding stage connection states, and consistent overall change relationship are selected. The process setting data of the reference cycle should differ from the process setting data of the main attribution cycle so that the two can be used to determine the correspondence between changes in process setting data and changes in quality status.
[0090] Quality status indicates the state of product quality data relative to quality benchmark data. Product quality data in the reference period conforms to quality benchmark data. The quality status of the main attribution period is represented by the quality deviation assigned to that period. The process setting data of the same molding stage and the same data item in the main attribution period and the reference period are compared item by item to determine the dimensions of process setting data that have differences.
[0091] For each process setting data dimension, it is further determined whether the corresponding quality status tends to the quality benchmark data when the process setting data changes from the process setting data of the main attribution period to the process setting data of the reference period. The process setting data dimension that satisfies this correspondence is determined as the candidate adjustment dimension, and the direction of change from the process setting data of the main attribution period to the process setting data of the reference period is determined as the candidate adjustment direction.
[0092] When multiple reference cycles are matched, the above comparisons are performed separately, and process setting data dimensions with the same candidate adjustment direction in multiple reference cycles are retained. For process setting data dimensions with inconsistent candidate adjustment directions, no adjustment is made in the current injection molding cycle. The retained candidate adjustment dimensions, the corresponding candidate adjustment directions and their respective molding stages are combined to form candidate parameter adjustment data, which is used to form compensation trend data in subsequent steps.
[0093] In step S7, the candidate parameter adjustment data formed in step S6 is applied to subsequent injection molding cycles, and compensation trend data is formed based on the adjusted process response data and product quality data.
[0094] First, based on the candidate adjustment dimensions, candidate adjustment directions, and molding stages contained in the candidate parameter adjustment data, the corresponding periodic state characteristic change direction is found from the set response reference, and this change direction is determined as the expected response direction. The expected response direction indicates the change direction that the corresponding process response data should produce in the corresponding molding stage after the candidate parameter adjustment data is executed when there is no continuous state shift in the injection molding system.
[0095] After completing a subsequent injection molding cycle, following step S3, the cycle state characteristics of the current injection molding cycle are differentiated from those of the previous injection molding cycle. The instantaneous response component directly caused by the candidate parameter adjustment data is separated from the state change. The instantaneous response component and the expected response direction are compared item by item according to the same process response data type, the same molding stage, and the same cycle state characteristics. When the instantaneous response component is consistent with the expected response direction, it indicates that the injection molding equipment has executed the candidate parameter adjustment data and formed the corresponding current cycle response.
[0096] Based on this, the cross-cycle residual amount after the separation of the immediate response component is further extracted, and the change direction of the cross-cycle residual amount is recorded according to the order of the injection molding cycle. At the same time, the quality data of the corresponding subsequent products is obtained, the quality deviation of the subsequent product quality data relative to the quality benchmark data is calculated, and the adjustment direction of the candidate parameter adjustment data, the comparison result of the immediate response component and the expected response direction, the continuation direction of the cross-cycle residual amount, and the continuous change result of the quality deviation are associated and saved in the cycle order to form compensation trend data.
[0097] A process disturbance flag is generated when the immediate response component corresponds to the expected response direction, the cross-cycle residual amount does not continue continuously in the opposite direction to the expected response direction, and the absolute value of the quality deviation in subsequent product quality data decreases. The process disturbance flag indicates that the state change corresponding to the quality deviation can change with the execution of candidate parameter adjustment data.
[0098] When the instantaneous response component corresponds to the expected response direction, and the residual amount across cycles continues in the opposite direction to the expected response direction in the continuous injection cycle, the correspondence between the continuous candidate parameter adjustment data and the quality deviation change is compared. After the candidate parameter adjustment data is continuously executed, when the absolute value of the quality deviation gradually decreases relative to the previous injection cycle, a mold degradation indicator is generated.
[0099] Among them, the immediate response component corresponds to the expected response direction and is used to exclude situations where candidate parameter adjustment data has not been executed or the equipment response is abnormal; the cross-cycle residual amount continues continuously in the opposite direction, indicating that the process state has not been eliminated with the parameter response of the current cycle; the absolute value of the quality deviation gradually decreases relative to the previous injection cycle, indicating that the degree of quality improvement generated by continuous parameter adjustment continues to decline. The above three types of data come from the same process response data type, the same molding stage and the corresponding injection cycle, and data from different molding stages or different collection locations are not cross-judged.
[0100] For example, when candidate parameter adjustment data is applied to the holding pressure stage, and the instantaneous response component is consistent with the expected response direction in the set response reference, while the cross-cycle residual amount corresponding to the cavity pressure in the holding pressure stage continues to change in the opposite direction to the expected response direction, and the absolute value of the product size deviation gradually decreases after continuous adjustment relative to the previous injection cycle, a mold degradation indicator is generated; when the cross-cycle residual amount no longer continues in the opposite direction, and the absolute value of the product size deviation decreases, a process disturbance indicator is generated.
[0101] In step S8, after generating the mold degradation identifier, the source of the cross-cycle residual amount that formed the mold degradation identifier is traced back, and mold maintenance data is generated.
[0102] When collecting process response data, the process response data type, molding stage, and sensor acquisition location are recorded together for each process response data. The process response data type is used to distinguish between cavity pressure timing, screw displacement timing, and mold temperature timing; the molding stage is used to indicate whether the corresponding data belongs to the filling stage, holding pressure stage, or cooling stage; the sensor acquisition location is used to indicate the detection location where the corresponding process response data is obtained. The same identifier is written into the cycle data record along with the process response data and maintains the corresponding relationship when generating cycle state characteristics and cross-cycle residual amounts.
[0103] After identifying the mold degradation indicator, the cross-cycle residual amount that continues in the same direction in the continuous injection cycle is read and grouped according to the process response data type, molding stage and sensor acquisition location. Grouping means that the cross-cycle residual amounts with the same three indicators are divided into the same data group, and the cross-cycle residual amounts of different molding stages or different sensor acquisition locations are not merged.
[0104] Check each data group in the order of the injection cycle. When the cross-cycle residual amount in the same data group maintains the same direction of change in consecutive injection cycles and continues until the injection cycle that determines the mold degradation indicator, the molding stage corresponding to the data group is determined as the abnormal molding stage, and the corresponding sensor acquisition position is determined as the abnormal acquisition position. If multiple data groups meet the above conditions, the abnormal molding stage and abnormal acquisition position corresponding to each data group are retained respectively.
[0105] Mold maintenance data is generated based on the abnormal molding stage and abnormal acquisition location. The mold maintenance data includes at least the process response data type corresponding to the mold degradation indicator, the abnormal molding stage, the abnormal acquisition location, and the quality-related cycle range. It is also associated with the injection cycle that determines the mold degradation indicator and saved accordingly. The mold maintenance data is used to indicate the molding stage and inspection location that need to be checked during maintenance, and does not directly identify the abnormal acquisition location as a specific failed component.
[0106] Simultaneously, attribution-related samples are searched based on the quality-related cycle range to form the current candidate parameter adjustment data, and a mold degradation sample mark is written for the attribution-related samples. The mold degradation sample mark indicates that the quality deviation in the sample includes the influence of the mold degradation state. When executing step S6 in the subsequent step, it is first checked whether the attribution-related samples have the mold degradation sample mark. Attribution-related samples with this mark continue to be stored in the historical production record, but do not participate in the reference cycle matching, candidate adjustment dimension determination, and candidate parameter adjustment data generation.
[0107] For example: when the cavity pressure at a certain sensor location during the holding pressure stage corresponds to the cross-cycle residual amount that continues in the same direction throughout the continuous injection molding cycle, the holding pressure stage is identified as an abnormal molding stage, the sensor location of the cavity pressure is identified as an abnormal acquisition location, and the attribution-related samples within the corresponding quality-related cycle range are marked as mold degradation samples.
[0108] Example 2:
[0109] like Figure 2 As shown, this embodiment further improves the design based on Embodiment 1. The difference is that in the actual operation of Embodiment 1, it was found that when the sensor acquisition channel corresponding to the process response data experiences a reference offset, response ratio change, or acquisition timing offset, the cross-cycle residual amount formed by the process response data may also continue in the same direction throughout consecutive injection molding cycles. This leads to a compensation trend similar to that of acquisition channel abnormalities and mold degradation, failing to distinguish between the cross-cycle continuation of the actual molding state and the data offset generated by a single sensor acquisition channel before generating mold maintenance data. Based on this, a process parameter self-optimization method for precision injection molding production further includes: after forming the cross-cycle residual amount in step S3 and before determining the quality-related cycle range in step S4, performing consistency verification on the cross-cycle residual amount based on the stage transmission relationship between different process response data in historical qualified cycles, forming a reliable residual amount or acquiring an abnormal residual amount, and using only the reliable residual amount to form the cross-cycle state sequence.
[0110] Specifically, from the historical qualified cycles where the process setting data is continuous and the product quality data meets the quality benchmark data, the cycle state characteristics corresponding to the cavity pressure timing, screw displacement timing and mold temperature timing are read. According to the process response data type, molding stage, sensor acquisition location and the arrangement position of cycle state characteristics, the cycle state characteristics with continuous sequential relationship in the same injection cycle and in adjacent injection cycles are matched.
[0111] When the periodic state characteristics of one process response data change, and another process response data continues to show corresponding changes in the subsequent position of the current molding stage, the next molding stage, or the next injection cycle, the former is identified as the source state characteristic, and the latter is identified as the supporting state characteristic. The data type, molding stage, sensor acquisition location, and change direction of the source state characteristic are associated and saved with the data type, molding stage, sensor acquisition location, change direction, and the temporal sequence relationship between the two, forming a stage transmission relationship.
[0112] For example: In a historical qualified cycle, if the screw displacement timing changes forward during the filling stage, and the cavity pressure timing changes upward in the subsequent timing position of the filling stage, then the screw displacement change and cavity pressure change are recorded as a set of stage transmission relationships. Similarly, if the cavity pressure state changes during the holding stage, and the mold temperature timing changes accordingly in the later part of the holding stage or the cooling stage, a stage transmission relationship can also be formed.
[0113] For each cross-cycle residual amount obtained in step S3, according to its process response data type, molding stage, sensor acquisition location and change direction, the corresponding support state characteristics are searched from the stage transfer relationship, and in the current injection molding cycle or the subsequent injection molding cycle determined according to the stage transfer relationship, it is checked whether there is a corresponding support cross-cycle residual amount.
[0114] When the order of appearance of the source cross-cycle residual and the supporting cross-cycle residual conforms to the stage transmission relationship, and the direction of change of the two maintains the corresponding relationship in the historical qualified cycle, the source cross-cycle residual and the supporting cross-cycle residual are determined as reliable residuals. The reliable residuals are connected in time sequence according to the method of Example 1, and are used to determine the quality association cycle range, construct the residual transmission chain, and calculate the cycle attribution coefficient.
[0115] When a cross-cycle residual quantity appears continuously only in the same process response data type and the same sensor acquisition location, while the supporting cross-cycle residual quantity that should appear according to the stage transmission relationship does not appear, further compare the corresponding product quality data. If the product quality data does not change along the quality change direction corresponding to the cross-cycle residual quantity in the historical qualified cycle, the cross-cycle residual quantity is determined as an abnormal acquisition residual quantity.
[0116] For abnormal residual amounts collected, they are not written into the cross-cycle state sequence, are not used to determine the quality-related cycle range, and are not used to form mold degradation indicators. The controller writes the corresponding process response data type, molding stage, sensor acquisition location, and abnormal duration cycle for the collected abnormal residual amount to form the acquisition verification data.
[0117] In subsequent injection molding cycles, the remaining process response data types continue to be used to form cycle state characteristics and reliable residual amounts. When the data re-acquired at the sensor acquisition location is restored to correspond with the stage transfer relationship, the data corresponding to that sensor acquisition location is allowed to participate in the formation of cross-cycle residual amounts again. Before restoration, the data corresponding to that sensor acquisition location is retained in the cycle data record, but does not participate in the quality deviation allocation and candidate parameter adjustment data generation.
[0118] When the same cross-cycle residual amount is supported by other process response data types at the same time, and the subsequent product quality data changes according to the corresponding relationship, it is not judged as an abnormal residual amount, but is treated as a real change in molding state and the steps S4 to S8 of Example 1 are continued.
[0119] For example: when the residual amount of cavity pressure corresponding to a certain sensor acquisition location continuously increases in a continuous injection cycle, while the screw displacement timing and mold temperature timing do not show corresponding changes in the stage transmission relationship, and the product quality data does not show corresponding shifts, the residual amount of cavity pressure corresponding to the cross-cycle is determined as an abnormal residual amount; when at least two types of process response data of cavity pressure, screw displacement or mold temperature change continuously according to the stage transmission relationship, and are accompanied by corresponding changes in product quality data, the relevant residual amount of cross-cycle is determined as a reliable residual amount.
[0120] Example 3:
[0121] like Figure 3 As shown, based on the same inventive concept as the process parameter self-optimization method for precision injection molding production in the foregoing embodiments, this application provides a process parameter self-optimization system for precision injection molding production. The system and method embodiments in this application are based on the same inventive concept. The system includes:
[0122] Cycle data acquisition module: Collects process setting data, process response data and product identification according to the injection molding cycle, and determines quality benchmark data from historical qualified cycles;
[0123] State feature construction module: Extracts the periodic state features of process response data, and establishes a set response reference based on the relationship between changes in process setting data and changes in periodic state features in historical qualified cycles;
[0124] Cross-cycle state module: Differentiate the cycle state characteristics of adjacent injection cycles, separate the instantaneous response components according to the set response reference, form cross-cycle residual quantity with the remaining state changes, and connect them in time sequence to form a cross-cycle state sequence.
[0125] Quality cycle association module: acquires product quality data including inspection time stamps and product identifiers, determines the production cycle based on the product identifiers, and determines the quality association cycle range based on the inspection time stamps and cross-cycle status sequences;
[0126] Quality attribution calculation module: Based on the transmission direction of residual amount across cycles and the cycle sequence relationship within the quality correlation cycle range, the quality deviation is allocated to obtain the cycle attribution coefficient;
[0127] Parameter adjustment generation module: Combines process setting data and quality deviations of each associated cycle according to the cycle attribution coefficient to form attribution correlation samples and generate candidate parameter adjustment data;
[0128] Compensation trend discrimination module: Based on the immediate response components after candidate parameter adjustment, cross-cycle residual amount and subsequent product quality data, compensation trend data is generated; when the process disturbance judgment condition is met, a process disturbance identifier is generated; when the mold degradation judgment condition is met, a mold degradation identifier is generated.
[0129] Optimize the maintenance execution module: update the process setting data for subsequent cycles based on the process disturbance flag; stop updating and generate mold maintenance data based on the mold degradation flag.
[0130] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.
[0131] The above description is merely a preferred embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present application, based on the technical solution and concept of the present application, should be covered within the scope of protection of the present application.
Claims
1. A method for self-optimizing process parameters in precision injection molding production, characterized in that, The methods include: Collect process setting data, process response data and product identification data according to the injection molding cycle, and determine quality benchmark data from historical qualified cycles; Extract the periodic state characteristics of the process response data, and establish a setting response reference based on the relationship between the changes in process setting data and the changes in periodic state characteristics of historical qualified cycles; The periodic state characteristics of adjacent injection molding cycles are differentially analyzed, and the instantaneous response components are separated according to the set response reference. The remaining state changes form the cross-cycle residual amount, and the components are connected in time sequence to form a cross-cycle state sequence. Obtain product quality data including detection time stamps and product identifiers, determine the production cycle based on the product identifiers, and determine the quality-related cycle range based on the detection time stamps and cross-cycle state sequences; Based on the transmission direction of the residual amount across cycles and the sequential relationship of cycles within the aforementioned quality correlation cycle, the quality deviation is allocated to obtain the cycle attribution coefficient; Based on the aforementioned cycle attribution coefficients, process setting data and quality deviations for each associated cycle are combined to form attribution association samples, and candidate parameter adjustment data are generated. Compensation trend data is generated based on the immediate response components after adjusting the candidate parameters, the residual amount across cycles, and the quality data of subsequent products. A process disturbance flag is generated when the process disturbance judgment condition is met, and a mold degradation flag is generated when the mold degradation judgment condition is met. Update the process setting data for subsequent cycles based on the process disturbance flag; Based on the mold degradation indicator, updates are stopped and mold maintenance data is generated.
2. The method for self-optimization of process parameters for precision injection molding production according to claim 1, characterized in that, The process setting data includes injection stage setting data, holding pressure stage setting data, and cooling stage setting data; the process response data includes at least two of the following: cavity pressure timing, screw displacement timing, and mold temperature timing; based on the injection molding machine's stage switching signal, the process response data is divided into filling stage data segments, holding pressure stage data segments, and cooling stage data segments, and each data segment is time-aligned, and the change direction, change rate, and stage connection state of each data segment are combined into the periodic state characteristics.
3. The method for self-optimization of process parameters for precision injection molding production according to claim 2, characterized in that, A setting response reference is established based on the correspondence between changes in process setting data and changes in cycle state characteristics during the historical qualified cycles; the cycle state characteristics of adjacent injection cycles are differentially processed to obtain the state change amount; according to the setting response reference, the instantaneous response component corresponding to the changes in process setting data is separated from the state change amount to obtain the cross-cycle residual amount; the cross-cycle state sequence is established based on the directional consistency and temporal continuity between the cross-cycle residual amounts of consecutive injection cycles.
4. The method for self-optimization of process parameters for precision injection molding production according to claim 3, characterized in that, A cycle timescale is configured for each injection molding cycle. Based on the interval between the detection timescale and the cycle timescale of the production cycle, the temporal correspondence between the product quality data and the production cycle is verified. Starting from the production cycle, injection molding cycles in which the cross-cycle residual amount maintains directional continuity are selected sequentially forward along the cross-cycle state sequence to form an initial association range. The positions where the directional continuity of the cross-cycle residual amount is interrupted and the non-continuous switching positions of the process setting data are determined respectively, and the position closer to the production cycle among the two positions is determined as the boundary of the quality association cycle range. When only one of the positions exists, the boundary is determined by that position. When neither of the two positions exists, the boundary is determined by the starting position of the initial association range.
5. The method for self-optimization of process parameters for precision injection molding production according to claim 4, characterized in that, Within the quality correlation cycle, according to the arrangement of process response data type, forming stage, and cycle state characteristics, the corresponding cross-cycle residual amounts in consecutive correlation cycles are connected sequentially to form a residual transfer chain. Taking the quality deviation direction corresponding to the production cycle as the end constraint, cross-cycle residual amounts corresponding to the quality deviation direction are selected forward along the residual transfer chain. Based on the continuation position of the cross-cycle residual amounts in the residual transfer chain of each correlation cycle, and the changing relationship of cross-cycle residual amounts between adjacent correlation cycles, the quality deviation allocation amount corresponding to each correlation cycle is determined. Based on the corresponding ratio between each quality deviation allocation amount and the quality deviation, the cycle attribution coefficient is obtained.
6. The method for self-optimization of process parameters for precision injection molding production according to claim 5, characterized in that, The associated cycle with the highest cycle attribution coefficient, and whose corresponding cross-cycle residual change direction corresponds to the quality deviation, is determined as the primary attribution cycle. The process setting data, cycle status characteristics, and quality deviation assigned to the primary attribution cycle are combined to form the attribution association sample. Using the cycle status characteristics of the primary attribution cycle as the matching condition, a reference cycle with corresponding cycle status characteristics and different process setting data is selected from the historical qualified cycles. The differences in process setting data and quality status between the primary attribution cycle and the reference cycle are compared to determine the candidate adjustment dimensions and candidate adjustment directions of the process setting data, forming the candidate parameter adjustment data.
7. The method for self-optimization of process parameters for precision injection molding production according to claim 6, characterized in that, The expected response direction corresponding to the candidate parameter adjustment data is determined according to the set response reference; the instantaneous response component obtained after using the candidate parameter adjustment data is compared with the expected response direction, and the continuation direction of the cross-cycle residual amount and the quality deviation change of the subsequent product quality data are recorded in the order of the injection molding cycle to form the compensation trend data; when the instantaneous response component corresponds to the expected response direction, the cross-cycle residual amount does not continue continuously in the direction opposite to the expected response direction, and the absolute value of the quality deviation decreases, the process disturbance indicator is generated; When the instantaneous response component corresponds to the expected response direction, the cross-cycle residual amount continues continuously in the direction opposite to the expected response direction, and after continuously executing the candidate parameter adjustment data, the absolute value of the quality deviation gradually decreases relative to the previous injection cycle, the residual amount of the mold degradation indicator generated still continues continuously in the direction opposite to the expected response direction, and after continuously executing the candidate parameter adjustment data, the absolute value of the quality deviation gradually decreases relative to the previous injection cycle, it is determined that the compensation trend data meets the mold degradation judgment condition, and the state judgment indicator is set as the mold degradation indicator.
8. The method for self-optimization of process parameters for precision injection molding production according to claim 7, characterized in that, After generating the mold degradation identifier, the process response data type, molding stage, and sensor acquisition location that formed the cross-cycle residual amount are read; the cross-cycle residual amounts that continue continuously in the same direction are aggregated according to the process response data type, molding stage, and sensor acquisition location to determine the corresponding abnormal molding stage and abnormal acquisition location; mold maintenance data containing the abnormal molding stage, abnormal acquisition location, process response data type, and quality correlation cycle range is generated; the attribution correlation samples corresponding to the quality correlation cycle range are marked as mold degradation samples, and the use of the mold degradation samples for the generation of subsequent candidate parameter adjustment data is stopped.
9. The method for self-optimization of process parameters for precision injection molding production according to claim 3, characterized in that, Also includes: Based on the stage transmission relationship of response data from different processes in historical qualified cycles, the consistency of the cross-cycle residual amount is verified. Cross-cycle residual quantities that support the stage transmission relationship and whose quality deviation direction is consistent are identified as reliable residual quantities, and only the reliable residual quantities are used to form the cross-cycle state sequence; cross-cycle residual quantities that do not have corresponding support for cross-cycle residual quantities and whose corresponding product quality data do not change accordingly are marked as abnormal residual quantities.
10. A process parameter self-optimization system for precision injection molding production, characterized in that, The system includes: Cycle data acquisition module: Collects process setting data, process response data and product identification according to the injection molding cycle, and determines quality benchmark data from historical qualified cycles; State feature construction module: Extracts the periodic state features of the process response data, and establishes a set response reference based on the relationship between the changes in process setting data and the changes in periodic state features in historical qualified cycles; Cross-cycle state module: Differentiate the cycle state characteristics of adjacent injection cycles, separate the instantaneous response components according to the set response reference, form cross-cycle residual quantity with the remaining state changes, and connect them in time sequence to form a cross-cycle state sequence. Quality cycle association module: acquires product quality data including detection time stamps and product identifiers, determines the production cycle based on the product identifiers, and determines the quality association cycle range based on the detection time stamps and cross-cycle state sequences; Quality attribution calculation module: Based on the transmission direction and cycle sequence relationship of the residual amount across cycles within the quality correlation cycle range, the quality deviation is allocated to obtain the cycle attribution coefficient; Parameter adjustment generation module: Combines the process setting data and quality deviation of each associated cycle according to the cycle attribution coefficient to form attribution association samples and generate candidate parameter adjustment data; Compensation trend discrimination module: Compensation trend data is generated based on the immediate response components after adjusting the candidate parameters, the residual amount across cycles, and the quality data of subsequent products; A process disturbance flag is generated when the process disturbance judgment condition is met, and a mold degradation flag is generated when the mold degradation judgment condition is met. The optimization and maintenance execution module updates the subsequent cycle process setting data based on the process disturbance flag; it stops updating and generates mold maintenance data based on the mold degradation flag.
