Search

CN-121996946-A - Hardware forming defect prediction method and system based on stamping parameters

CN121996946ACN 121996946 ACN121996946 ACN 121996946ACN-121996946-A

Abstract

The invention discloses a hardware forming defect prediction method and a hardware forming defect prediction system based on stamping parameters, and particularly relates to the technical field of hardware forming defect prediction, wherein the hardware forming defect prediction method comprises the steps of obtaining stamping parameters corresponding to each stroke frequency and forming response data corresponding to the stamping parameters in a hardware forming process, combining the stamping parameters and the forming response data to form a stroke frequency data item, and writing the stroke frequency data item into a cache queue for edge calculation; and acquiring stamping parameters and forming response data in real time in the continuous times and constructing a separability index reflecting the sensitivity of the material to the parameter change, and judging whether the deformation freedom degree of the material contracts according to the continuous descending trend of the separability index so as to identify the state that the material is forced to enter an irreversible failure path before the occurrence of a dominant defect.

Inventors

  • YE CHAOYI
  • LUO JIANLUN
  • ZHANG XIAOMEI
  • CHEN YING

Assignees

  • 佛山市顺德区百仕盛五金实业有限公司

Dates

Publication Date
20260508
Application Date
20251225

Claims (9)

  1. 1. The hardware forming defect prediction method based on the stamping parameters is characterized by comprising the following steps of: s1, obtaining stamping parameters corresponding to each stroke frequency and forming response data corresponding to the stamping parameters in a hardware forming process, combining the stamping parameters and the forming response data to form a stroke frequency data item, and writing the stroke frequency data item into a cache queue of edge calculation; s2, reading a plurality of continuous impulse data items from a cache queue calculated by the edge according to the impulse sequence, and calculating the difference value of stamping parameters in adjacent impulse data items to obtain parameter change data; s3, based on the parameter change data, grouping processing is carried out on the impulse data items according to the magnitude of the parameter change amplitude, the impulse data items with the parameter change amplitude in different change intervals are divided into at least two impulse data subsets with different parameter change conditions, and response feature extraction is carried out on the formed response data in each impulse data subset respectively, so that a corresponding response feature set is obtained; S4, executing the separability calculation on the response characteristic sets corresponding to the different stroke frequency data subsets, and forming a separability index by calculating the difference degree of the response characteristics among the different stroke frequency data subsets and the discrete degree of the response characteristics inside each stroke frequency data subset; S5, carrying out change trend analysis on the separable index in a plurality of continuous calculation periods, judging that the response capability of the material to the change of the stamping parameters in the current hardware forming process is contracted when the separable index is in a continuously descending state in the plurality of continuous calculation periods and reaches a preset judging condition, and marking the corresponding stroke frequency data item as a hardware forming defect prediction result and outputting the result.
  2. 2. The hardware forming defect prediction method based on stamping parameters according to claim 1, wherein the hardware forming defect prediction method based on stamping parameters is characterized in that: In S1, the process of writing into the cache queue of edge computation includes: s1-1, collecting stamping parameters corresponding to each punching frequency through a signal collecting interface of an edge computing side and stamping equipment, recording sampling time information of each stamping parameter during collection, and taking the stamping parameters and the corresponding sampling time information as original sampling data of the stamping parameters; S1-2, in a preset time window range of the punching frequency, carrying out continuous sampling on load measurement signals and slide block position measurement signals corresponding to punching parameters, organizing the load measurement signals obtained by continuous sampling into a load curve sampling sequence according to time sequence, organizing the slide block position measurement signals obtained by continuous sampling into a slide block position curve sampling sequence according to time sequence, and recording sampling time information corresponding to each sampling point to be used as forming response original sampling data; S1-3, based on a time window of the current time of the stroke, associating punching parameter original sampling data and forming response original sampling data with sampling time falling into the same time window of the stroke to the same time of the stroke, and generating a time mark corresponding to the current time of the stroke; s1-4, taking a stroke frequency identifier as an index, respectively gathering punching parameter original sampling data and forming response original sampling data which are related to the same stroke frequency identifier in a corresponding stroke frequency time window according to a corresponding relation of the stroke frequency identifier, and forming a punching parameter data set and a forming response data set; S1-5, taking a time window of a stroke frequency as a time range, sequentially generating a unified time sequence covering the time window of the stroke frequency according to a preset time interval by taking the starting time of the time window of the stroke frequency as a starting point based on sampling time information of each sampling point; S1-6, packaging the aligned stamping parameter data and the forming response data according to a unified data structure to generate a stroke frequency data item, sequentially writing the stroke frequency data item into a pre-allocated cache area in an edge computing node working memory according to the sequence of sampling time information, and sequentially inserting the stroke frequency data item into a cache queue according to the writing sequence.
  3. 3. The hardware forming defect prediction method based on stamping parameters according to claim 2, wherein the hardware forming defect prediction method based on stamping parameters is characterized in that: in S2, the process of obtaining parameter variation data includes: S2-1, in an edge computing node, based on a buffer queue buffered in a working memory, taking the enqueuing sequence of the impulse data items in the buffer queue as an impulse sequence basis, sequentially reading a plurality of impulse data items which are continuously arranged from the head of the queue, and temporarily storing each read impulse data item in a sequence buffer area; S2-2, based on the stroke frequency identification fields carried by each stroke frequency data item in the sequence buffer, executing sequence continuity check on the stroke frequency identification fields, judging whether the stroke frequency identifications of two adjacent stroke frequency data items are continuously arranged according to an increasing rule, and eliminating the stroke frequency data items with discontinuous stroke frequency identifications; s2-3, respectively extracting the aligned punching parameter data in each punching data item based on the punching parameter field and the sampling time index field carried in the punching data item in the sequence buffer, and writing the punching parameter data into the punching parameter sequence list according to the punching sequence.
  4. 4. A method for predicting a hardware forming defect based on stamping parameters according to claim 3, wherein: in S2, the process of obtaining parameter variation data further includes: s2-4, taking stamping parameter data corresponding to any two adjacent punching times in the stamping parameter sequence list as a group, and based on the sampling time index field, checking each sampling time index position of the two adjacent stamping parameter data in the stamping parameter sequence list on a unified time sequence one by one to judge whether corresponding numerical values exist in the two stamping parameter data at each sampling time index position; Judging that the alignment judgment is passed only when all sampling time index positions in the unified time sequence meet that corresponding values exist in two pieces of punching parameter data at the same time, otherwise judging that the alignment judgment is not passed, and eliminating adjacent punching times corresponding to the two pieces of punching parameter data from the current difference value calculation object; S2-5, on the premise that alignment judgment is passed, two pieces of stamping parameter data of adjacent times are taken as objects, difference calculation is carried out for each sampling time index position, and the stamping parameter value of the sampling time index position corresponding to the next time is subtracted from the stamping parameter value of the sampling time index position corresponding to the previous time, so that a stamping parameter difference value entry list corresponding to the sampling time index one by one is formed; S2-6, organizing all stamping parameter difference value entries corresponding to the same adjacent stroke frequency into a parameter difference value sequence according to a sampling time index sequence, and writing each parameter difference value sequence into a parameter difference value sequence list according to the stroke frequency sequence to serve as parameter change data.
  5. 5. The hardware forming defect prediction method based on stamping parameters according to claim 4, wherein the hardware forming defect prediction method based on stamping parameters is characterized in that: In S3, the process of obtaining the corresponding response feature set includes: s3-1, based on the parameter change data and the corresponding stroke frequency identification, associating the parameter difference value sequence of each stroke frequency with the corresponding stroke frequency identification to form a parameter change record list arranged according to the sequence of the stroke frequency; S3-2, aiming at each record in the parameter change record list, in a parameter difference value sequence corresponding to the current record, absolute value operation is carried out on the stamping parameter difference value of each sampling time index position, the upper limit value in the absolute value is selected in each parameter difference value sequence as a parameter change amplitude value corresponding to the stroke frequency, and the parameter change amplitude value and the corresponding stroke frequency identification are written into a parameter change amplitude mapping table; S3-3, presetting a plurality of non-overlapping parameter variation amplitude intervals, wherein each parameter variation amplitude interval is determined by a pair of parameter variation amplitude lower limit values and parameter variation amplitude upper limit values, and based on parameter variation amplitude values in a parameter variation amplitude mapping table, comparing the parameter variation amplitude values corresponding to each impulse with the plurality of parameter variation amplitude intervals one by one, judging the parameter variation amplitude interval in which the current parameter variation amplitude value falls, classifying the impulse identifications falling into the same parameter variation amplitude interval into the same group, and forming impulse identification groups with different parameter variation amplitude intervals.
  6. 6. The method for predicting hardware forming defects based on stamping parameters according to claim 5, wherein the method comprises the following steps: in S3, the process of obtaining the corresponding response feature set further includes: s3-4, taking the stroke frequency identification packet as an index, reading corresponding stroke frequency data items from a cache queue of the edge computing node according to the stroke frequency identification, and gathering the stroke frequency data items belonging to the same stroke frequency identification packet into a corresponding stroke frequency data subset; S3-5, respectively extracting aligned forming response data in forming response fields from each of the stroke frequency data items in the corresponding stroke frequency data subsets according to the stroke frequency sequence and the sampling time index sequence, and organizing the extracted forming response data into a response data matrix, wherein a row index of the response data matrix corresponds to the stroke frequency identification, and a column index corresponds to the sampling time index position on the uniform time sequence; S3-6, based on a response data matrix corresponding to each impulse data subset, performing response feature extraction processing on each line of formed response data to obtain a plurality of response feature quantities, wherein the plurality of response feature quantities comprise peak values, peak value occurrence time, rising stage slopes and curve integral values in a preset time window of a formed response on a unified time sequence, combining the plurality of response feature quantities into response feature vectors corresponding to the current impulse, and combining the response feature vectors corresponding to each impulse in the same impulse data subset according to the impulse sequence to form a response feature set corresponding to the impulse data subset; Traversing forming response data of the current stroke frequency at all sampling time index positions corresponding to the uniform time sequence, comparing forming response values of all sampling time index positions, selecting a sampling point at an upper limit value from the forming response values, and taking the forming response value corresponding to the sampling point as a forming response peak value of the current stroke frequency; After determining the sampling time index position corresponding to the forming response peak value, reading the time value in the unified time sequence corresponding to the sampling time index position, and taking the corresponding time value as the forming response peak value appearance time; The rising stage slope is that in the forming response data of the current stroke frequency, the initial sampling time index position on the unified time sequence is taken as the rising stage starting point, the sampling time index position corresponding to the forming response peak value is taken as the rising stage end point, the forming response values of the starting point and the end point and the corresponding time values are respectively read, the ratio of the forming response value increment to the time increment is calculated, and the corresponding ratio is taken as the rising stage slope of the current stroke frequency; The curve integral value in the preset time window is that each sampling time index position between the starting time and the ending time corresponding to the preset time window is selected on the unified time sequence, the forming response value of each sampling time index position is obtained according to the sampling time index sequence, the time difference between the adjacent sampling time index positions is used as an integral step length by adopting a preset time interval, trapezoidal approximate area calculation is carried out on the forming response value corresponding to each adjacent sampling time index position, the trapezoidal approximate area calculation is accumulated section by section in the preset time window, and the accumulated area sum is used as the forming response curve integral value of the current impulse in the preset time window.
  7. 7. The hardware forming defect prediction method based on stamping parameters according to claim 6, wherein the hardware forming defect prediction method based on stamping parameters is characterized in that: In S4, the process of forming the separability index includes: s4-1, taking response feature vectors arranged according to the order of the times of the strokes in the response feature set as objects, and writing the response feature vectors into a response feature vector order list; s4-2, calculating the absolute value of the numerical value difference between the response characteristic vector elements by taking any two different impulse data subsets as objects, based on a corresponding response characteristic vector sequence list and taking the response characteristic quantity of the same index position as a comparison object, and writing the absolute value of the difference on all the index positions into a cross-subset difference sequence in sequence; S4-3, selecting a difference absolute value with a numerical value at an upper limit position in a cross-subset difference sequence, and taking the corresponding difference absolute value as an upper limit difference degree between current two impulse data subsets; s4-4, taking any one of the stroke frequency data subsets as an object, taking response characteristic quantities of the same index positions as comparison objects in a response characteristic vector sequence list corresponding to the corresponding stroke frequency data subset, calculating numerical value difference absolute values among different response characteristic vector elements, and writing the difference absolute values corresponding to the index positions into a subset internal difference sequence in sequence; S4-5, selecting a difference absolute value with a numerical value at an upper limit position from a difference sequence in the subset, and taking the corresponding difference absolute value as an upper limit dispersion in the current impulse data subset; S4-6, taking the value of the upper limit difference degree crossing the subsets as a subtracted number, taking the upper limit value in the upper limit dispersion degree values inside all the impulse data subsets as a subtracted number, performing numerical subtraction to obtain a difference value, and performing absolute value operation on the difference value, wherein the obtained value is used as a separability index.
  8. 8. The hardware forming defect prediction method based on stamping parameters according to claim 7, wherein the hardware forming defect prediction method based on stamping parameters is characterized in that: In S5, the process of outputting the hardware forming defect prediction result includes: S5-1, based on a calculation result of the separable indexes, sequentially writing the separable indexes sequentially generated in the continuous calculation period into a separable index sequence according to the calculation period sequence; S5-2, taking any two separable indexes in adjacent sequence index positions in the separable index sequence as objects, executing numerical difference operation, subtracting the numerical value of the former separable index from the numerical value of the latter separable index to obtain the numerical value difference of the separable index in the adjacent calculation period, and writing the obtained numerical value difference into a separable index difference sequence; S5-3, in the sequence of the differential values of the separable indexes, judging all the numerical differences, judging that the numerical difference is smaller than zero as a descending difference, judging that the numerical difference is larger than zero as an ascending difference, judging that the numerical difference is equal to zero as an leveling difference, and writing each judging result into a change judging sequence; S5-4, in the change judging sequence, according to the sequence index order, carrying out continuity judgment on judging results on adjacent index positions, wherein the judging results corresponding to the current index position and the next index position are all descending difference values, forming a continuous descending section from the current index position to the next index position, continuing to recursively judge whether the same descending state is met between the next index position and the next index position, and connecting a plurality of continuous descending sections of the adjacent index positions continuously meeting the descending state to form a descending section; S5-5, comparing the sequence section length of each descending segment in the descending segment set, taking the descending segment with the sequence section length reaching or exceeding a preset descending judgment threshold as the descending segment meeting the descending continuous condition, and defining the corresponding partitional index sequence section as the partitional index section reaching the descending continuous condition; S5-6, positioning a stroke frequency identification corresponding to a calculation period based on the index position of the separable index sequence corresponding to the separable index section reaching the descending continuous condition, judging that the response capability of the material to the change of the punching parameters in the corresponding stroke frequency period is contracted, and marking the corresponding stroke frequency data item as a hardware forming defect prediction result.
  9. 9. The hardware forming defect prediction system based on the stamping parameters comprises a hardware forming defect prediction method based on the stamping parameters, and is characterized by comprising a punching frequency acquisition module, a change calculation module, a characteristic grouping module, an indexable quantity module and a shrinkage early-warning module, wherein the hardware forming defect prediction method based on the stamping parameters is characterized in that: the punching frequency acquisition module is used for acquiring punching parameters corresponding to each punching frequency and forming response data corresponding to the punching parameters in the hardware forming process, combining the punching parameters and the forming response data to form punching frequency data items, and writing the punching frequency data items into a cache queue for edge calculation; The change calculation module is used for reading a plurality of continuous punching frequency data items according to the punching frequency sequence from the cache queue calculated by the edge, and calculating the difference value of punching parameters in adjacent punching frequency data items to obtain parameter change data representing the change condition of the punching parameters among the continuous punching frequencies; The characteristic grouping module is used for performing grouping processing on the impulse data items according to the parameter variation amplitude based on the parameter variation data, dividing the impulse data items with the parameter variation amplitude in different variation intervals into different impulse data subsets, and respectively performing response characteristic extraction on the formed response data in each impulse data subset to obtain a corresponding response characteristic set; The indexable quantity module is used for executing the separable calculation on the response characteristic sets corresponding to the different stroke frequency data subsets, and forming a separable index by calculating the difference degree of the response characteristics among the different stroke frequency data subsets and the discrete degree of the response characteristics in each stroke frequency data subset; The shrinkage early warning module is used for executing change trend analysis on the separable index in a plurality of continuous calculation periods, judging that the response capability of the material to the change of the stamping parameters in the current hardware forming process is shrunk when the separable index is in a continuous descending state in the plurality of continuous calculation periods and reaches a preset judging condition, and marking the corresponding stroke frequency data item as a hardware forming defect prediction result and outputting the result.

Description

Hardware forming defect prediction method and system based on stamping parameters Technical Field The invention relates to the technical field of hardware forming defect prediction, in particular to a hardware forming defect prediction method and system based on stamping parameters. Background In the existing hardware stamping forming defect prediction technology, whether forming results have failure phenomena such as cracks, wrinkles or corners collapse or not is generally taken as a judgment target, the prediction process mainly depends on abnormal changes of stamping parameters, equipment running states and forming response signals to identify risks in advance, and technically and logically default materials always have adjustable deformation spaces in the whole forming process, so that the forming results still have the possibility of being corrected as long as the parameters are still in reasonable intervals; However, in actual stamping production, plastic flow of the material in the die is gradually limited along with repeated forming, local constraint enhancement and continuous accumulation of strain, when the deformation coordination capacity in the material is weakened to a certain degree, the subsequent deformation path is fixed by a process structure and a historical strain state, even if no detectable defect characteristic exists, the material cannot be restored to a stable forming state through parameter fine adjustment, and the occurrence of the subsequent defect in the state has certainty but not accidents; Because the existing prediction method only pays attention to whether a single forming result approaches a failure threshold value, the continuous characterization capability of the evolution process of the deformation degree of freedom of the material is lacking, and whether the material is forced to enter an irreversible failure path under the normal parameter condition cannot be identified, the prediction result cannot provide effective early warning before the defect really appears; Therefore, the current hardware forming defect prediction technology cannot process and judge the forming process from the angle of material deformation path selectivity, and cannot recognize the state that the forming process inevitably goes to failure in advance. Disclosure of Invention In order to overcome the defects in the prior art, the embodiment of the invention provides a hardware forming defect prediction method and a hardware forming defect prediction system based on punching parameters, which are characterized in that punching parameters and forming response data are collected in real time in continuous punching times, a separable index reflecting the sensitivity of a material to parameter change is constructed, and whether the deformation freedom degree of the material is contracted is judged according to the continuous descending trend of the separable index so as to identify the state that the material is forced to enter an irreversible failure path before the occurrence of a dominant defect. In order to achieve the purpose, the invention provides the following technical scheme that the hardware forming defect prediction method based on stamping parameters comprises the following steps: s1, obtaining stamping parameters corresponding to each stroke frequency and forming response data corresponding to the stamping parameters in a hardware forming process, combining the stamping parameters and the forming response data to form a stroke frequency data item, and writing the stroke frequency data item into a cache queue of edge calculation; s2, reading a plurality of continuous impulse data items from a cache queue calculated by the edge according to the impulse sequence, and calculating the difference value of stamping parameters in adjacent impulse data items to obtain parameter change data; s3, based on the parameter change data, grouping processing is carried out on the impulse data items according to the magnitude of the parameter change amplitude, the impulse data items with the parameter change amplitude in different change intervals are divided into at least two impulse data subsets with different parameter change conditions, and response feature extraction is carried out on the formed response data in each impulse data subset respectively, so that a corresponding response feature set is obtained; S4, executing the separability calculation on the response characteristic sets corresponding to the different stroke frequency data subsets, and forming a separability index by calculating the difference degree of the response characteristics among the different stroke frequency data subsets and the discrete degree of the response characteristics inside each stroke frequency data subset; S5, carrying out change trend analysis on the separable index in a plurality of continuous calculation periods, judging that the response capability of the material to the change of the stamping parameters in the