Stamping data low-cost acquisition method and process energy spectrum construction method

Abstract

The present application belongs to the field of stamping forming production monitoring, and particularly relates to a low-cost stamping data acquisition method and a process energy spectrum construction method. The low-cost stamping data acquisition method comprises the following steps: S1: acquiring a plurality of discrete sample data in the stamping process through actual stamping test, including measuring stamping force and measuring process energy. S2: constructing a simulation model for simulating the stamping forming process, the simulation model comprising a stamping geometric model and a finite element model. S3: generating simulation stamping data using the simulation model, including simulation process energy and simulation thickness change. S4: calculating the deviation between the simulation process energy and the measured process energy; iteratively correcting the simulation model to obtain the required numerical model. S5: generating simulation stamping data required for constructing the process energy spectrum using the numerical model. The present application solves the problem that it is difficult to accurately and low-costly monitor the processing quality of stamping workpieces on-line under the existing conditions.

Description

Technical Field

[0001] This invention belongs to the field of stamping production monitoring, specifically involving a low-cost method and system for acquiring stamping data, as well as corresponding data processing equipment. It also relates to a method for constructing a process energy spectrum and a method for online monitoring of the processing quality of stamped workpieces. Background Technology

[0002] In stamping processes, quality monitoring has always been a challenge for those skilled in the art. Due to the enclosed nature of stamping dies, stamped workpieces are typically processed rapidly within a closed environment; directly measuring the quality of the stamping process is technically difficult. Existing technologies usually predict the thickness changes of the stamped workpiece during processing based on equipment operating parameters or feedback signals, thereby assessing the quality of the stamped workpiece.

[0003] Based on this, the present invention proposes a method for monitoring the processing quality of stamped workpieces based on process energy maps. This method mainly studies the relationship between thickness changes and process energy and stamping depth during the processing of stamped workpieces. In subsequent quality monitoring, it combines the process energy and stamping depth changes of the stamped workpiece during processing to predict the thickness changes. Then, based on the maximum thickness increase rate and maximum thinning rate of the stamped workpiece, it determines whether wrinkling or cracking has occurred, thereby obtaining the processing quality assessment result of the stamped workpiece.

[0004] The online monitoring method for stamping workpiece processing quality based on process energy mapping boasts high monitoring accuracy, but it also suffers from a significant drawback: the accuracy of the quality monitoring scheme is highly dependent on the scale of the sample data. This scheme requires extensive actual testing of each type of stamping workpiece under specified stamping process conditions to obtain sufficient sample data. This real sample data is then used to create a process energy map and guide actual production. Taking the construction of a process energy map for a car door-like inner panel made of AA5052 aluminum alloy with a stamping depth of 28mm as an example, technicians need to select stamping depths from 1mm to 28mm at 1mm intervals to determine the horizontal axis of the process energy map. With 16 sets of process conditions set for each stamping depth, this requires processing 448 car door-like inner panels and collecting process energy and thickness variation data for each of these 448 panels. Furthermore, if the material type, workpiece shape, structure, or size changes, the technicians need to repeat the above steps. All of these factors significantly increase the construction cost of the process energy map.

[0005] Therefore, it is evident that the data acquisition strategy of obtaining sample data through actual stamping tests requires a stamping process at each stamping depth. After processing, the stamped workpiece also needs to be cut to collect the thickness variation data required for constructing the process energy profile. This results in a significant consumption of human and material resources, greatly increasing the application cost of the solution, leading to excessively long testing cycles and high testing costs. Therefore, those skilled in the art urgently need a low-cost method to obtain a large amount of reliable stamping data to avoid the inefficiency and resource waste caused by actual testing. Summary of the Invention

[0006] To address the problems of low efficiency and high cost associated with traditional methods of obtaining stamping data through actual stamping tests, this invention provides a low-cost method for obtaining stamping data and a method for constructing process energy maps.

[0007] This invention is achieved using the following technical solution:

[0008] A low-cost method for acquiring stamping data is disclosed, which is used to obtain process energy and thickness variation data with mapping relationships required for constructing a process energy profile; the thickness variation data includes the maximum thinning rate and the increase in thickness rate. The low-cost method for acquiring stamping data includes the following steps:

[0009] S1: Under the preset stamping process conditions, obtain several discrete sample data through actual stamping tests. The sample data includes the measured stamping force and the measured process energy.

[0010] S2: Construct a simulation model to simulate the stamping process. The simulation model includes a stamping geometry model and a finite element model. The stamping geometry model reflects the shape, structure, and dimensions of the stamping die and sheet metal. The finite element model simulates the stamping of the sheet metal to obtain the mapping relationship between the thickness variation of the stamped workpiece and the process energy.

[0011] S3: Use the simulation model to generate the simulated maximum thinning rate, simulated maximum thickness increase rate, and simulated process energy.

[0012] S4: Calculate the deviation between the simulated process energy and the measured process energy; then make the following decision:

[0013] A. When the deviation is greater than the correction threshold, the model parameters that require manual input in the stamping geometry model are actually measured, and the actual measured values ​​of the model parameters are input into the stamping geometry model for correction; and the friction coefficient μ in the finite element model is iteratively corrected.

[0014] B. When the deviation is not greater than the correction threshold, save the current simulation model as the numerical model.

[0015] S5: Using numerical models as a tool, generate process energy and thickness variation data with mapping relationships required for constructing process energy maps.

[0016] As a further improvement of the present invention, the friction coefficient correction strategy in step S4 is as follows:

[0017] S01: Initialize the friction coefficient population μ(g) and mutation probability F a Crossover probability CR, maximum number of iterations ger, individual learning factor b1, population learning factor b2, and inertia weight q.

[0018] S02: Calculate the original friction coefficient individual μ in the current friction coefficient population μ(g). i (g) Average absolute percentage error of corresponding process energy e MA,i (g) and identify the individual friction coefficient μ with the smallest mean absolute percentage error. m (g)

[0019] S03: Update the original friction coefficient individual μ in the friction coefficient population μ(g) based on the following formula. i (g) evolutionary direction v i (g+1):

[0020]

[0021] In the above formula, r1 and r2 are random values ​​between 0 and 1; E is the identity matrix; e MA,m (g) represents the average absolute percentage error of the individual with the optimal friction coefficient; μ i (g), μ q (g) and μ r (g) represents three original friction coefficient individuals randomly selected from the friction coefficient population μ(g).

[0022] S04: According to the direction of evolution v i (g+1) Select the original friction coefficient individuals μ from the current friction population μ(g). i (g) Mutation is performed to obtain individual mutant friction coefficients μ. i The formula for mutation operation (g) is as follows:

[0023]

[0024] In the above formula, F a (g+1) represents the mutation probability in the (g+1)th iteration.

[0025] S05: For individuals with abruptly changed friction coefficients in the current friction population μ(g), i '(g) and the original friction coefficient of the individual μ i(g) Perform crossover operations to generate new individuals with different friction coefficients. The strategy for crossover operations is as follows:

[0026]

[0027] In the above formula, rand i It is the original friction coefficient individual μ i (g) and the individual friction coefficient of mutation μ i '(g) A random number between 0 and 1 generated during the crossover operation.

[0028] S06: Comparison of individual friction coefficients with abrupt changes (μ) i '(g) and the original friction coefficient of the individual μ i The mean absolute percentage error of the process energy in (g) is used to update the individual friction coefficients to obtain the friction coefficient population μ(g+1) at iteration g+1:

[0029]

[0030] In the above formula, e MA,i '(g) represents the individual friction coefficient μ of the mutation. i The average absolute percentage error of the process energy corresponding to '(g)'.

[0031] S07: Determine if the current iteration number g is less than the maximum iteration number ger:

[0032] If so, then repeat steps S02 to S06.

[0033] Otherwise, end the iteration and output the individual friction coefficient μ with the smallest mean absolute percentage error of process energy in the current friction coefficient population μ(g). m (g) is the updated coefficient of friction.

[0034] This invention also includes a method for constructing a process energy map, which comprises the following steps:

[0035] 1. Under the preset stamping process conditions, obtain several discrete sample data through actual stamping tests.

[0036] Second, adopt the aforementioned low-cost method for obtaining stamping data, and establish a numerical model based on the collected sample data.

[0037] This numerical model is used to generate an array consisting of the maximum thinning rate, the maximum thickening rate, the stamping depth, and the stamping force, which have a mapping relationship.

[0038] III. Based on the stamping depth h p A blank process energy map is plotted with process energy E as the horizontal and vertical axes, and a boundary-bounded region to be filled is determined based on the process constraints on parameters.

[0039] IV. Using a numerical model to fill in the unfilled areas in the blank process energy map, the process is as follows:

[0040] 4.1: Using a numerical model to generate the associated simulated stamping depth h ps Simulated impact force F s Simulated maximum thickness increase rate Δ MTCs and the simulated maximum thinning rate Δ MTNs The first array U1 is formed as follows: U1 = {h ps F s Δ MTCs Δ MTNs}

[0041] 4.2: Based on the simulated stamping depth h ps and simulated impact force F s Calculate the corresponding simulated process energy E s :

[0042]

[0043] In the above formula, F s (x) represents the simulated impact force F. s Regarding the simulated stamping depth h ps The fitting function;

[0044] 4.3: Convert the first array U1 into the second array U2: U2 = {h ps E s Δ MTCs Δ MTNs}

[0045] 4.4: Using the simulated stamping depth h in the second array ps and simulation process energy E s Using the x and y coordinates of the pixels within the area to be filled as the coordinates, the maximum thickness increase Δ is simulated. MTCs With the simulated maximum thinning rate Δ MTNs These are the first and second attribute values ​​for the corresponding pixels; this completes the pixel filling of the blank process energy spectrum.

[0046] 5. Using the vertical axis as the axis of symmetry, mirror the image filled in the above steps into two symmetrical parts: the wrinkle recognition area and the breakage recognition area.

[0047] 6. According to the preset color mapping relationship, each pixel in the wrinkle recognition area is colored according to the first attribute value, and each pixel in the crack recognition area is colored according to the second attribute value.

[0048] 7. Based on the preset safety threshold, the boundaries between the wrinkled area and the safe area in the wrinkled identification area, and the boundaries between the fractured area and the safe area in the fracture identification area are delineated to obtain the required process energy spectrum.

[0049] This invention also includes a method for online monitoring of the processing quality of stamped workpieces, comprising the following steps:

[0050] Step 1: Obtain the process energy map of the current stamping workpiece to be processed using the process energy map construction method described above.

[0051] Step 2: Real-time acquisition of the stamping depth h of the workpiece to be processed during the actual processing. real and real-time impact pressure F real .

[0052] Step 3: Based on the real-time stamping depth h during the processing... real and real-time impact pressure F real Calculate the real-time process energy E real .

[0053] Step 4: Based on the real-time stamping depth h of the workpiece to be processed during the processing... real and real-time process energy E real The state changes are plotted in the wrinkling and cracking identification areas of the process energy map to create corresponding state trajectories.

[0054] Step 5: Evaluate the machining quality of the stamped workpiece based on the state trajectory:

[0055] (1) When any point in the state trajectory passes through the fracture zone, it is determined that the processed stamped workpiece has fractured.

[0056] (2) When the endpoint of the state trajectory is located in the wrinkled area, it is determined that the processed stamped workpiece has local wrinkling.

[0057] This invention also includes a low-cost stamping data acquisition system, which employs the aforementioned low-cost stamping data acquisition method and combines measurement process energy and simulation model to obtain a numerical model for low-cost stamping data acquisition. The low-cost stamping data acquisition system includes: a measurement process energy acquisition module, a simulation model construction module, an error calculation module, a friction coefficient update module, and a numerical model correction module.

[0058] The process energy measurement module is used to calculate the process energy based on several discrete sample data obtained from the actual stamping test.

[0059] The simulation model building module includes a stamping geometry model and a finite element model. The stamping geometry model reflects the shape, structure, and dimensions of the stamping die and sheet metal. The finite element model simulates the stamping deformation of the sheet metal, thereby obtaining the mapping relationship between the thickness change of the stamped workpiece and the process energy. The simulation model is used to generate the simulated maximum thinning rate, simulated maximum thickness increase rate, and simulated process energy.

[0060] The error calculation module is used to calculate the deviation between the simulated process energy generated by the constructed simulation model and the measured process energy.

[0061] The friction coefficient update module is used to iteratively correct the friction coefficient in the finite element model using the correction strategy described above when the deviation calculated by the error calculation module exceeds the correction threshold.

[0062] The numerical model correction module is used to reset the manually input parameters in the stamping geometry model and update the friction coefficient in the finite element model when the deviation calculated by the error calculation module exceeds the correction threshold. When the deviation calculated by the error calculation module does not exceed the correction threshold, the corresponding parameters in the simulation model are retained to obtain a numerical model that reflects the mapping relationship between the maximum thinning rate, maximum thickening rate and process energy of the stamped workpiece.

[0063] The present invention also includes a data processing device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it creates a low-cost stamping data acquisition system as described above. Then, based on several discrete sample data collected during the stamping test process, it automatically creates a numerical model that can generate a mapping relationship between the maximum thinning rate, the maximum thickening rate, and the process energy of the stamped workpiece under specified process conditions.

[0064] The technical solution provided by this invention has the following beneficial effects:

[0065] The present invention proposes a low-cost method for acquiring stamping data. This method utilizes a high-precision stamping simulation model to obtain a large amount of process energy and thickness variation data, which is then used to construct a process energy map at low cost. This process energy map enables online monitoring of the processing quality of stamped workpieces.

[0066] In the process of constructing and refining the simulation model, to ensure the accuracy and reliability of the simulation model, this invention designs a new model iterative correction strategy and an improved particle swarm optimization (PSO-DE) algorithm for iteratively optimizing the friction coefficient in the finite element model. This invention treats the simulated process energy of the stamping simulation model as a dynamic response, uses the friction coefficient as an optimization parameter, and continuously optimizes the friction coefficient through the improved algorithm. This ensures that the simulated process energy generated by the constructed simulation model is consistent with the measured process energy obtained from actual stamping tests, thereby guaranteeing the accuracy of the final generated process energy and thickness variation data.

[0067] The numerical model obtained from the low-cost stamping data acquisition method provided by this invention can replace the measured process energy and thickness change data obtained from actual stamping tests in the traditional process energy map construction process. Thus, while ensuring similar data accuracy, it significantly reduces the high data acquisition cost generated by actual measurements and improves the practical value of the constructed process energy map. Attached Figure Description

[0068] Figure 1 This is a flowchart of the steps of a low-cost method for obtaining stamping data provided in Embodiment 1 of the present invention.

[0069] Figure 2 This is a flowchart of the steps for obtaining and measuring process energy in Embodiment 1 of the present invention.

[0070] Figure 3 This is a flowchart of the simulation model creation and application process in Embodiment 1 of the present invention.

[0071] Figure 4 This is a schematic diagram illustrating the accuracy evaluation process of the stamping forming simulation model in Embodiment 1 of the present invention.

[0072] Figure 5 This is a flowchart of the steps involved in the iterative correction process of the friction coefficient in the finite element model of Embodiment 1 of the present invention.

[0073] Figure 6 This is a framework diagram of a low-cost stamping data acquisition system provided in Embodiment 2 of the present invention.

[0074] Figure 7 This is a flowchart of the steps involved in constructing a process energy map as provided in Embodiment 4 of the present invention.

[0075] Figure 8 This is a case study diagram illustrating the implementation process of the method for constructing the process energy map in Embodiment 4 of the present invention.

[0076] Figure 9This is a flowchart of the steps of an online monitoring method for the processing quality of stamped workpieces provided in Embodiment 5 of the present invention.

[0077] Figure 10 The simulation process energy and error diagrams for four different combinations of stamping control parameters (CB1, CB2, CB3, CB4) in the performance test experiment are shown.

[0078] Figure 11 This invention compares the accuracy and running speed of the improved PSO-DE algorithm with the traditional PSO algorithm in finding the individual with the optimal friction coefficient.

[0079] in, Figure 11 Part (a) shows the average absolute percentage error of the process energy corresponding to the individual optimal friction coefficients obtained by the two algorithms. Figure 11 Part (b) shows the convergence curves of the two algorithms.

[0080] Figure 12 The simulated thickness variation and error are shown for four different combinations of stamping control parameters (CB1, CB2, CB3, CB4) in the performance test experiment.

[0081] Figure 13 A comparison of process energy maps constructed using three different methods.

[0082] Among them, MAP1 is a process energy spectrum constructed using a large amount of measured stamping data; MAP2 is a process energy spectrum constructed using stamping data generated by the numerical model in this invention; and MAP3 is a process energy spectrum constructed using stamping data generated by the uncorrected initial simulation model.

[0083] Figure 14 This is a spatial distribution map showing the thickness variation monitoring results of different process energy spectra in MAP1 to MAP3.

[0084] Figure 15 A comparison of the monitoring accuracy of energy spectra of different processes in MAP1 to MAP3.

[0085] (a) corresponds to the average absolute percentage error of different monitoring results; (b) corresponds to the defect identification accuracy of different monitoring results.

[0086] Figure 16 This invention provides a visualization of the thickness variation path of stamped workpieces using a process energy map constructed in accordance with the present invention.

[0087] Figure 17 for Figure 16 A comparison chart of the machining quality of actual stamped workpieces produced under different process paths. Detailed Implementation

[0088] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0089] Example 1

[0090] This embodiment provides a low-cost method for acquiring stamping data, which is used to acquire process energy and thickness variation data with mapping relationships required for constructing a process energy profile; the thickness variation data includes the maximum thinning rate and the increase in thickness rate. For example... Figure 1 As shown, the low-cost method for obtaining stamping data includes the following process:

[0091] Under the preset stamping process conditions, several discrete sample data are obtained through actual stamping tests. The sample data includes the measured stamping force and the measured process energy.

[0092] like Figure 2 As shown, the data collection method for each group of samples is as follows:

[0093] A. Set L stamping depths.

[0094] B. Measure the stamping force of the stamped workpiece at L stamping depths through actual stamping tests.

[0095] C. Calculate the process energy corresponding to L stamping depths based on L stamping depths and their corresponding stamping forces:

[0096]

[0097] In the above formula, E l ' is the measured process energy when the stamping depth is l; F(x) is the stamping force F with respect to the stamping depth h. p The fitting function.

[0098] A simulation model is constructed to simulate the stamping process. The simulation model includes a stamping geometry model and a finite element model. The stamping geometry model reflects the shape, structure, and dimensions of the stamping die and sheet metal. The finite element model is used to simulate the stamping of the sheet metal, thereby obtaining the mapping relationship between the thickness variation of the stamped workpiece and the process energy.

[0099] like Figure 3As shown, the stamping geometry model in this embodiment is constructed using any three-dimensional geometry modeling software, including SolidWorks, 3DMax, and Unity. The parameters for constructing the stamping geometry model mainly include the structural parameters of the die, such as the structural parameters of the punch, die, and blank holder, as well as the structural parameters of the sheet metal. The finite element model is established using any finite element analysis software, including Dynaform and Abaqus. To drive the operation of the finite element model, stamping control parameters, material performance parameters, friction coefficients, etc., need to be input into the finite element model. These parameters need to be reasonably designed based on the actual stamping workpiece, which will not be elaborated here.

[0100] The simulation model is used to generate the simulated maximum thinning rate, the simulated maximum thickness increase rate, and the simulated process energy.

[0101] In the scheme provided in this embodiment, the process energy is actually the vertical axis of the process energy spectrum, and its accuracy directly determines the monitoring range of the spectrum. The simulation model provided in this embodiment essentially needs to replace actual stamping tests to generate relevant stamping data; therefore, the simulated process energy and simulated thickness changes obtained from the simulation model should be consistent with the measured process energy and measured thickness changes obtained from actual stamping tests. In subsequent work on this embodiment, the main focus is on iteratively correcting the simulation model to make the simulated process energy and simulated thickness changes generated by the simulation model more closely resemble the measured process energy and measured thickness changes.

[0102] like Figure 4 As shown, the accuracy of the stamping simulation model is evaluated, and then the friction coefficient μ in the finite element model is iteratively corrected. In each iteration, the deviation between the simulated process energy and the measured process energy is calculated first; then the following decisions are made:

[0103] A. When the deviation is greater than the correction threshold, the model parameters that require manual input in the stamping geometry model are actually measured, and the actual measured values ​​of the model parameters are input into the stamping geometry model for correction; and the friction coefficient after iteration is used to correct the finite element model.

[0104] B. When the deviation is not greater than the correction threshold, save the current simulation model as the numerical model.

[0105] In the solution provided in this embodiment, the deviation between the simulated process energy and the measured process energy is expressed as the mean absolute percentage error e. MA To measure this, the specific calculation formula is as follows:

[0106]

[0107] Among them, E p,i'E' represents the measured process energy obtained from an actual stamping test when the stamping depth is i; p,i These represent the simulated process energy generated by the simulation model when the stamping depth is i; N represents the number of sets of measured process energy data obtained from actual stamping tests. In this embodiment, the preset correction threshold is 10%.

[0108] After calculating the deviation between the simulated process energy and the measured process energy, if the deviation exceeds the set correction threshold, the stamping simulation model needs to be corrected. Since the deviation between the stamping simulation model and the measurement data obtained from actual stamping tests mainly stems from the inconsistency between the modeling parameters and the actual stamping parameters, the correction of the stamping simulation model primarily involves correcting the modeling parameters.

[0109] This embodiment divides the correction of the simulation model into two parts: one is the correction of the stamping geometry model, and the other is the correction of the finite element model.

[0110] The structural parameters of the stamping geometry model can be ensured to match reality through precise measurements of the die and sheet metal. For control parameters in the finite element model's driving parameters, such as stamping speed and blank holder force, since these are active input parameters, it is usually sufficient to ensure that the input values ​​of the simulation model match the actual input values. However, fluctuations in control parameters are inevitable during the actual forming process, leading to deviations between simulation and measurement results. To avoid misjudgments, multiple combinations of stamping speed and blank holder force can be set, and the simulated process energy and measured process energy under different control parameter combinations can be compared to eliminate deviations caused by accidental fluctuations. Regarding the material property parameters of the sheet metal, since the material properties of each sheet metal may differ, even if the material property parameters are corrected during a single stamping process, deviations from reality may still exist in the next forming. Therefore, default values ​​can be used for the sheet metal material property parameters.

[0111] During the correction of the finite element model, the technicians in this embodiment discovered that, regarding the friction coefficient between the die and the sheet metal, although the constructed process energy spectrum can effectively monitor the thickness change of the stamped workpiece under different friction coefficients, the change in the friction coefficient still affects the quantitative relationship between process energy and thickness change, thus leading to a decrease in the monitoring accuracy of the constructed process energy spectrum. Therefore, correcting the friction coefficient in the finite element model of stamping is a necessary measure to ensure data accuracy.

[0112] In this embodiment, setting a fixed friction coefficient solely based on experience often results in a deviation from the actual friction coefficient. Furthermore, since the mold inevitably wears down during operation, human experience often fails to account for the resulting changes in the friction coefficient. Therefore, this embodiment employs an improved Particle Swarm Optimization (PSO-DE) algorithm to assist in iterative optimization of the friction coefficient. During the iteration process, this embodiment uses the process energy of the stamping process as the dynamic response and the friction coefficient as the optimization parameter. Figure 5 As shown, the iterative correction process for the friction coefficient in the finite element model correction process of this embodiment is as follows:

[0113] S01: Initialize the friction coefficient population μ(g) and mutation probability F a Crossover probability CR, maximum number of iterations ger, individual learning factor b1, population learning factor b2, and inertia weight q.

[0114] The friction coefficient population μ(g) is:

[0115] μ(g)=(μ1(g),μ2(g),…,μ i (g),…,μ τ (g))

[0116] In the above formula, g is the iteration number; when g = 1, the iteration is in the initial state; τ is the number of original friction coefficient individuals in the friction coefficient population μ(g).

[0117] The i-th individual with the friction coefficient μ in the friction coefficient population μ(g) i (g) is:

[0118] μ i (g)=(μ 1,i (g),μ 2,i (g))

[0119] In the above formula, μ 1,i (g) represents the i-th individual initial friction coefficient μ. i (g) The coefficient of friction between the sheet metal and the blank holder; μ 2,i (g) represents the i-th individual initial friction coefficient μ. i (g) The coefficient of friction between the sheet metal and the die.

[0120] In this embodiment, the i-th individual with the original friction coefficient μ i (g) The coefficient of friction μ between the sheet metal and the blank holder 1,i (g) and the i-th original friction coefficient individual μ i (g) The coefficient of friction μ between the sheet metal and the die 2,i The value of (g) is within the preset maximum friction coefficient μ. maxWith minimum friction coefficient μ min Randomly select from the given range, using the following formula:

[0121]

[0122] S02: Calculate the original friction coefficient individual μ in the current friction coefficient population μ(g). i (g) Average absolute percentage error of corresponding process energy e MA,i (g) and identify the individual friction coefficient μ with the smallest mean absolute percentage error. m (g)

[0123] In this embodiment, the original friction coefficient individuals μ in the friction coefficient population μ(g) i (g) The average absolute percentage error of the corresponding process energy e MA,i The formula for calculating (g) is:

[0124]

[0125] In the above formula, E l (μ i (g) When the stamping depth is l, the stamping forming simulation model adopts the original friction coefficient individual μ. i (g) Process energy as the coefficient of friction; h p,max This represents the total stamping depth.

[0126] S03: Update the original friction coefficient individual μ in the friction coefficient population μ(g) based on the following formula. i (g) evolutionary direction v i (g+1):

[0127]

[0128] In the above formula, r1 and r2 are random values ​​between 0 and 1; E is the identity matrix; e MA,m (g) represents the average absolute percentage error of the individual with the optimal friction coefficient; μ i (g), μ q (g) and μ r (g) represents three original friction coefficient individuals randomly selected from the friction coefficient population μ(g).

[0129] S04: According to the direction of evolution v i (g+1) Select the original friction coefficient individuals μ from the current friction population μ(g). i (g) Mutation is performed to obtain individual mutant friction coefficients μ. i The formula for mutation operation (g) is as follows:

[0130]

[0131] In the above formula, F a (g+1) represents the mutation probability in the (g+1)th iteration.

[0132] S05: For individuals with abruptly changed friction coefficients in the current friction population μ(g), i '(g) and the original friction coefficient of the individual μ i (g) Perform crossover operations to generate new individuals with different friction coefficients. The strategy for crossover operations is as follows:

[0133]

[0134] In the above formula, rand i It is the original friction coefficient individual μ i (g) and the individual friction coefficient of mutation μ i '(g) A random number between 0 and 1 generated during the crossover operation.

[0135] S06: Comparison of individual friction coefficients with abrupt changes (μ) i '(g) and the original friction coefficient of the individual μ i The mean absolute percentage error of the process energy in (g) is used to update the individual friction coefficients to obtain the friction coefficient population μ(g+1) at iteration g+1:

[0136]

[0137] In the above formula, e MA,i '(g) represents the individual friction coefficient μ of the mutation. i The average absolute percentage error of the process energy corresponding to '(g)'.

[0138] S07: Determine if the current iteration number g is less than the maximum iteration number ger:

[0139] If so, then repeat steps S02 to S06.

[0140] Otherwise, end the iteration and output the individual friction coefficient μ with the smallest mean absolute percentage error of process energy in the current friction coefficient population μ(g). m (g) is the corrected coefficient of friction.

[0141] Finally, when the maximum number of iterations is reached or the deviation between the simulated process energy obtained from the corrected simulation model and the actual measured process energy is less than a preset correction threshold, the simulation model is saved as the required numerical model. Then, using the numerical model as a tool, process energy and thickness variation data with mapping relationships are generated, which are needed to construct the process energy map.

[0142] Example 2

[0143] Based on the solution in Example 1, this embodiment further provides a low-cost system for acquiring stamping data, such as... Figure 6 As shown, it employs the low-cost stamping data acquisition method described in Example 1, combining process energy measurement and simulation modeling to obtain a numerical model for low-cost stamping data acquisition. The low-cost stamping data acquisition system in this embodiment includes: a process energy measurement module, a simulation model construction module, an error calculation module, a friction coefficient update module, and a numerical model correction module.

[0144] The process energy measurement module is used to calculate the process energy based on several discrete sample data obtained from the actual stamping test.

[0145] The simulation model construction module includes a stamping geometry model and a finite element model. The stamping geometry model reflects the shape, structure, and dimensions of the stamping die and sheet metal. The finite element model simulates the stamping process of the sheet metal, thereby obtaining the mapping relationship between the thickness variation of the stamped workpiece and the process energy. The simulation model is used to generate the simulated maximum thinning rate, the simulated maximum thickness increase rate, and the simulated process energy.

[0146] The error calculation module is used to calculate the deviation between the simulated process energy generated by the constructed simulation model and the measured process energy.

[0147] The friction coefficient update module is used to iteratively correct the friction coefficient in the finite element model by adopting the correction strategy as in Example 1 when the deviation calculated by the error calculation module exceeds the correction threshold.

[0148] The numerical model correction module is used to reset the manually input parameters in the stamping geometry model and correct the friction coefficient in the finite element model when the deviation calculated by the error calculation module exceeds the correction threshold. When the deviation calculated by the error calculation module does not exceed the correction threshold, the corresponding parameters in the simulation model are retained to obtain a numerical model that reflects the mapping relationship between the maximum thinning rate, maximum thickening rate and process energy of the stamped workpiece.

[0149] Example 3

[0150] This embodiment provides a data processing device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it creates a low-cost stamping data acquisition system as described in Embodiment 2. Then, based on several discrete sample data obtained from actual stamping tests, it automatically constructs a numerical model that can generate a mapping relationship between the maximum thinning rate, maximum thickening rate, and process energy of the stamped workpiece under specified process conditions. This numerical model can be used to fill in the data in the process energy map during the construction process.

[0151] The data processing device provided in this embodiment is essentially a computer device. A computer device can be a smart terminal, tablet computer, laptop computer, desktop computer, rack server, blade server, tower server, or cabinet server (including independent servers or server clusters composed of multiple servers), etc.

[0152] The computer device in this embodiment includes, but is not limited to, a memory and a processor that can communicate with each other via a system bus.

[0153] In this embodiment, the memory (i.e., the readable storage medium) includes flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory can be an internal storage unit of a computer device, such as the hard disk or RAM of the computer device. In other embodiments, the memory can also be an external storage device of the computer device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the computer device. Of course, the memory can also include both internal storage units and external storage devices of the computer device. In this embodiment, the memory is typically used to store the operating system and various application software installed on the computer device. In addition, the memory can also be used to temporarily store various types of data that have been output or will be output.

[0154] In some embodiments, the processor may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip. The processor is typically used to control the overall operation of a computer device. In this embodiment, the processor is used to run program code stored in memory or process data.

[0155] Example 4

[0156] Based on the scheme in Example 1, this embodiment further provides a method for constructing a process energy map, such as... Figure 7 and Figure 8 As shown, it includes the following steps:

[0157] 1. Under the preset stamping process conditions, obtain several discrete sample data through actual stamping tests.

[0158] Second, adopt the low-cost stamping data acquisition method as in Example 1, and establish a numerical model based on the collected sample data.

[0159] This numerical model is used to generate an array consisting of the maximum thinning rate, the maximum thickening rate, the stamping depth, and the stamping force, which have a mapping relationship.

[0160] III. Based on the stamping depth h p A blank process energy map was plotted using process energy E as the horizontal and vertical axes, respectively. Based on the process's parameter constraints, a boundary-bounded region to be filled was identified within the process energy map. The results are as follows: Figure 8 As shown in section (a).

[0161] IV. Using a numerical model to fill in the unfilled areas in the blank process energy map, the process is as follows:

[0162] 4.1: Using a numerical model to generate the associated simulated stamping depth h ps Simulated impact force F s Simulated maximum thickness increase rate Δ MTCs and the simulated maximum thinning rate Δ MTNs The first array U1 is formed as follows: U1 = {h ps F s Δ MTCs Δ MTNs}

[0163] 4.2: Based on the simulated stamping depth h ps and simulated impact force F s Calculate the corresponding simulated process energy E s :

[0164]

[0165] In the above formula, F s (x) represents the simulated impact force F. s Regarding the simulated stamping depth h ps The fitting function.

[0166] 4.3: The simulated impact force F in the first array s Replace with simulated process energy E s This is then transformed into the second array U2: U2 = {h ps E s Δ MTCs Δ MTNs}

[0167] 4.4: Using the simulated stamping depth h in the second array ps and simulation process energy E sUsing the x and y coordinates of the pixels within the area to be filled as the coordinates, the maximum thickness increase Δ is simulated. MTCs With the simulated maximum thinning rate Δ MTNs These are the first and second attribute values ​​for the corresponding pixels; this completes the pixel filling of the blank process energy spectrum.

[0168] 5. Using the vertical axis as the axis of symmetry, mirror the map filled in the above steps into two symmetrical parts: the wrinkling recognition area and the fracture recognition area. The result is as follows. Figure 8 As shown in part (b) of the document.

[0169] VI. According to the preset color mapping relationship, each pixel in the wrinkle recognition area is colored according to the first attribute value, and each pixel in the crack recognition area is colored according to the second attribute value. The result is as follows. Figure 8 As shown in section (c) of the document.

[0170] 7. Based on the preset safety threshold, the boundaries between the wrinkled area and the safe area in the wrinkled identification area, and the boundaries between the fractured area and the safe area in the fracture identification area are delineated to obtain the required process energy spectrum.

[0171] Example 5

[0172] This embodiment provides a method for online monitoring of the processing quality of stamped workpieces, such as... Figure 9 As shown, it includes the following steps:

[0173] Step 1: Obtain the process energy map of the current stamping workpiece to be processed using the process energy map construction method as described in Example 4.

[0174] Step 2: Real-time acquisition of the stamping depth h of the workpiece to be processed during the actual processing. real and real-time impact pressure F real .

[0175] Step 3: Based on the real-time stamping depth h during the processing... real and real-time impact pressure F real Calculate the real-time process energy E real .

[0176] Step 4: Based on the real-time stamping depth h of the workpiece to be processed during the processing... real and real-time process energy E real The state changes are plotted in the wrinkling and cracking identification areas of the process energy map to create corresponding state trajectories.

[0177] Step 5: Evaluate the machining quality of the stamped workpiece based on the state trajectory:

[0178] (1) When any point in the state trajectory passes through the fracture zone, it is determined that the processed stamped workpiece has fractured.

[0179] (2) When the endpoint of the state trajectory is located in the wrinkled area, it is determined that the processed stamped workpiece has local wrinkling.

[0180] Performance testing

[0181] To verify the effectiveness of the low-cost stamping data acquisition method proposed in this embodiment in reducing the cost of process energy map construction and its practical application, a test experiment for constructing a process energy map was conducted, using a car door inner panel as an example and selecting AA5052 aluminum alloy as the material. The test experiment determined the contour of the process energy map using 16 sets of stamping process conditions composed of stamping speed and blank holder force combinations. The horizontal axis of the process energy map is the stamping depth, and stamping depths from 1mm to 28mm were selected in 1mm intervals to determine the horizontal axis of the process energy map. Then, the numerical model obtained by the low-cost stamping data acquisition method provided in this embodiment was used to fill in the required data in the process energy map.

[0182] I. Control Parameter Settings for Simulation Model

[0183] In this experiment, the initial value of the friction coefficient between the die and the sheet metal was set to 0.14 in the stamping simulation model, while the measurement data of process energy and thickness change were obtained from actual stamping tests under 16ml lubrication.

[0184] In this case, 10% was selected as the correction threshold for the stamping simulation model. This avoids inaccurate process energy maps built based on the simulation model due to an excessively large threshold, while also preventing the waste of human and material resources caused by frequent corrections due to an excessively small threshold. To avoid misjudgments in the accuracy assessment of the stamping simulation model due to fluctuations in control parameters, this experiment also selected four different combinations of control parameters, namely CB1 (v = 2 mm / s, F...). h =6kN), CB2 (v=10mm / s, F h =18kN), CB3 (v=22mm / s, F h =36kN) and CB4(v=30mm / s,F h The measured process energy at 48kN was compared with the simulated process energy obtained from the initial stamping simulation model. The results and deviations are as follows: Figure 10 As shown. In the control parameter combination, v represents the stamping speed, and F... h This indicates the blank holder force. Figure 10 In the middle, e MA-修正后 The mean absolute percentage error of the simulation process energy in the corrected stamping simulation model can be calculated as follows:

[0185]

[0186] In the above formula, E l 'E' represents the measured process energy when the stamping depth is l. l-修正后 h represents the simulated process energy obtained from the modified simulation model when the stamping depth is l. p,max Here, h represents the total stamping depth. p,max =28mm; L is the number of sets of measured process energy data. e MA-修正前 The mean absolute percentage error of the simulation process energy of the initial stamping simulation model before correction can be calculated as follows:

[0187]

[0188] In the above formula, E l-修正前 The simulated process energy is the initial simulation model before correction when the stamping depth is l.

[0189] Depend on Figure 10 The data shows that the simulated process energy of the initial stamping forming simulation model is greater than the measured process energy under different combinations of control parameters, and there is a significant deviation. The measured process energy was obtained under lubrication conditions of 16ml of lubricant, while the friction coefficient set in the initial stamping forming simulation model is 0.14, which corresponds to lubrication conditions of approximately 8ml of lubricant. Analysis shows that the worse the lubrication conditions, the more energy input is required to overcome friction, thus increasing the process energy. Therefore, the simulated process energy of the initial stamping forming simulation model will be greater than the measured process energy. Furthermore, under various combinations of control parameters, the average absolute percentage error of the simulated process energy of the initial stamping forming simulation model before correction exceeds 10%, which is the set correction threshold. Therefore, the initial stamping forming simulation model needs to be corrected.

[0190] II. Revision of the Simulation Model

[0191] Since the simulation process energy deviation is largest in the initial stamping simulation model under the control parameter combination CB1, this experiment continues to use the measured process energy under CB1 as the benchmark to correct the friction coefficient of the stamping forming simulation model. Among them, the friction coefficient correction of the initial stamping forming simulation model using the PSO-DE iterative algorithm is used as the experimental group of this embodiment, and the parameter settings of the experimental group are shown in Table 1.

[0192] Table 1: Initial parameter settings for the PSO-DE algorithm in the experimental group

[0193]

[0194] In addition, this embodiment also uses the classic Particle Swarm Optimization (PSO) algorithm to correct the friction coefficient of the initial stamping simulation model, serving as a control group for this embodiment. Both the experimental and control groups were run 200 times in Matlab. The mean absolute percentage error of the simulation process energy corresponding to the optimal friction coefficient in each run is as follows: Figure 11 As shown in section (a), the convergence curves of the two are as follows: Figure 11 As shown in section (b).

[0195] from Figure 11 As can be seen, the average absolute percentage error of the simulated process energy corresponding to the optimal friction coefficient of the PSO-DE algorithm fluctuates steadily between 2% and 5%, while the average absolute percentage error of the simulated process energy corresponding to the optimal friction coefficient of the PSO algorithm fluctuates between 4% and 10%. The accuracy of the friction coefficient optimized by the PSO-DE algorithm is significantly higher than that of the PSO algorithm. Furthermore, in the convergence curves of the PSO-DE and PSO algorithms, the PSO-DE algorithm converges after approximately 10 iterations, while the PSO algorithm converges after approximately 55 iterations. In terms of running speed, the PSO-DE algorithm is also significantly superior to the PSO algorithm.

[0196] In the average absolute percentage error of the simulation process energy corresponding to the individual optimal friction coefficient obtained from 200 runs of the PSO-DE algorithm, the minimum value and its corresponding optimal friction coefficient are identified as μ1 = 0.077 and μ2 = 0.086; where μ1 is the friction coefficient between the sheet metal and the pressure ring, and μ2 is the friction coefficient between the sheet metal and the die.

[0197] After obtaining the optimal friction coefficient, this coefficient was input into the stamping simulation model to correct the model. The corrected simulation model shows the simulated process energy under CB1, CB2, CB3, and CB4 as follows: Figure 12 As shown. From Figure 12 As can be seen, under different combinations of control parameters, the simulated process energy obtained by the corrected simulation model is significantly closer to the measured process energy than that obtained by the initial simulation model. The maximum mean absolute percentage error of the simulated process energy obtained by the corrected simulation model does not exceed 4.51%, which is less than the correction threshold of 10%. Furthermore, a comparison of the maximum simulated thinning rate and maximum simulated thickening rate between the original and corrected simulation models under CB1, CB2, CB3, and CB4 is shown below. Figure 12 As shown. From Figure 12As can be seen, under different combinations of control parameters, the simulated maximum thinning rate and simulated maximum thickness rate obtained by the modified simulation model are significantly closer to the measured maximum thinning rate and measured maximum thickness rate than those obtained by the initial simulation model. The comparison of simulated process energy and simulated thickness changes obtained by the simulation models before and after modification proves that the modified simulation model meets the usage requirements and can be used for further acquisition of simulated process energy and simulated thickness change data. This confirms the effectiveness of the solution provided in this embodiment.

[0198] III. Constructing Process Energy Maps Based on the Modified Simulation Model

[0199] After obtaining the corrected stamping simulation model, the set stamping process conditions can be input into the simulation model to acquire simulation process energy and thickness variation data. After data acquisition, a process energy map is constructed. In the constructed process energy map, the horizontal axis represents the stamping depth h. p The vertical axis represents the process energy E; the color of the pixels in the interval is used to characterize the thickness change of the stamped workpiece corresponding to each stamping depth and its process energy. The thickness change of the stamped workpiece includes the maximum thinning rate Δ. MTN Maximum thickness increase Δ MTC The two are represented in two coordinate graphs. The process energy map constructed based on the modified stamping simulation model in this embodiment is roughly similar to the process energy maps constructed using two other different data acquisition methods. Figure 13 As shown, the images of MAP2 are very similar to those of MAP1, and both differ significantly from MAP3.

[0200] IV. Accuracy Analysis of Process Energy Map

[0201] To verify the effectiveness of the process energy map constructed using the modified stamping simulation model provided in this embodiment, this experiment selected the thickness change of a door-like inner panel under a stamping depth of 27.5 mm to identify the monitoring accuracy and defect identification accuracy of the process energy map constructed using the modified stamping simulation model.

[0202] Monitoring results of process energy spectrum as follows Figure 14 As shown, from Figure 14 It can be seen that the maximum thinning rate and maximum thickening rate monitored by the scheme in this embodiment are very close to the measurement results. Only 4 unbroken parts were identified as broken parts by the spectrum, with a fracture identification accuracy of 87.5%. 5 wrinkled parts were not correctly identified by the spectrum, with a wrinkle identification accuracy of 84.4%. The process energy spectrum constructed by the corrected stamping simulation model can effectively identify defects in stamped workpieces.

[0203] The mean absolute percentage error of the process energy spectrum monitoring results constructed in this embodiment is as follows: Figure 15 As shown in MAP2 in (a), the maximum error in thickness variation monitoring does not exceed 8%, demonstrating the high monitoring accuracy of the process energy map constructed using the corrected stamping simulation model for thickness variation in stamped workpieces. The defect identification accuracy of this embodiment is as follows: Figure 15 As shown in MAP2 in (b), all exceed 80%.

[0204] In addition, from Figure 14 It can also be observed that: because MAP1 is established based on actual measurement results, its monitoring results are actually closer to the measurement results. And from... Figure 15 It can also be seen that MAP1's monitoring error for the maximum thinning rate of stamped workpieces (3.09%) is 2.02% lower than MAP2's (5.11%), and MAP1's monitoring error for the maximum thickening rate of stamped workpieces (3.07%) is 4.74% lower than MAP2's (7.81%). MAP1's fracture detection rate (93.75%) is 6.25% higher than MAP2's (87.5%), and MAP1's wrinkling detection rate (93.75%) is 9.35% higher than MAP2's (84.4%).

[0205] The reason why the monitoring error and defect identification accuracy of the process energy map constructed using the method of this embodiment are lower than those of MAP1 is that, in the actual stamping process, the friction coefficient between the sheet metal and the die, as well as the material properties of the sheet metal, are constantly changing. With changes in control parameters such as blank holder force and stamping speed, the friction state between the die and the sheet metal also changes continuously, thus affecting the quantitative relationship between the process energy and thickness changes of the stamped workpiece. However, in the stamping simulation model, the friction coefficient is set as a constant, which is inconsistent with the actual stamping process. This leads to errors in the simulated process energy and simulated thickness changes obtained by the simulation model, thereby affecting the monitoring accuracy of the process energy map constructed from the simulation results. Furthermore, the material properties of the sheet metal change continuously during the stamping process, and these changing material properties affect the process energy and thickness changes of the stamped workpiece. Since these material properties are also constant in the stamping simulation model, this also leads to a decrease in the monitoring accuracy of the process energy map constructed from the simulation results. The decrease in monitoring accuracy also leads to a decrease in defect identification accuracy.

[0206] However, overall, the monitoring error of the process energy map constructed using the low-cost stamping data acquisition method in this case is still below 8%, and the defect identification rate is above 80%, maintaining a high level and significantly better than MAP3. Furthermore, the construction cost of the process energy map constructed using this embodiment can be significantly reduced. Specific comparisons of process energy and sheet metal consumption are shown in Table 2.

[0207] Table 2: Cost Comparison of Different Process Energy Mapping Construction Methods

[0208]

[0209] Analysis of the data in the table above shows that the traditional method of constructing a process energy map based on test data requires stamping 448 door-like inner panels. After each stamping, the formed stamped workpiece needs to be cut and its thickness measured, which consumes material resources such as the electrical energy required for the press operation, the material consumption of sheet metal, and human resources. However, the process energy map constructed using the method described in this case only requires stamping 4 door-like inner panels. Compared to MAP1, MAP2 reduces the sheet metal consumption rate by 99.11%. Since each forming of a door-like inner panel requires a certain amount of process energy, the 448 formings of door-like inner panels consumed approximately 111.43 kJ of process energy, while the 4 formings of door-like inner panels used for stamping simulation model correction consumed only 3.79 kJ, saving 107.64 kJ of process energy. Based on the existing energy efficiency analysis results of medium-sized hydraulic presses, the process energy consumption of hydraulic presses accounts for about 7% of the total energy consumption of press operation. This means that the process energy map method provided in this case can save about 1537.71kJ of energy when constructing a process energy map.

[0210] In summary, the process energy map constructed by the low-cost stamping data acquisition method provided in this embodiment can significantly reduce the construction cost of the process energy map and greatly shorten the test cycle for acquiring test data while maintaining a similar monitoring accuracy to the traditional process energy map based on test data. Therefore, it has great practical value and can generate good economic benefits.

[0211] V. Application of the Process Energy Map Constructed in this Case

[0212] The process energy map constructed using the scheme in this embodiment can visualize the thickness change of the stamped workpiece during processing, thereby supporting production decisions during the stamping process to avoid the generation of defective parts. When the stamping speed is 14mm / s and the blank holder force is 42kN, the thickness change path 1 of the car door inner panel is as follows. Figure 16As indicated by circle 1, the maximum thinning rate of the stamped workpiece is located in the safe zone at a stamping depth of 26mm, while it is located in the fracture zone at 27mm. The maximum thinning rate of the stamped workpiece moves further into the fracture zone at 28mm, meaning that the stamped workpiece fractured at 27mm, and the crack further expanded at 28mm.

[0213] like Figure 17 As shown, when inspecting the actual processing effect of the stamped workpiece, it was found that when the inner panel of this type of car door was stamped to 27mm, under the thickness change path 1, the stamped workpiece produced obvious small cracks, and when it was stamped to 28mm, the cracks further expanded into large cracks, proving the effectiveness and accuracy of the process energy map constructed in this embodiment for monitoring the thickness change and defects of the stamped workpiece.

[0214] To suppress defects generated during the stamping process, technicians can implement process control over the stamping process using a constructed process energy map. When the workpiece is stamped to 26mm, the blank holder force is reduced by 10kN. Under this control operation, the thickness change path of the stamped workpiece is as follows: Figure 16 The area marked by circle 2 is called thickness variation path 2. Under thickness variation path 2, when the stamped workpiece is stamped to 27mm and 28mm, although the risk of wrinkling increases, that is, the maximum thickness increase rate approaches the wrinkling zone, the maximum thinning rate falls within the safe zone.

[0215] Combination Figure 17 The actual processing results showed that: under thickness change path 2, the stamped workpiece did not crack or wrinkle when stamped to 27mm and 28mm, and the maximum thinning rate of the stamped workpiece decreased by 14.56%.

[0216] This demonstrates that the process energy map constructed using the solution provided in this embodiment can achieve "visualization" of the thickness variation of stamped workpieces, thereby intuitively observing the trend of defect generation in stamped workpieces and enabling timely control of the stamping process, effectively avoiding the generation of defects in stamped workpieces.

[0217] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A low-cost method for acquiring stamping data, used to acquire, at low cost, process energy and thickness variation data with mapping relationships required for constructing a process energy profile; the thickness variation data includes the maximum thinning rate and the maximum thickening rate; characterized in that, The method for obtaining stamping data at low cost includes the following steps: S1: Under preset stamping process conditions, several discrete sample data are obtained through actual stamping tests. The sample data includes the measured stamping force and the measured process energy. S2: Construct a simulation model for simulating the stamping process. The simulation model includes a stamping geometry model and a finite element model. The stamping geometry model is used to reflect the shape, structure, and size of the stamping die and sheet metal. The finite element model is used to simulate the stamping of the sheet metal, thereby obtaining the mapping relationship between the thickness change of the stamped workpiece and the process energy. S3: Use simulation models to generate the simulated maximum thinning rate, simulated maximum thickness increase rate, and simulated process energy; S4: Calculate the deviation between the simulated process energy and the measured process energy; then make the following decision: A. When the deviation exceeds the correction threshold, the model parameters that require manual input in the stamping geometry model are actually measured, and the actual measured values ​​of the model parameters are input into the stamping geometry model for correction; and the friction coefficient μ in the finite element model is iteratively corrected. B. When the deviation is not greater than the correction threshold, save the current simulation model as the numerical model; S5: Using the numerical model as a tool, generate process energy and thickness variation data with mapping relationships required for constructing the process energy map.

2. The method for low-cost acquisition of stamping data as described in claim 1, characterized in that: In step S1, the sample data is obtained using the following method: A. Set L stamping depths; B. Measure the stamping force of the stamped workpiece at L stamping depths through actual stamping tests; C. Calculate the process energy corresponding to the L stamping depths based on the L stamping depths and their corresponding stamping forces: In the above formula, E l ' is the measured process energy when the stamping depth is l; F(x) is the stamping force F with respect to the stamping depth h. p The fitting function.

3. The method for low-cost acquisition of stamping data as described in claim 1, characterized in that: In step S4, the correction strategy for the friction coefficient μ is as follows: S01: Initialize the friction coefficient population μ(g) and mutation probability F a Crossover probability CR, maximum number of iterations ger, individual learning factor b1, population learning factor b2, and inertia weight q; S02: Calculate the original friction coefficient individual μ in the current friction coefficient population μ(g). i (g) Average absolute percentage error of corresponding process energy e MA,i (g) and identify the individual friction coefficient μ with the smallest mean absolute percentage error. m (g); S03: Update the original friction coefficient individual μ in the friction coefficient population μ(g) based on the following formula. i (g) evolutionary direction v i (g+1): In the above formula, r1 and r2 are random values ​​between 0 and 1; E is the identity matrix; e MA,m (g) represents the average absolute percentage error of the individual with the optimal friction coefficient; μ i (g), μ q (g) and μ r (g) represents three original friction coefficient individuals randomly selected from the friction coefficient population μ(g); S04: According to the direction of evolution v i (g+1) Select the original friction coefficient individual μ from the current friction coefficient population μ(g). i (g) Mutation is performed to obtain individual mutant friction coefficients μ. i The formula for mutation operation (g) is as follows: In the above formula, F a (g+1) is the mutation probability in the (g+1)th iteration; S05: For the mutant friction coefficient individuals μ in the current friction coefficient population μ(g) i '(g) and the original friction coefficient of the individual μ i (g) Perform crossover operations to generate new individuals with different friction coefficients. The strategy for crossover operations is as follows: In the above formula, rand i It is the original friction coefficient individual μ i (g) and the individual friction coefficient of mutation μ i (g) A random number between 0 and 1 generated during the crossover operation; S06: Comparison of individual friction coefficients with abrupt changes (μ) i '(g) and the original friction coefficient of the individual μ i The mean absolute percentage error of the process energy in (g) is used to update the individual friction coefficients to obtain the friction coefficient population μ(g+1) at iteration g+1: In the above formula, e MA,i '(g) represents the individual friction coefficient μ of the mutation. i '(g) corresponds to the average absolute percentage error of process energy; S07: Determine if the current iteration number g is less than the maximum iteration number ger: If yes, repeat steps S02 to S06; otherwise, end the iteration and output the optimal friction coefficient individual μ with the smallest mean absolute percentage error of process energy in the current friction coefficient population μ(g). m (g) is the corrected coefficient of friction.

4. The method for low-cost acquisition of stamping data as described in claim 3, characterized in that: In step S01, the friction coefficient population μ(g) is: μ(g)=(μ1(g),μ2(g),…,μ i (g),…,μ τ (g)) In the above formula, g is the iteration number; when g = 1, the iteration is in the initial state; τ is the number of original friction coefficient individuals in the friction coefficient population μ(g); Among them, the i-th original friction coefficient individual μ in the friction coefficient population μ(g) i (g) is: m i (g)=(μ 1,i (g),μ 2,i (g)) In the above formula, μ 1,i (g) represents the i-th individual initial friction coefficient μ. i (g) The coefficient of friction between the sheet metal and the blank holder; μ 2,i (g) represents the i-th individual initial friction coefficient μ. i (g) The coefficient of friction between the sheet metal and the die.

5. The method for low-cost acquisition of stamping data as described in claim 4, characterized in that: The i-th initial friction coefficient individual μ i (g) The coefficient of friction μ between the sheet metal and the blank holder 1,i (g) and the i-th original friction coefficient individual μ i (g) The coefficient of friction μ between the sheet metal and the die 2,i The value of (g) is within the preset maximum friction coefficient μ. max With minimum friction coefficient μ min Randomly select from the given range, using the following formula:

6. The method for low-cost acquisition of stamping data as described in claim 5, characterized in that: In step S02, the initial friction coefficient individuals μ in the friction coefficient population μ(g) i (g) Corresponding average absolute percentage error of process energy e MA,i The formula for calculating (g) is: In the above formula, E l (μ i (g) When the stamping depth is l, the stamping forming simulation model adopts the original friction coefficient μ. i (g) Process energy as the coefficient of friction; h p,max This represents the total stamping depth.

7. A method for constructing a process energy map, comprising the following steps:

1. Under the preset stamping process conditions, obtain several discrete sample data through actual stamping tests; 2. Using the low-cost stamping data acquisition method described in any one of claims 1-6, a numerical model is established by combining the collected sample data; The numerical model is used to generate an array consisting of the maximum thinning rate, the maximum thickening rate, the stamping depth, and the stamping force, which have a mapping relationship. III. Based on the stamping depth h p A blank process energy map is plotted with process energy E as the horizontal and vertical axes, and a boundary-bounded area to be filled is determined based on the process constraints on parameters. IV. Using the aforementioned numerical model, data is filled into the unfilled areas of the blank process energy map. The process is as follows: 4.1: Using the numerical model, generate the associated simulated stamping depth h ps Simulated impact force F s Simulated maximum thickness increase rate Δ MTCs and the simulated maximum thinning rate Δ MTNs The first array U1 is formed as follows: U1 = {h ps F s Δ MTCs Δ MTNs }; 4.2: Based on the simulated stamping depth h ps and simulated impact force F s Calculate the corresponding simulated process energy E s : In the above formula, F s (x) represents the simulated impact force F. s Regarding the simulated stamping depth h ps The fitting function; 4.3: Convert the first array U1 into the second array U2: U2 = {h ps E s Δ MTCs Δ MTNs }; 4.4: Using the simulated stamping depth h in the second array ps and simulation process energy E s Using the x and y coordinates of the pixels within the area to be filled as the coordinates, the maximum thickness increase Δ is simulated. MTCs With the simulated maximum thinning rate Δ MTNs These are the first and second attribute values ​​for the corresponding pixels; this completes the pixel filling of the blank process energy map.

5. Using the vertical axis as the axis of symmetry, mirror the map filled in the above steps into two symmetrical parts; these are the wrinkle recognition area and the breakage recognition area, respectively.

6. According to the preset color mapping relationship, each pixel in the wrinkle recognition area is colored according to the first attribute value, and each pixel in the crack recognition area is colored according to the second attribute value.

7. Based on the preset safety threshold, the boundaries between the wrinkled area and the safe area in the wrinkled identification area, and the boundaries between the fractured area and the safe area in the fracture identification area are delineated to obtain the required process energy spectrum.

8. A method for online monitoring of the processing quality of stamped workpieces, characterized in that, It includes the following steps: Step 1: Obtain the process energy map of the current stamping workpiece to be processed, generated using the process energy map construction method as described in claim 7; Step 2: Real-time acquisition of the stamping depth h of the workpiece to be processed during the actual processing. real and real-time impact pressure F real ; Step 3: Based on the real-time stamping depth h during the processing... real and real-time impact pressure F real Calculate the real-time process energy E real ; Step 4: Based on the real-time stamping depth h of the workpiece to be processed during the processing... real and real-time process energy E real The state changes are plotted in the wrinkling and cracking identification areas of the process energy map to show the corresponding state trajectories. Step 5: Evaluate the machining quality of the stamped workpiece based on the stated state trajectory. (1) When any point in the state trajectory passes through the fracture zone, it is determined that the processed workpiece has fractured; (2) When the endpoint of the state trajectory is located in the wrinkled area, it is determined that the processed workpiece has local wrinkles.

9. A low-cost system for acquiring stamping data, characterized in that: It employs the low-cost stamping data acquisition method described in any one of claims 1 to 6, combining measurement process energy and simulation models to obtain a numerical model for low-cost stamping data acquisition, which includes: The process energy acquisition module is used to calculate the process energy based on several discrete sample data obtained from actual stamping tests. The simulation model construction module includes a stamping geometry model and a finite element model. The stamping geometry model reflects the shape, structure, and size of the stamping die and sheet metal. The finite element model simulates the stamping of the sheet metal to obtain the mapping relationship between the thickness change of the stamped workpiece and the process energy. The simulation model generates the simulated maximum thinning rate, the simulated maximum thickness increase rate, and the simulated process energy. The error calculation module is used to calculate the deviation between the simulated process energy generated by the constructed simulation model and the measured process energy; The friction coefficient update module is used to iteratively correct the friction coefficient in the finite element model using any one of the correction strategies included in claims 3 to 6 when the deviation calculated by the error calculation module exceeds the correction threshold; and the numerical model correction module is used to reset the manually input model parameters in the stamping geometry model and correct the friction coefficient in the finite element model when the deviation calculated by the error calculation module exceeds the correction threshold; when the deviation calculated by the error calculation module does not exceed the correction threshold, the parameters in the corresponding simulation model are retained to obtain a numerical model for generating a mapping relationship between the maximum thinning rate, the maximum thickening rate and the process energy of the stamped workpiece.

10. A data processing apparatus comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, it creates a low-cost stamping data acquisition system as described in claim 9; then, based on several discrete sample data obtained from actual stamping tests, it automatically constructs a numerical model that can generate a mapping relationship between the maximum thinning rate, maximum thickening rate and process energy of the stamped workpiece under specified process conditions.

Images