Zip-top can double-seaming process parameter optimization method and device

Abstract

The invention provides a ring-pull can double rolling and sealing process parameter optimization method and device, relates to the technical field of double rolling and sealing, and solves the technical problem that only a single target is focused in the ring-pull can rolling and sealing process parameter optimization process in the prior art. The method specifically comprises the following steps: acquiring key process parameters and target process parameters; performing finite element simulation on the key process parameters to obtain a preliminary relationship between the key process parameters and the target process parameters; in finite element simulation, the number of wave crests and the maximum residual wave crest height are extracted, and a compactness evaluation data set is constructed; training a DFNN proxy model based on the compactness evaluation data set; and based on the DFNN proxy model, a series of Pareto optimal solutions are obtained by adopting an NSGA-II algorithm, and an equilibrium solution is selected from the series of Pareto optimal solutions as an optimal process parameter. The method is used for optimizing the parameters of the double rolling and sealing process of the zip-top can.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of double-seaming, and in particular to a method and device for optimizing double-seaming process parameters of a zip-top can. BACKGROUND

[0002] The double-seaming quality of a zip-top can directly determines the sealing reliability thereof, and is crucial to food packaging safety and equipment service life; currently, process parameter setting is mostly dependent on the experience of engineers and a tedious trial-and-error method, and some research attempts to adjust parameters by combining simulation with an optimization algorithm, but the traditional method lacks objective and fine quantitative means for the tightness, a core index of seaming, the optimization process often only focuses on a single target while ignoring the coordinated trade-off of multiple indexes, and the high cost of finite element simulation calculation leads to an excessively long optimization period, so the prior art has the technical problem that the optimization process of the double-seaming process parameters of the zip-top can only focuses on a single target. SUMMARY

[0003] The application provides a method and device for optimizing double-seaming process parameters of a zip-top can, solving the technical problem that the prior art has the technical problem that the optimization process of the double-seaming process parameters of the zip-top can only focuses on a single target.

[0004] To achieve the above-mentioned purpose, the application adopts the following technical solutions: In a first aspect, a method for optimizing double-seaming process parameters of a zip-top can is provided, comprising: obtaining key process parameters and target process parameters; the key process parameters include the number of feedings of a first roller, the number of feedings of a second roller, the speed of the can body, and the friction coefficient; the target process parameters include the overlap rate, the number of wave crests, and the maximum contact stress; performing finite element simulation on the key process parameters to obtain the preliminary relationship between the key process parameters and the target process parameters; in the finite element simulation, extracting the number of wave crests and the maximum residual wave crest height, and constructing a tightness evaluation dataset; the tightness evaluation dataset includes the key process parameters, the target process parameters, and the maximum residual wave crest height; training a DFNN surrogate model based on the tightness evaluation dataset; the DFNN surrogate model is used to represent the mapping relationship between the key process parameters and the target process parameters; based on the DFNN surrogate model, a series of Pareto optimal solutions are obtained by using an NSGA-II algorithm, and a balanced solution is selected from the series of Pareto optimal solutions as the optimal process parameters.

[0005] In combination with the first aspect described above, in a possible implementation manner, the finite element simulation is performed on the key process parameters to obtain the preliminary relationship between the key process parameters and the target process parameters, comprising: adjusting the values of the key process parameters respectively and repeating the finite element simulation; recording the changes of the target process parameters corresponding to different key process parameter values to obtain the preliminary relationship between the key process parameters and the target process parameters.

[0006] In a possible implementation manner of the first aspect, in the finite element simulation, the number of wave crests and the maximum residual wave crest height are extracted, including: extracting the number of wave crests of one lap seal rear cover hook through post-processing; if the number of wave crests is the same, extracting the maximum residual wave crest height after two lap seals.

[0007] In a possible implementation manner of the first aspect, the DFNN surrogate model is trained based on the tightness evaluation data set, including: based on the tightness evaluation data set, a simulation experiment is designed by using an L16 orthogonal table to obtain training samples; and the DFNN surrogate model is trained based on the training samples.

[0008] In a possible implementation manner of the first aspect, the DFNN surrogate model includes one input layer, one output layer, and four hidden layers; the input layer has four nodes; the output layer has three nodes; and the four hidden layers have structures of 8 nodes, 16 nodes, 16 nodes, and 8 nodes in sequence.

[0009] In a possible implementation manner of the first aspect, a series of Pareto optimal solutions are obtained by using the NSGA-II algorithm based on the DFNN surrogate model, including: generating a plurality of groups of key process parameter combinations that meet process constraints; inputting each group of key process parameter combinations into the trained DFNN surrogate model to obtain corresponding target process parameters; taking the key process parameters as a population, performing non-dominated sorting on individuals in the population, and screening high-quality individuals; generating a new generation of population through selection, crossover, and mutation operations, and cyclically performing the step of screening high-quality individuals until a preset evolution number is reached; and collecting all non-dominated solutions in the iteration process to form a series of Pareto optimal solutions.

[0010] In a possible implementation manner of the first aspect, the population size of the NSGA-II algorithm is 100, the preset evolution number is 200, and the optimization target is to maximize the overlap rate, maximize the number of wave crests, and minimize the maximum contact stress.

[0011] In a possible implementation manner of the first aspect, one balanced solution is selected from the series of Pareto optimal solutions as the optimal process parameter, including: obtaining extreme values of the target process parameters in the series of Pareto optimal solutions; the extreme values include a maximum value of the overlap rate, a maximum value of the number of wave crests, a minimum value of the maximum contact stress, and a maximum value of the maximum contact stress; based on the extreme values, calculating a score value of each normalized Pareto optimal solution; and selecting the key process parameters corresponding to the Pareto optimal solution with the highest score value as the optimal process parameters.

[0012] In a possible implementation manner of the first aspect, the score value of each normalized Pareto optimal solution satisfies the following formula:

[0013] wherein S is a score value, is an overlap rate, is a number of peaks, is a maximum contact stress, is a maximum value of the overlap rate, is a maximum value of the number of peaks, is a minimum value of the maximum contact stress, is a maximum value of the maximum contact stress, is an overlap rate weight, is a number of peaks weight, is a maximum contact stress weight.

[0014] In a second aspect, a two-step can end process parameter optimization device is provided, comprising: a communication unit and a processing unit; the communication unit is used to obtain key process parameters and target process parameters; the key process parameters include a first roller feed number of turns, a second roller feed number of turns, a can body rotation speed, and a friction coefficient; the target process parameters include an overlap rate, a number of peaks, and a maximum contact stress; the processing unit is used to perform finite element simulation on the key process parameters, to obtain a preliminary relationship between the key process parameters and the target process parameters; in the finite element simulation, the number of peaks and the maximum residual peak height are extracted, and a tightness evaluation dataset is constructed; the tightness evaluation dataset includes the key process parameters, the target process parameters, and the maximum residual peak height; a DFNN proxy model is trained based on the tightness evaluation dataset; the DFNN proxy model is used to represent the mapping relationship between the key process parameters and the target process parameters; based on the DFNN proxy model, a series of Pareto optimal solutions are obtained by using an NSGA-II algorithm, and a balanced solution is selected from the series of Pareto optimal solutions as an optimal process parameter.

[0015] The application provides a two-step can end process parameter optimization method and device, which can realize objective and fine quantification of end tightness, give consideration to multi-objective collaborative optimization of sealing quality and equipment stress, solve the technical problem that the prior art focuses on a single target in the process of optimizing can end process parameters, and improve multi-objective optimization efficiency through a DFNN proxy model, quickly obtain global optimal process parameters, and further improve can sealing reliability and equipment durability.

[0016] It should be understood that the description of technical features, technical solutions, advantages or similar language in this application does not imply that all features and advantages can be realized in any single embodiment. On the contrary, it can be understood that the description of a feature or advantage means that the specific technical feature, technical solution or advantage is included in at least one embodiment. Therefore, the description of technical features, technical solutions or advantages in this specification does not necessarily refer to the same embodiment. Further, the technical features, technical solutions and advantages described in this embodiment can be combined in any appropriate manner. Those skilled in the art will understand that the embodiments can be implemented without one or more specific technical features, technical solutions or advantages of the specific embodiments. In other embodiments, additional technical features and advantages can be identified in specific embodiments that do not embody all embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 A flowchart of a method for optimizing process parameters of a two-fold can end of a can is provided for an embodiment of the present application; Figure 2 A flowchart of another method for optimizing process parameters of a two-fold can end of a can is provided for an embodiment of the present application; Figure 3 A schematic diagram of a first can end after a first can end is provided for an embodiment of the present application; Figure 4 A schematic diagram of a residual wave crest after a second can end is provided for an embodiment of the present application; Figure 5 A flowchart of another method for optimizing process parameters of a two-fold can end of a can is provided for an embodiment of the present application; Figure 6 A schematic diagram of a DFNN agent model structure is provided for an embodiment of the present application; Figure 7 A flowchart of another method for optimizing process parameters of a two-fold can end of a can is provided for an embodiment of the present application; Figure 8 A Pareto frontier diagram is provided for an embodiment of the present application; Figure 9 A flowchart of another method for optimizing process parameters of a two-fold can end of a can is provided for an embodiment of the present application; Figure 10 A schematic diagram of a device for optimizing process parameters of a two-fold can end of a can is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0018] In the description of the present application, unless otherwise specified, " / " means "or", for example, A / B can mean A or B. "And / or" herein is only a description of the relationship between the associated objects, which means that there can be three relationships, for example, A and / or B, which can mean: A exists alone, A and B exist together, and B exists alone. In addition, "at least one" means one or more, and "multiple" means two or more. "First", "second", and the like do not limit the quantity and execution order, and "first", "second", and the like do not necessarily mean different.

[0019] It should be noted that in the present application, the words "exemplary" or "for example" are used to mean serving as an example, instance, or illustration. Any embodiment or design scheme described as "exemplary" or "for example" in the present application should not be interpreted as being more preferred or having more advantages than other embodiments or design schemes. Rather, the use of "exemplary" or "for example" is intended to present the relevant concept in a specific manner.

[0020] To solve the technical problem that the existing technology focuses on a single target in the optimization process of the can end double-seaming process parameters, the present application provides a can end double-seaming process parameter optimization method, which comprises the following steps: first, obtaining key process parameters and target process parameters, constructing a tightness evaluation dataset through finite element simulation, training a DFNN surrogate model representing the mapping relationship of the parameters, and then obtaining a Pareto optimal solution based on the model using the NSGA-II algorithm, and selecting a balanced solution as the optimization parameter. Based on this, the multi-objective collaborative optimization of the overlap rate, tightness and equipment contact stress is realized, breaking through the limitations of single target optimization. At the same time, the objective and fine characterization of tightness is realized through the double-parameter quantification strategy of the number of wave peaks and the maximum residual wave peak height, the optimization efficiency is improved with the help of the DFNN surrogate model, and finally the globally optimal process parameters are obtained, which take into account the sealing quality of the can end and the durability of the equipment.

[0021] As shown in Figure 1 The can end double-seaming process parameter optimization method provided by the present application comprises the following steps: S101, obtaining key process parameters and target process parameters.

[0022] Among them, the key process parameters are adjustable variables that affect the sealing quality, including the number of feedings of the first roller, the number of feedings of the second roller, the speed of the can body, and the friction coefficient; the target process parameters are core indicators for measuring the sealing effect and equipment loss, including the overlap rate, the number of wave peaks, and the maximum contact stress.

[0023] In the present application, the key process parameters can be determined through production experience, literature research or pre-experiment screening, and the target process parameters can be obtained by double-seaming the can end based on the determined key process parameters.

[0024] Based on the above steps, the optimized input variables and evaluation criteria are determined, ensuring that the subsequent process is optimized.

[0025] S102, finite element simulation is performed on the key process parameters to obtain the preliminary relationship between the key process parameters and the target process parameters.

[0026] Wherein, the finite element simulation is a technique for simulating material deformation and contact behavior during the sealing process through numerical calculation, and the preliminary relationship refers to the correlation trend of parameter change and index fluctuation, i.e. the correlation trend of key process parameters and target process parameters.

[0027] In the embodiments of the present application, a sealing finite element model is first established according to the actual working conditions, then the values of each key process parameter are adjusted and the finite element simulation is repeated, the changes of the target process parameters corresponding to different key process parameter values are recorded, and the correlation trend of the key process parameters and the target process parameters is obtained.

[0028] It should be pointed out that the specific process of establishing the finite element model is not limited in the present application.

[0029] As an example, the number of roller feed laps has the greatest impact on the sealing index, and the tank speed has the most significant impact on the contact stress.

[0030] Based on the above steps, the correlation between parameters and indexes is obtained, avoiding the blindness of subsequent optimization and improving the rationality of optimization.

[0031] S103, in the finite element simulation, the number of wave peaks and the maximum residual wave peak height are extracted to construct a tightness evaluation data set.

[0032] Wherein, the number of wave peaks is the number of effective protrusions of the cover hook after one-way sealing, and the maximum residual wave peak height is the maximum residual height of the un-compressed wave peak after two-way sealing, both of which are quantitative indexes of tightness.

[0033] In the embodiments of the present application, the number of wave peaks of the cover hook section is counted and the residual wave peak height is measured in all finite element simulations of S102, and then associated with the corresponding key process parameters and target process parameters to form a structured tightness evaluation data set.

[0034] Based on the above steps, the number of wave peaks and the residual wave peak height are extracted to realize the breakthrough from qualitative judgment to quantitative analysis of sealing performance, laying a foundation for high-precision optimization.

[0035] S104, training a DFNN proxy model based on the tightness evaluation data set.

[0036] Among them, the DFNN surrogate model uses a multi-layer neural network to simulate the nonlinear relationship between parameters and indicators, replacing the high-cost finite element simulation model, and is used to characterize the mapping relationship between key process parameters and target process parameters.

[0037] In this embodiment of the application, the density evaluation dataset is divided into a training set and a validation set, and then the model network structure and training hyperparameters are determined, and the model is iteratively trained until it converges.

[0038] Based on the above steps, an accurate parameter and indicator mapping model was constructed, which significantly reduced the computational cost of subsequent optimization and improved optimization efficiency.

[0039] S105. Based on the DFNN surrogate model, the NSGA-II algorithm is used to obtain a series of Pareto optimal solutions, and an equilibrium solution is selected from the series of Pareto optimal solutions as the optimal process parameters.

[0040] Among them, the Pareto optimal solution is the solution in multi-objective optimization where there is no better alternative, while the equilibrium solution is the engineering feasible solution that takes into account all objectives.

[0041] In this embodiment, the algorithm optimization objective and parameter constraints are set, an initial population is generated, the target index is obtained through the model, and non-dominated solutions are collected through sorting, screening, and evolutionary iteration. Then, the equilibrium solution is selected according to the evaluation rules as the optimal process parameters.

[0042] Based on the above steps, globally optimal parameters that balance sealing quality and equipment durability are obtained, avoiding the pitfalls of single-objective optimization and enhancing the application value of process parameters.

[0043] Based on the above technical solution, it is possible to achieve objective and precise quantification of sealing tightness, and to coordinate the optimization of multiple objectives, including sealing quality and equipment stress. This solves the technical problem that existing technologies only focus on a single objective in the optimization process of can sealing process parameters. Furthermore, by using the DFNN proxy model, the efficiency of multi-objective optimization is improved, and the globally optimal process parameters are quickly obtained, thereby improving the sealing reliability of cans and the durability of equipment.

[0044] In one possible implementation of the embodiments of this application, combined with Figure 1 ,like Figure 2 As shown, the above S103 can be implemented through the following S201 and S202, which are explained in detail below: S201. Extract the number of peaks of the cover hook after the roll seal through post-processing.

[0045] Among them, the first roll seal is the first stage of the double roll seal process, the cover hook is the structure on the edge of the can cover used to roll with the can body, the crest is the protruding structure formed at the cover hook after the first roll seal, and the post-processing refers to the operation of data extraction and analysis of the finite element simulation results.

[0046] In the embodiment of the present application, after the finite element simulation of the first crimping is completed, the cross-sectional shape data of the cover hook is called, the cover hook area is located through the graphical analysis tool of the software, the effective wave peaks are identified, and the number thereof is counted.

[0047] As an example, as shown in Figure 3 , the distribution of the wave peaks at the cover hook after the first crimping is shown, and the position and number characteristics of the extractable wave peaks are intuitively presented. The number of the wave peaks in the statistical chart is 6.

[0048] Based on the above steps, the core quantitative index of tightness is directly obtained, the accuracy and repeatability of the extraction of the wave peak number are ensured, and the basic data for the tightness evaluation is provided.

[0049] S202, if the number of the wave peaks is the same, the maximum residual wave peak height after the second crimping is extracted.

[0050] The second crimping is the second stage of the double crimping process, and the maximum residual wave peak height refers to the maximum vertical height of the residual wave peak at the cover hook after the second crimping.

[0051] In the embodiment of the present application, the number of the wave peaks of multiple simulation results is compared first. If the number is consistent, the cross-sectional data of the cover hook after the second crimping of the corresponding working condition is called, the vertical distance between the wave peak vertex and the adjacent wave valley is selected through the software measurement tool, and the maximum value is recorded.

[0052] It should be noted that the measurement needs to be performed at the same cross-sectional position, and the accuracy of the measurement tool needs to match the grid accuracy of the simulation model to avoid data distortion due to measurement position or accuracy problems.

[0053] As an example, as shown in Figure 4 , the cross-section of the cover hook after the second crimping is called in the ANSYS post-processing, the residual height of each wave peak is measured, and the residual height of the wave peak corresponding to the 100% area is taken as the maximum residual wave peak height.

[0054] Based on the above steps, the tightness quantitative dimension is further refined for the working condition with the same number of wave peaks, and the accurate extraction of the residual wave peak height is realized.

[0055] Based on the above technical solution, the layered quantification of the crimping tightness is realized, the problem that a single index cannot accurately represent the tightness is solved, and high-quality quantitative indexes are provided for the construction of the subsequent tightness evaluation data set.

[0056] In a possible implementation manner of the embodiment of the present application, in combination with Figure 1 , as shown in Figure 5As shown, S104 can be implemented by S501 and S502, which are described below. S501, based on the tightness evaluation dataset, a simulation experiment is designed by using an L16 orthogonal table to obtain training samples.

[0057] The L16 orthogonal table is a 4-factor 4-level orthogonal experimental design table, which can uniformly cover the parameter space within a limited number of experiments; the training sample is a structured data pair containing the key process parameters and the corresponding target process parameters and tightness index.

[0058] In the embodiments of the present application, the data in the tightness evaluation dataset is selected, the factors of the L16 orthogonal table are four key process parameters, the levels are the reasonable value intervals of each parameter, and then finite element simulation is carried out according to the experimental combination of the orthogonal table. The target process parameters obtained by simulation, the extracted tightness index and the corresponding key process parameters are associated to form training samples.

[0059] As an example, taking four key process parameters of one roller feed number (3 / 4 / 5 / 6 turns), two roller feed number (1 / 2 / 3 / 4 turns) and the like as factors, each with four levels, 16 simulation experiments are designed by using an L16 orthogonal table. The lamination rate, wave crest number and maximum contact stress data are obtained in each experiment, and 16 training samples are formed in combination with the tightness index.

[0060] Based on the above steps, comprehensive training samples are obtained with fewer experiments, the samples have the characteristics of uniform dispersion and neat comparison, the data acquisition cost is reduced, and the representativeness of the samples is ensured.

[0061] S502, training the DFNN proxy model based on the training samples.

[0062] The DFNN proxy model is a deep feedforward neural network, including an input layer, four hidden layers and an output layer. The input layer has four nodes (corresponding to four key process parameters), the output layer has three nodes (corresponding to three target process parameters), and the number of nodes in the four hidden layers is 8, 16, 16 and 8, respectively.

[0063] In the embodiments of the present application, the training samples are first normalized, abnormal value cleaning and pretreated, the training set and the validation set are divided, the optimizer, the learning rate and the loss function are configured, and the model is iteratively trained.

[0064] It should be pointed out that the specific process of iteratively training the model is not limited in the present application.

[0065] As an example, after normalizing the 16 training samples, the training set and the validation set are divided according to 7:3, the Adam optimizer (learning rate 0.001) is used, and the mean square error is used as the loss function. The model is iteratively trained as follows: Figure 6The DFNN model shown is trained.

[0066] Based on the above steps, the mapping relationship between the key process parameters and the target process parameters can be accurately captured, replacing the high-cost finite element simulation and greatly improving the calculation efficiency of subsequent optimization.

[0067] Based on the above technical solution, a comprehensive training sample is efficiently obtained through an L16 orthogonal table, and a high-precision DFNN proxy model is combined, both of which cooperatively reduce the experimental and calculation costs and ensure the reliability and fitting accuracy of the model, thereby providing efficient and accurate calculation support for subsequent multi-objective optimization.

[0068] In a possible implementation manner of the embodiment of the application, the NSGA-II algorithm is combined with the DFNN proxy model to obtain a series of Pareto optimal solutions. Figure 1 As shown in Figure 7 As shown in the above S105, the DFNN proxy model is used to obtain a series of Pareto optimal solutions through the NSGA-II algorithm, which can be implemented through the following S701 to S705, which will be described in detail below. S701, a plurality of sets of key process parameter combinations meeting process constraints are generated.

[0069] The process constraint refers to the parameter value range set based on the production equipment capacity, material characteristics and industry standards.

[0070] In the embodiment of the application, the constraint intervals of the key process parameters are first determined (for example, the first roller feed number is 3-7, the friction coefficient is 0.1-0.4, etc.), and then a plurality of parameter combinations meeting all the constraint conditions are generated.

[0071] As an example, the process constraints are set as the first roller feed number 3-7, the second roller feed number 1-5, the tank body speed 100-220 rad / s, and the friction coefficient 0.1-0.4, and 100 sets of key process parameter combinations are generated by using Latin hypercube sampling.

[0072] Based on the above steps, a compliant and comprehensive initial parameter combination is quickly obtained, which provides a basic sample conforming to actual production for subsequent optimization and avoids the occupation of calculation resources by invalid parameters.

[0073] S702, each set of key process parameter combination is input into the trained DFNN proxy model to obtain the corresponding target process parameter.

[0074] The target process parameter is a core index for measuring the sealing quality and equipment loss, including the overlap rate, the number of wave peaks, and the maximum contact stress, which correspond to the evaluation requirements of "sealing reliability", "tightness", and "equipment durability", respectively.

[0075] In the embodiments of the present application, each generated set of key process parameters is sequentially combined and input into the trained DFNN agent model, and the model outputs the specific values of the three target process parameters corresponding to each set of combinations through a preset parameter mapping relationship.

[0076] As an example, the 100 sets of key process parameter combinations generated in S701 are sequentially input into the DFNN agent model, and the model outputs 100 corresponding target process parameters, wherein the overlap rate ranges from 53% to 60%, the number of wave crests ranges from 60 to 85, and the maximum contact stress ranges from 205 to 285 MPa.

[0077] Based on the above steps, a large amount of target process parameter data is quickly and efficiently obtained, the parameter evaluation period is greatly shortened, and the overall efficiency of the optimization process is improved.

[0078] S703, taking the key process parameters as a population, non-dominant sorting is performed on the individuals in the population, and high-quality individuals are screened.

[0079] Among them, the population refers to a set composed of all key process parameter combinations, each parameter combination is an "individual"; non-dominant sorting refers to a process of judging whether an individual is dominated by other individuals (if the overlap rate and the number of wave crests of individual A are not lower than those of individual B, and the maximum contact stress is not higher than that of individual B, then A dominates B), and dividing Pareto levels according to the domination relationship; high-quality individuals refer to individuals in high Pareto levels, which are not dominated by other individuals or have very low domination degree.

[0080] In the embodiments of the present application, all individuals in the population are traversed, and the target process parameter values of each individual are compared with each other to determine the domination and dominated relationship of each individual, and multiple Pareto levels are divided. The first 1-2 high-level individuals are selected as high-quality individuals and reserved to the next generation population. The optimization target is to maximize the overlap rate, maximize the number of wave crests, and minimize the maximum contact stress.

[0081] As an example, 100 individuals are taken as a population, and the overlap rate, the number of wave crests, and the maximum contact stress are compared with each other to divide 3 Pareto levels. 30 non-dominated individuals in the first level are selected as high-quality individuals.

[0082] Based on the above steps, individuals with better performance are accurately screened out, and dominated low-efficiency parameter combinations are eliminated, effectively improving the overall quality of the population and providing high-quality parent samples for subsequent evolution.

[0083] S704, a new generation population is generated through selection, crossover, and mutation operations, and the step of screening high-quality individuals is repeatedly executed until a preset evolution generation number is reached.

[0084] The selection operation is a process of selecting parent individuals from high-quality individuals, the crossover operation is a process of recombining parameter fragments of two parent individuals to generate offspring individuals, and the mutation operation is a process of randomly adjusting part of parameter values of the offspring individuals.

[0085] In the embodiment of the application, parent individuals are selected from high-quality individuals through tournament selection or roulette selection, then the parameters of the parents are recombined in a single-point crossover or multi-point crossover manner, and finally part of the parameter values of the offspring are randomly adjusted according to a preset mutation probability to generate a new population; the non-dominated sorting and high-quality individual screening steps of S701 to S703 are executed in a loop until the number of population iterations reaches the preset evolution generation, the population size is set to 100, and the preset evolution generation is 200.

[0086] As an example, parent individuals are selected from 30 high-quality individuals by using tournament selection (selection pressure is 2), the crossover probability is set to 0.8, the mutation probability is set to 0.1, the offspring are generated by single-point crossover, and the parameter values are randomly adjusted (for example, the number of roller feed circles is ±0.2 circles) during mutation to generate a new population of 100 individuals, and the screening step is executed in a loop until 200 iterations.

[0087] Based on the above steps, the population is evolved towards the optimal solution while the population diversity is maintained, the local optimal trap is avoided, and the comprehensiveness and depth of the optimization process are ensured.

[0088] S705, collect all non-dominated solutions in the iteration process to form a series of Pareto optimal solutions.

[0089] The non-dominated solution refers to a parameter combination that is not dominated by any other individual in the iteration process, and the series of Pareto optimal solutions refers to a set formed by de-duplication and integration of all non-dominated solutions.

[0090] In the embodiment of the application, all non-dominated solutions in each population are recorded, and de-duplication is performed on all recorded non-dominated solutions after the iteration is completed to form a final set of Pareto optimal solutions.

[0091] As an example, as shown in a Pareto front diagram obtained by the NSGA-II optimization, Figure 8 As shown in a Pareto front diagram obtained by the NSGA-II optimization,

[0092] Based on the above steps, all optimal solutions in the iteration process are integrated to form a set of Pareto optimal solutions covering the multi-objective trade-off relationship, and sufficient and high-quality options are provided for accurate selection of subsequent equilibrium solutions.

[0093] Based on the above technical scheme, the pareto optimal solution set considering the sealing quality, tightness and equipment durability is efficiently obtained, the efficiency and comprehensiveness of optimization are ensured, and clear basis is provided for subsequent balanced solution selection, and the problems of low efficiency and solution set in traditional multi-objective optimization are effectively solved.

[0094] In a possible implementation manner of the embodiment of the application, the method is combined with Figure 1 As shown in the S105, the selection of the balanced solution from the series of pareto optimal solutions as the optimal process parameter can be implemented by the following S901, S902 and S903, which are described in detail as follows: Figure 9 As shown in the S105, the selection of the balanced solution from the series of pareto optimal solutions as the optimal process parameter can be implemented by the following S901, S902 and S903, which are described in detail as follows: S901, obtaining the extreme value of the target process parameter in the series of pareto optimal solutions.

[0095] The extreme value refers to the boundary value of each target process parameter in the pareto optimal solution set, including the maximum value of the overlap rate, the maximum value of the wave peak number, the minimum value and the maximum value of the maximum contact stress.

[0096] In the embodiment of the application, all the pareto optimal solutions are traversed, the values of the overlap rate, the wave peak number and the maximum contact stress are extracted, the extreme values of each index are screened out and recorded.

[0097] It should be noted that it is necessary to ensure that all the solutions are traversed without omission to avoid the influence of the extreme value deviation on subsequent calculation.

[0098] As an example, after traversing the 34 pareto optimal solutions, the maximum value of the overlap rate is 60%, the maximum value of the wave peak number is 85, the minimum value of the maximum contact stress is 205 MPa, and the maximum value is 285 MPa.

[0099] Based on the above steps, the unified reference benchmark of the target parameter is obtained, which provides a reliable basis for subsequent normalized calculation.

[0100] S902, based on the extreme value, calculating the score value of each normalized pareto optimal solution.

[0101] The normalization is a processing method of mapping the target process parameter value to the 0-1 interval, and the score value is a quantitative index for comprehensively evaluating the multi-objective performance of each optimal solution.

[0102] In the embodiment of the application, the target parameters of each solution are normalized according to the extreme value, the overlap rate and the wave peak number are normalized in a positive direction, and the maximum contact stress is normalized in a reverse direction, and then the comprehensive score value is calculated in combination with the weight.

[0103] Optionally, the score value calculation satisfies the following formula:

[0104] S is the score value, is a lapping rate, is a number of wave crests, is a maximum contact stress, is a maximum value of the lapping rate, is a maximum value of the number of wave crests, is a minimum value of the maximum contact stress, is a maximum value of the maximum contact stress, is a lapping rate weight, is a number of wave crests weight, is a maximum contact stress weight.

[0105] It should be noted that the weights can be adjusted according to the preliminary relationship in S102, and the normalization processing can eliminate the influence of dimensional difference.

[0106] As an example, let = 0.45, = 0.35, = 0.2, the normalized lapping rate of a certain solution is 0.97, the normalized number of wave crests is 0.79, and the normalized maximum contact stress is 0.82, and the calculated score is 0.92.

[0107] Based on the above steps, the comprehensive evaluation of multi-objective performance is realized in a quantitative way, and the advantages and disadvantages of each optimal solution are more intuitive.

[0108] S903, selecting the key process parameters corresponding to the Pareto optimal solution with the highest score value as the optimal process parameters.

[0109] Among them, the optimal process parameters refer to the key process parameter combination that is optimal in comprehensive multi-objective performance and meets the requirements of engineering practical application.

[0110] In the embodiments of the present application, the score values of all Pareto optimal solutions are compared, the solution with the highest score is selected, and its corresponding one-way roller feed number, two-way roller feed number, tank body speed, and friction coefficient key process parameters are extracted.

[0111] As an example, as shown in Table 1, the highest score among the 34 optimal solutions is 0.95, and the corresponding key process parameters are one-way roller feed 4.47 turns, two-way roller feed 2 turns, friction coefficient 0.19, and tank body speed 161.85 rad / s.

[0112] It should be noted that the key process parameters before optimization are used to generate multiple key process parameter combinations that meet the process constraints in S701, and the corresponding target process parameters are the output values of the DFNN proxy model.

[0113] Table 1

[0114] Based on the above steps, the optimal parameters of multi-objective balance are quickly locked, the deviation of subjective selection is avoided, and the reliability of process parameters is improved.

[0115] Based on the above technical scheme, through the processes of extracting extreme values, normalized score calculation, and highest score screening, objective and accurate screening of the Pareto optimal solution is realized, the final optimal process parameters consider sealing quality, tightness, and equipment durability, and the subjectivity problem of balanced solution selection in multi-objective optimization is solved.

[0116] The above describes the scheme of the embodiments of the present application mainly from the perspective of device implementation. It can be understood that each device, for example, the easy-open can double-seal process parameter optimization device, includes at least one of the corresponding hardware structure and software module for executing each function. Those skilled in the art should easily realize that, in combination with the units and algorithm steps of each example described in the embodiments disclosed herein, the present application can be realized in the form of hardware or a combination of hardware and computer software. Whether a certain function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical scheme. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0117] The embodiments of the present application can divide the functional units of the easy-open can double-seal process parameter optimization device according to the above method examples. For example, each functional unit can be divided according to each function, or two or more functions can be integrated in one processing unit. The integrated unit can be realized in the form of hardware or software functional unit. It should be noted that the division of units in the embodiments of the present application is illustrative, and is only a logical functional division. Actual implementation can have another division method.

[0118] In the case of using integrated units, Figure 10 A possible structural schematic diagram of the easy-open can double-seal process parameter optimization device (denoted as easy-open can double-seal process parameter optimization device 100) involved in the above embodiments is shown, which includes a processing unit 1001 and a communication unit 1002, and can also include a storage unit 1003. Figure 10 The structural schematic diagram shown can be used to illustrate the structure of the easy-open can double-seal process parameter optimization device involved in the above embodiments.

[0119] When Figure 10The structure diagram shown is used to show the structure of the easy-open can double-seam process parameter optimization device involved in the above embodiment. The processing unit 1001 is used to control and manage the action of the easy-open can double-seam process parameter optimization device. The communication unit 1002 is used for communication between the easy-open can double-seam process parameter optimization device and other devices. The storage unit 1003 is used to store the program code and data of the easy-open can double-seam process parameter optimization device.

[0120] For example, the communication unit 1002 is configured to obtain key process parameters and target process parameters. The key process parameters include the number of first roller feed circles, the number of second roller feed circles, the can body rotation speed, and the friction coefficient. The target process parameters include the overlap rate, the number of wave peaks, and the maximum contact stress. The processing unit 1001 is configured to perform finite element simulation on the key process parameters to obtain a preliminary relationship between the key process parameters and the target process parameters. In the finite element simulation, the number of wave peaks and the maximum residual wave peak height are extracted to construct a tightness evaluation dataset. The tightness evaluation dataset includes the key process parameters, the target process parameters, and the maximum residual wave peak height. A DFNN surrogate model is trained based on the tightness evaluation dataset. The DFNN surrogate model is used to represent the mapping relationship between the key process parameters and the target process parameters. Based on the DFNN surrogate model, a series of Pareto optimal solutions are obtained by using the NSGA-II algorithm. One balanced solution is selected from the series of Pareto optimal solutions as the optimal process parameters.

[0121] In a possible implementation, the processing unit 1001 is further configured to perform finite element simulation on the key process parameters to obtain a preliminary relationship between the key process parameters and the target process parameters, including: adjusting the values of the key process parameters respectively and repeating the finite element simulation; recording the changes of the target process parameters corresponding to different key process parameter values to obtain the preliminary relationship between the key process parameters and the target process parameters.

[0122] In a possible implementation, the processing unit 1001 is further configured to extract the number of wave peaks and the maximum residual wave peak height in the finite element simulation, including: extracting the number of wave peaks after the first sealing by post-processing; if the number of wave peaks is the same, extracting the maximum residual wave peak height after the second sealing.

[0123] In a possible implementation, the processing unit 1001 is further configured to train a DFNN surrogate model based on the tightness evaluation dataset, including: based on the tightness evaluation dataset, using an L16 orthogonal table to design a simulation experiment to obtain training samples; and training the DFNN surrogate model based on the training samples.

[0124] In a possible implementation, the DFNN agent model comprises one input layer, one output layer, and four hidden layers; the input layer has four nodes; the output layer has three nodes; and the four hidden layers have 8 nodes, 16 nodes, 16 nodes, and 8 nodes in sequence.

[0125] In a possible implementation, the processing unit 1001 is further configured to obtain a series of Pareto optimal solutions based on the DFNN agent model and using the NSGA-II algorithm, including: generating a plurality of sets of key process parameter combinations that meet process constraints; inputting each set of key process parameter combinations into the trained DFNN agent model to obtain corresponding target process parameters; taking the key process parameters as a population, performing non-dominated sorting on individuals in the population, and screening high-quality individuals; generating a new generation of population through selection, crossover, and mutation operations, and repeatedly performing the step of screening high-quality individuals until a preset evolution number is reached; and collecting all non-dominated solutions in the iteration process to form a series of Pareto optimal solutions.

[0126] In a possible implementation, the population size of the NSGA-II algorithm is 100, the preset evolution number is 200, and the optimization objectives are to maximize the overlap rate, maximize the number of wave peaks, and minimize the maximum contact stress.

[0127] In a possible implementation, the processing unit 1001 is further configured to select one balanced solution as the optimal process parameters from the series of Pareto optimal solutions, including: obtaining extreme values of the target process parameters in the series of Pareto optimal solutions; the extreme values include the maximum value of the overlap rate, the maximum value of the number of wave peaks, the minimum value of the maximum contact stress, and the maximum value of the maximum contact stress; calculating a score value of each normalized Pareto optimal solution based on the extreme values; and selecting the key process parameters corresponding to the Pareto optimal solution with the highest score value as the optimal process parameters.

[0128] In a possible implementation, the calculation of the score value of each normalized Pareto optimal solution satisfies the following formula:

[0129] wherein S is the score value, is the overlap rate, is the number of wave peaks, is the maximum contact stress, is the maximum value of the overlap rate, is the maximum value of the number of wave peaks, is the minimum value of the maximum contact stress, is the maximum value of the maximum contact stress, is the weight of the overlap rate, is the weight of the number of wave peaks, is the weight of the maximum contact stress.

[0130] The processing unit 1001 can be a processor or a controller, and the communication unit 1002 can be a communication interface, a transceiver, a transceiver, a transceiver circuit, a transceiver device, etc. The communication interface is collectively referred to, and can include one or more interfaces. The storage unit 1003 can be a memory. When the easy-open can two-seam seaming process parameter optimization device 100 is a chip, the processing unit 1001 can be a processor or a controller, and the communication unit 1002 can be an input interface and / or an output interface, a pin or a circuit, etc. The storage unit 1003 can be a storage unit (for example, a register, a cache, etc.) within the chip, or a storage unit (for example, a read-only memory (ROM), a random access memory (RAM), etc.) located outside the chip.

[0131] The communication unit can also be referred to as a transceiving unit. The antenna and control circuit with transceiving function in the easy-open can two-seam seaming process parameter optimization device 100 can be regarded as the communication unit 1002 of the easy-open can two-seam seaming process parameter optimization device 100, and the processor with processing function can be regarded as the processing unit 1001 of the easy-open can two-seam seaming process parameter optimization device 100. Optionally, the device for realizing the receiving function in the communication unit 1002 can be regarded as a communication unit, and the communication unit is used to execute the receiving steps in the embodiments of the application, and the communication unit can be a receiver, a receiver, a receiving circuit, etc. The device for realizing the sending function in the communication unit 1002 can be regarded as a sending unit, and the sending unit is used to execute the sending steps in the embodiments of the application, and the sending unit can be a transmitter, a transmitter, a sending circuit, etc.

[0132] Figure 10 The integrated units in the above embodiments can be stored in a computer readable storage medium if they are realized in the form of software function modules and sold or used as independent products. Based on such understanding, the technical solutions of the embodiments of the application or the whole or part of the technical solutions that make essential contributions to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to execute all or part of the steps of the method described in the embodiments of the application. The storage medium storing the computer software product includes a U disk, a mobile hard disk, a read-only memory, a random access memory, a magnetic disk or an optical disk, and various media that can store program codes.

[0133] Figure 10 The units in the above embodiments can also be referred to as modules, for example, the processing unit can be referred to as a processing module.

[0134] Although the application has been described in connection with the embodiments thereof with reference to the various drawings, it will be understood that other variations and modifications of the details, and specific examples can be resorted to by those skilled in the art without departing from the spirit and scope of the application. In its broadest form, the application is directed to all new and useful processes, machines, articles of manufacture, compositions of matter, and methods that fall within the scope of the claims. The specification and drawings are, accordingly, to be regarded as illustrative rather than restrictive. For a better understanding of the application, its operating advantages, and the specific objects attained by its uses, reference should be made to the drawings and to the implementation in the claims. In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite articles "a" or "an" do not exclude a plurality. A single processor or other unit can fulfil the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

[0135] Although the application has been described in connection with specific embodiments thereof, it will be understood that it is capable of modifications and alternative constructions and combinations of parts herein described, drawing upon the spirit and scope of the application. Accordingly, the specification and drawings are to be regarded as illustrative in nature and not as restrictive. Obviously, modifications and variations of the present application are possible in light of the above teachings. It is, therefore, to be understood that changes can be made in the particular embodiments of the application recited herein and still be within the scope of the application.

Claims

1. A method for optimizing process parameters of double-sealing process for aluminum cans, characterized in that, include: Obtain key process parameters and target process parameters; the key process parameters include the number of feed revolutions of the first roller, the number of feed revolutions of the second roller, the tank rotation speed, and the coefficient of friction. The target process parameters include overlap ratio, number of peaks, and maximum contact stress. Finite element simulation was performed on the key process parameters to obtain the preliminary relationship between the key process parameters and the target process parameters; In the finite element simulation, the number of peaks and the maximum residual peak height are extracted to construct a compactness evaluation dataset; the compactness evaluation dataset includes the key process parameters, target process parameters, and the maximum residual peak height; A DFNN proxy model is trained based on the density evaluation dataset; the DFNN proxy model is used to characterize the mapping relationship between the key process parameters and the target process parameters; Based on the DFNN surrogate model, the NSGA-II algorithm is used to obtain a series of Pareto optimal solutions, and an equilibrium solution is selected from the series of Pareto optimal solutions as the optimal process parameters.

2. The method according to claim 1, characterized in that, Finite element simulations are performed on the key process parameters to obtain a preliminary relationship between the key process parameters and the target process parameters, including: Adjust the values ​​of each key process parameter and repeat the finite element simulation. Record the changes in the target process parameters corresponding to different values ​​of key process parameters, and obtain the preliminary relationship between key process parameters and target process parameters.

3. The method according to claim 1, characterized in that, In the finite element simulation, the number of wave crests and the maximum residual wave crest height are extracted, including: The number of wave peaks of the cover hook after the roll seal is extracted through post-processing; If the number of peaks is the same, then the maximum residual peak height after two roll seals is extracted.

4. The method according to claim 1, characterized in that, Training a DFNN proxy model based on the aforementioned tightness evaluation dataset includes: Based on the density evaluation dataset, an L16 orthogonal array was used to design a simulation experiment to obtain training samples. The DFNN proxy model is trained based on the training samples.

5. The method according to claim 4, characterized in that, The DFNN surrogate model includes an input layer, an output layer, and four hidden layers; the input layer has four nodes; the output layer has three nodes; and the four hidden layers have the following structures in sequence: eight nodes, 16 nodes, 16 nodes, and eight nodes.

6. The method according to claim 1, characterized in that, Based on the DFNN surrogate model, the NSGA-II algorithm is used to obtain a series of Pareto optimal solutions, including: Generate multiple sets of key process parameter combinations that meet process constraints; Each set of key process parameters is input into the trained DFNN proxy model to obtain the corresponding target process parameters; The key process parameters are used as a population, and individuals in the population are sorted non-dominated to select high-quality individuals. A new generation of population is generated through selection, crossover, and mutation operations. The process of selecting high-quality individuals is repeated until the preset number of generations is reached. Collect all non-dominated solutions during the iteration process to form a series of Pareto optimal solutions.

7. The method according to claim 6, characterized in that, The NSGA-II algorithm has a population size of 100 and a preset number of generations of evolution of 200. The optimization objectives are to maximize the overlap rate, maximize the number of peaks, and minimize the maximum contact stress.

8. The method according to claim 1, characterized in that, Selecting an equilibrium solution from the series of Pareto optimal solutions as the optimal process parameters includes: Obtain the extreme values ​​of the target process parameters in the series of Pareto optimal solutions; the extreme values ​​include the maximum value of the overlap ratio, the maximum value of the number of peaks, and the minimum and maximum values ​​of the maximum contact stress; Based on the extreme values, calculate the score for each normalized Pareto optimal solution; The key process parameters corresponding to the Pareto optimal solution with the highest score are selected as the optimal process parameters.

9. The method according to claim 8, characterized in that, The score for each normalized Pareto optimal solution is calculated according to the following formula: Where S is the score value. The overlap ratio is... The number of wave peaks, The maximum contact stress, This is the maximum value of the overlap ratio. The maximum value of the number of wave peaks. This is the minimum value of the maximum contact stress. The maximum value of the maximum contact stress. Weighted by overlap rate. Weighted by the number of peaks. This represents the maximum contact stress weight.

10. A device for optimizing process parameters of double-sealing process for beverage cans, characterized in that, The device includes: a communication unit and a processing unit; The communication unit is used to acquire key process parameters and target process parameters; the key process parameters include the number of feed revolutions of the first roller, the number of feed revolutions of the second roller, the tank rotation speed, and the friction coefficient; the target process parameters include the overlap ratio, the number of peaks, and the maximum contact stress. The processing unit is used to perform finite element simulation on the key process parameters to obtain a preliminary relationship between the key process parameters and the target process parameters; in the finite element simulation, the number of peaks and the maximum residual peak height are extracted to construct a compactness evaluation dataset; the compactness evaluation dataset includes the key process parameters, the target process parameters, and the maximum residual peak height; a DFNN surrogate model is trained based on the compactness evaluation dataset; the DFNN surrogate model is used to characterize the mapping relationship between the key process parameters and the target process parameters; based on the DFNN surrogate model, the NSGA-II algorithm is used to obtain a series of Pareto optimal solutions, and an equilibrium solution is selected from the series of Pareto optimal solutions as the optimal process parameters.

Images