A Method for Optimizing 3D Printing Process Parameters of a Protective Face Mask

Abstract

The present invention relates to the field of data processing. More specifically, the present invention relates to a method for optimizing the process parameters of 3D printing of a protective face mask. The method includes: obtaining the feasible value range of each process parameter for 3D printing of the protective face mask, and taking the spatial range formed by the feasible value ranges of all process parameters as the solution space; using a genetic algorithm to iteratively optimize based on the solution space until the stopping condition is met to obtain the optimal solution, and taking the process parameters in the optimal solution as the optimized process parameters; in one optimization process, taking the quality index of the protective face mask corresponding to each current feasible solution as the fitness, calculating the selection possibility of any current feasible solution, and selecting part of the current feasible solutions as the retained individuals according to the selection possibility. Avoid falling into local optima and more accurately locate the appropriate process parameters.

Description

Technical Field

[0001] The present invention relates to the field of data processing. More specifically, the present invention relates to a method for optimizing the 3D printing process parameters of a protective face mask. Background Art

[0002] With the advantages of rapid response and flexible production, 3D printing technology has played an important role in the manufacturing of personal protective equipment (PPE) such as protective face masks. However, the quality and performance of 3D printed protective face masks are affected by various process parameters. For example, an unreasonable printing temperature setting may lead to a decrease in strength or surface roughness. Therefore, it is necessary to optimize the process parameters to ensure their protective effect and wearing comfort.

[0003] Due to the differences in the types, sizes, and printing materials of protective face masks, there are differences in the processing process parameters of different protective face masks. In order to set appropriate process parameters according to the individual processing requirements of protective face masks, it is necessary to evaluate the quality and performance of the protective face masks obtained from each process parameter to obtain better process parameters. As an optimization algorithm, the genetic algorithm can effectively improve the optimization efficiency. However, this algorithm is easily trapped in a local optimum during the optimization process, resulting in the inability to obtain better process parameters. How to jump out of the local optimum and obtain better process parameters has become the research focus of the present invention.

[0004] The patent document with the authorization announcement number CN118428244B discloses a method for adaptively optimizing the reflow soldering process parameters. The key point of the method in this patent document is to achieve the adaptive adjustment of the reflow soldering process parameters by adjusting the reflow optimization step size. The method in this patent document does not involve how to solve the problem of jumping out of the local optimum. Therefore, the method in this patent document cannot solve the technical problems of this solution. Summary of the Invention

[0005] To solve the problem of how to jump out of the local optimum and obtain better process parameters, the present invention proposes a method for optimizing the 3D printing process parameters of a protective face mask. The method includes the following steps:

[0006] Obtain the feasible value range of each process parameter for 3D printing of a protective face mask, and use the spatial range formed by the feasible value ranges of all process parameters as the solution space;

[0007] Based on the solution space, use the genetic algorithm to iteratively optimize until the stop condition is met, obtain the optimal solution, and use the process parameters in the optimal solution as the optimized process parameters;

[0008] During an optimization process, the quality indicators of the protective masks corresponding to each current feasible solution are used as fitness values. The solution space is divided into several subspaces, and the exploration degree is calculated based on the distribution of the feasible solutions traversed in the previous iteration process in the subspaces. The variation degree is calculated based on the variation differences of the traversed feasible solutions in different regions within the subspaces. The selection probability of the current feasible solution is calculated based on the exploration degree, the variation degree, the fitness value of the current feasible solution, and the predicted fitness value at the peak position of the peak where the current feasible solution is located. Among them, the selection probability is positively correlated with the variation degree, the fitness value of the current feasible solution, and the predicted fitness value at the peak position of the peak where the current feasible solution is located, and is negatively correlated with the exploration degree.

[0009] Based on the selection probability, some of the current feasible solutions are selected as the retained individuals.

[0010] The present invention controls the retention of individuals by adjusting the selection probability of each current feasible solution, thereby effectively jumping out of the local optimum and more accurately locating the global optimum. Further, when calculating the selection probability, the exploration degree of the subspace is considered, so as to explore each subspace in the solution space to a certain extent, thereby effectively preventing the situation of falling into the local optimum caused by insufficient exploration of some subspaces. Further, when calculating the selection probability, a peak prediction part is introduced, so as to effectively solve the situation where the current feasible solution cannot reflect the existence of the optimal solution in each subspace, so as to set a greater selection probability for the current feasible solution in the subspace with a greater possibility of the existence of the optimal solution, avoid falling into the local optimum, and more accurately obtain the optimal solution.

[0011] Preferably, the selection probability of the current feasible solution satisfies the relational expression:

[0012] ;

[0013] wherein, T represents the exploration degree of the subspace where the current feasible solution is located, C represents the variation degree in the subspace where the current feasible solution is located, represents the predicted fitness value at the peak position of the peak where the current feasible solution is located, represents the prediction accuracy of the predicted fitness value, represents the fitness value of the current feasible solution, represents the selection probability of the current feasible solution.

[0014] Preferably, the division of the solution space into several subspaces includes:

[0015] The solution space is evenly divided into a preset first number of subspaces.

[0016] The present invention obtains subspaces by means of uniform division, effectively avoiding the inaccurate analysis caused by the volume difference of subspaces, and providing a basis for subsequent accurate analysis.

[0017] Preferably, the method for obtaining the exploration degree includes:

[0018] Obtain the positions of the feasible solutions traversed in the previous iteration process and record them as the traversed positions. Divide each subspace evenly into a preset second number of small spaces and record them as analysis spaces. Count the number of traversed positions included in each analysis space, and take the reciprocal of the variance of the number of traversed positions in all analysis spaces of each subspace as the exploration uniformity degree of each subspace;

[0019] Multiply the number of traversed positions included in each subspace by the exploration uniformity degree to obtain the exploration degree of each subspace.

[0020] When analyzing the exploration degree of the present invention, not only the exploration uniformity degree is introduced, but also the number of traversed positions is introduced, so as to more comprehensively and accurately analyze the exploration situation of each sub-region and provide a basis for subsequent analysis.

[0021] Preferably, the method for obtaining the variation difference degree includes:

[0022] Use the traversed feasible solutions and the corresponding fitness values in the solution space to fit a polynomial, which is recorded as the overall relationship; divide the fitting model of the overall variation relationship at the subspace into several independent peaks, obtain the kurtosis and skewness of each independent peak, multiply the kurtosis difference between every two independent peaks by the skewness difference to obtain the rule difference degree between the two independent peaks, and take the mean value of the rule difference degrees of all two independent peaks as the rule difference degree of the subspace;

[0023] Take the mean value of the interval distances between each independent peak and its surrounding independent peaks as the interval distance of each independent peak, and take the variance of the interval distances of all independent peaks in the subspace as the distribution difference degree of the subspace;

[0024] Multiply the rule difference degree of the subspace by the distribution difference degree to obtain the variation difference degree of the subspace.

[0025] When analyzing the variation difference degree of the present invention, not only the rule difference degree is introduced to reflect the fluctuation difference, but also the distribution interval difference is introduced to reflect the interval distribution information of different fluctuation rules, and the variation difference situation in the subspace is evaluated more comprehensively and accurately.

[0026] Preferably, the step of dividing the fitting model of the overall variation relationship at the subspace into several independent peaks includes:

[0027] Obtain the closed region enclosed by the positions corresponding to the minimum points of the overall change relationship as the independent region, and take the part of the relationship model corresponding to the overall change relationship within the independent region as the independent peak.

[0028] Preferably, the prediction accuracy of the fitness prediction value includes:

[0029] Obtain the fitness prediction values of each feasible solution within the independent peak where the current feasible solution is located, take the absolute value of the difference between the fitness of each traversed feasible solution and the fitness prediction value as the fitness value prediction deviation of each traversed feasible solution, and take the reciprocal of the fitness prediction deviation of all traversed feasible solutions within the peak where the current feasible solution is located as the prediction accuracy of the fitness prediction value.

[0030] Preferably, the step of selecting some current feasible solutions as reserved individuals according to the selection possibility includes:

[0031] Divide the selection possibility of each current feasible solution by the sum of the selection possibilities of all current feasible solutions to obtain the selection probability of each current feasible solution;

[0032] Take the selection probability of each current feasible solution as the extraction probability, and randomly extract a preset third number of current feasible solutions from all current feasible solutions as reserved individuals.

[0033] Preferably, take the number of iterations reaching the preset cut-off number as the stop condition.

[0034] Preferably, the method for obtaining the quality index of the protective mask includes:

[0035] On the simulation software, use the printing parameters corresponding to each process parameter setting in the current feasible solution to control the printing of the protective mask, and obtain the processed protective mask;

[0036] Conduct various performance simulation tests on the protective mask, obtain the performance scores of each performance test, and take the average value of the performance scores of all performance tests as the quality index of the protective mask.

[0037] The present invention has the following beneficial effects:

[0038] The present invention controls the retention of individuals by adjusting the selection possibility of each current feasible solution, thereby effectively jumping out of the local optimum and more accurately locating the global optimum;

[0039] Furthermore, when calculating the selection possibility, the exploration degree of the subspace is considered, so as to explore each subspace in the solution space to a certain extent, thereby effectively preventing the situation of falling into the local optimum caused by insufficient exploration of some subspaces;

[0040] Furthermore, when calculating the selection possibility, a peak prediction part is introduced, which effectively solves the situation that the current feasible solution cannot reflect the existence of the optimal solution in each subspace, so as to set a greater selection possibility for the current feasible solution in the subspace with a high possibility of the existence of the optimal solution, avoid falling into the local optimum, and obtain the optimal solution more accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 is a flowchart of the steps of a method for optimizing the 3D printing process parameters of a protective mask according to an embodiment of the present invention. DETAILED DESCRIPTION

[0042] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0043] The following will describe in detail the specific implementation manners of the present invention with reference to the accompanying drawings.

[0044] Please refer to Figure 1 , which shows a flowchart of the steps of a method for optimizing the 3D printing process parameters of a protective mask provided by an embodiment of the present invention. The method includes the following steps:

[0045] S1: Obtain the feasible value range of each process parameter for 3D printing of the protective mask, and use the spatial range formed by the feasible value ranges of all process parameters as the solution space.

[0046] Specifically, obtain the feasible value range of each process parameter for 3D printing of the protective mask, and use the spatial range formed by the feasible value ranges of all process parameters as the solution space.

[0047] The types of process parameters include, but are not limited to, the following aspects: printing temperature, printing speed, filling density, and support structure. The feasible value range of each process parameter is the value range of each process parameter obtained based on experience.

[0048] S2: Based on the solution space, use the genetic algorithm to iteratively optimize until the stop condition is met to obtain the optimal solution, and use the process parameters in the optimal solution as the optimized process parameters.

[0049] Preferably, as an example, based on the solution space, use the genetic algorithm to iteratively optimize until the stop condition is met to obtain the optimal solution, and use the process parameters in the optimal solution as the optimized process parameters, including:

[0050] Taking the solution space as the optimization search space, the genetic algorithm is used to perform iterative optimization within the optimization search space until the stop condition is met to obtain the optimal solution, and each process parameter in the optimal solution is used as the optimized process parameter.

[0051] It should be added that in this embodiment, the iteration count reaching the preset cut-off count is used as the stop condition. Other embodiments can set other cut-off conditions, and this embodiment does not make specific limitations.

[0052] Among them, using the genetic algorithm to perform iterative optimization within the optimization search space includes steps S20 - step S21.

[0053] It should be noted that the traditional genetic algorithm selects individuals with large fitness values to be retained for genetic reproduction. In this selection method, it is very easy for some individuals in the local optimum to be retained, making the retained individuals overly concentrated in a small area, thus not searching other areas and resulting in being unable to jump out of the local optimum. For example, in some areas, the individuals being traversed are at the peak, so the fitness value of this individual is large, while in some areas, the individuals being traversed are at the valley, so the fitness value of this individual is small. Therefore, by selecting individuals to be retained based on the fitness values of the individuals being traversed, it is very easy for there to be no individuals retained in some areas, resulting in the phenomenon of missing the optimal solution due to weak search intensity.

[0054] It should be further noted that in order to prevent falling into the local optimum, while ensuring that better individuals are retained, it is also necessary to judge the distribution of the traversed solutions, so as to search each area of the solution space. At the same time, when judging the retention of individuals, not only the fitness value of the currently traversed individuals in each area needs to be considered, but also the maximum value situation at the peak where the individual is located needs to be appropriately estimated, so as to prevent the phenomenon of missing the optimal solution due to the currently traversed individual being at the valley and thus the individual being removed.

[0055] S20: In an optimization process, taking the protective mask quality index corresponding to each current feasible solution as the fitness, calculate the selection possibility of any current feasible solution.

[0056] S200: In an optimization process, taking the protective mask quality index corresponding to each current feasible solution as the fitness.

[0057] Optionally, as an example, taking the protective mask quality index corresponding to each current feasible solution as the fitness includes:

[0058] Among the historical process data, obtain the historical process data with the smallest difference from the current feasible solution. Use the protective mask printed with the historical process data as a reference physical object, conduct various performance tests on the reference physical object, obtain the performance scores for each performance test, and take the average of the performance scores obtained from all performance tests as the quality index of the protective mask corresponding to the current feasible solution.

[0059] It should be noted that there is still a difference between the current feasible solution and the historical process data with the smallest difference. Therefore, the protective mask printed with the historical process data with differences cannot fully reflect the physical object corresponding to the current feasible solution, and thus the obtained quality index of the protective mask is also inaccurate.

[0060] Preferably, as an example, during an optimization process, take the quality index of the protective mask corresponding to each current feasible solution as the fitness, including:

[0061] On the simulation software, use the printing parameters corresponding to each process parameter setting in the current feasible solution to control the printing of the protective mask and obtain the printed protective mask.

[0062] Conduct various performance simulation tests on the protective mask, obtain the performance scores for each performance test, and take the average of the performance scores obtained from all performance tests as the quality index of the protective mask.

[0063] It should be added that the types of performance simulation tests for the protective mask include but are not limited to the following aspects:

[0064] Filtration efficiency, sealing performance, impact resistance, weight index, air permeability index.

[0065] S201: During an optimization process, calculate the selection probability of any current feasible solution.

[0066] It should be noted that to prevent falling into a local optimum, a certain degree of search needs to be ensured in each subspace. At the same time, to prevent the differences in peaks and valleys among the current feasible solutions in different subspaces from causing the fitness values of the current feasible solutions to be unable to reflect the existence of the optimal value in the subspace, it is necessary to estimate the fitness value at the peak position to represent the existence of the optimal value in the subspace, and then more accurately set the selection probability of the current feasible solution.

[0067] Optionally, as an example, during an optimization process, calculate the selection probability of any current feasible solution, including:

[0068]

[0069] Among them, the solution space is divided into several subspaces, and the exploration degree of each subspace is obtained. T represents the exploration degree of the subspace where the current feasible solution is located, and C represents the degree of variation difference in the subspace where the current feasible solution is located. It represents the estimated fitness value at the peak position of the peak where the current feasible solution is located. It represents the fitness value of the feasible solution. It represents the selection possibility of the current feasible solution.

[0070] It can be understood that the degree of variation difference C reflects the difference in the variation laws of different regions in the subspace where the current feasible solution is located. The larger this value, the greater the difference in the variation laws. Therefore, the fitness information at the current feasible solution cannot represent the fitness information of other regions in this subspace. Therefore, it is necessary to explore this subspace in detail to prevent falling into a local optimum. Therefore, it is necessary to increase the selection possibility of the current feasible solution, so that the current feasible solution has a greater probability of being retained, thus providing a data basis for subsequent traversal. It reflects the traversed situation in this subspace. The larger this value, the greater the traversed degree of this subspace. Therefore, the possibility of falling into a local optimum in this subspace is smaller. Therefore, it is necessary to appropriately reduce the selection possibility of the current feasible solution, so that the retention probability of the current feasible solution is reduced, and then save more energy to traverse other regions.

[0071] Since the current feasible solution may be at the valley position, the fitness value of the current feasible solution at the valley position cannot accurately reflect the situation of the existence of the optimal solution in this subspace. Therefore, it is necessary to estimate the information of the peak where the current feasible solution is located, so as to judge the situation of the existence of the optimal solution in this subspace according to the peak information, and then set the appropriate selection possibility. Due to the existence of deviation in the estimation, relying solely on the estimated value cannot accurately reflect the situation of the existence of the optimal solution in this subspace. Therefore, it is also necessary to combine the real value to comprehensively reflect the situation of the existence of the optimal solution in this subspace. It reflects the situation of the existence of the optimal solution in this subspace. The larger this value, the greater the possibility of the existence of the optimal solution in this subspace. Therefore, the more carefully this subspace should be searched. Therefore, it is necessary to increase the selection possibility of the current feasible solution, so as to provide a data basis for subsequent searching of this subspace.

[0072] It should be noted that in the above process, the estimated fitness value at the peak position and the current feasible solution are combined by taking the average value. This combination method does not consider the influence of the estimation accuracy, resulting in inaccurate calculation results.

[0073] Preferably, as an example, in an optimization process, calculating the selection possibility of any current feasible solution includes:

[0074]

[0075] Among them, C represents the degree of variation difference in the subspace where the current feasible solution is located, represents the predicted fitness value at the peak position of the peak where the current feasible solution is located, represents the prediction accuracy of the predicted fitness value, represents the fitness value of the feasible solution, represents the selection possibility of the current feasible solution.

[0076] It can be understood that in the process of calculating the selection possibility, according to the prediction accuracy situation, the fitness value of the current feasible solution and the predicted value of the fitness value at the peak position are combined. When the prediction accuracy is high, more reference is made to the predicted fitness value at the peak position of the peak where the feasible solution is located, so as to prevent skipping the optimal solution just because the current feasible solution takes a smaller value for a while and is not retained. At the same time, when the prediction accuracy is low, more reference is made to the fitness value of the current feasible solution, so as to prevent incorrect prediction information from interfering with the selection of the feasible solution.

[0077] The above embodiments involve subspaces, exploration degrees, variation difference degrees, predicted fitness values at peak positions, and prediction accuracies. Next, the determination methods of subspaces, exploration degrees, variation difference degrees, predicted fitness values at peak positions, and prediction accuracies need to be described.

[0078] First, the method for obtaining subspaces is introduced.

[0079] Preferably, as an example, the solution space is divided into several subspaces, including:

[0080] The solution space is evenly divided into a preset first number of subspaces. In this embodiment, the preset first number is taken as 25 for description. Other values can be taken in other embodiments, and this embodiment does not make specific limitations.

[0081] Then, the method for obtaining the exploration degree is introduced.

[0082] Optionally, as an example, obtaining the exploration degree of each subspace includes:

[0083] Obtain the positions of the feasible solutions that have been traversed in the previous iteration process and record them as the traversed positions, and use the number of traversed positions in this subspace as the exploration degree.

[0084] It should be noted that uneven search will also lead to falling into local optimality. Therefore, when judging the exploration degree, the search uniformity situation needs to be evaluated.

[0085] Preferably, as an example, obtaining the exploration degree of each subspace includes:

[0086] The positions of the feasible solutions traversed in the previous iteration are recorded as the traversed positions. Each subspace is evenly divided into a preset second number of small spaces, which are recorded as analysis spaces. The number of traversed positions included in each analysis space is used, and the reciprocal of the variance of the number of traversed positions in all analysis spaces of each subspace is used as the exploration uniformity degree of each subspace;

[0087] Multiply the number of traversed positions included in each subspace by the exploration uniformity degree to obtain the exploration degree of each subspace.

[0088] In this embodiment, the preset second number is taken as 9 for description, and other values can be taken in other embodiments, and this embodiment does not make specific restrictions.

[0089] After that, the method for obtaining the variation difference degree is introduced.

[0090] Optionally, as an example, the method for obtaining the variation difference degree in the subspace where the current feasible solution is located includes:

[0091] Use the least squares method to perform polynomial fitting on all feasible solutions and their corresponding fitness values in the solution space, and record the fitted polynomial as the overall variation relationship; divide the fitting model of the overall variation relationship at the subspace into several independent peaks, obtain the kurtosis and skewness of each independent peak, multiply the kurtosis difference between two independent peaks by the skewness difference to obtain the variation difference degree between the two independent peaks, and use the mean value of the variation law differences between all two independent peaks in the subspace as the variation difference degree of the subspace.

[0092] It should be noted that when analyzing the variation law differences in different regions of the subspace, the distribution situation of the variation law is not considered, resulting in inaccurate measurement results.

[0093] Preferably, as an example, the method for obtaining the variation difference degree in the subspace where the current feasible solution is located includes:

[0094] Use the least squares method to perform polynomial fitting on all feasible solutions and their corresponding fitness values in the solution space, and record the fitted polynomial as the overall variation relationship. Divide the fitting model of the overall variation relationship at the subspace into several independent peaks, obtain the kurtosis and skewness of each independent peak, multiply the kurtosis difference between every two independent peaks by the skewness difference to obtain the law difference degree between the two independent peaks, and use the mean value of the law difference degrees between all two independent peaks as the law difference degree of the subspace;

[0095] Use the mean value of the interval distances between each independent peak and its surrounding independent peaks as the interval distance of each independent peak, and use the variance of the interval distances of all independent peaks in the subspace as the distribution difference degree of the subspace;

[0096] Multiply the degree of difference in the law of the subspace by the degree of difference in the distribution to obtain the degree of change difference of the subspace.

[0097] It should be added that the fitting model of the overall change relationship formula at the subspace is divided into several independent peaks, including:

[0098] Obtain the closed region enclosed by the positions corresponding to the minimum value points of the overall change relationship formula as the independent region, and take the part of the relationship model corresponding to the overall change relationship formula within the independent region as the independent peak.

[0099] Subsequently, introduce the method for obtaining the estimated fitness value at the peak position.

[0100] Preferably, as an example, the method for obtaining the estimated fitness value at the peak position of the peak where the current feasible solution is located includes:

[0101] Obtain the independent peak where the current feasible solution is located, and use the overall change relationship formula to fit the fitness value at the peak position of the independent peak where the current feasible solution is located, which is denoted as the estimated fitness value at the peak position of the peak where the current feasible solution is located.

[0102] Finally, introduce the method for obtaining the prediction accuracy.

[0103] Preferably, as an example, the method for obtaining the prediction accuracy of the estimated fitness value includes:

[0104] Divide the selection possibility of each current feasible solution by the sum of the selection possibilities of all current feasible solutions to obtain the selection probability of each current feasible solution;

[0105] Use the selection probability of each current feasible solution as the extraction probability, and randomly extract a preset third number of current feasible solutions from all current feasible solutions as the retained individuals.

[0106] S21: In one optimization process, select some current feasible solutions as the retained individuals according to the selection possibility.

[0107] Preferably, as an example, in one optimization process, selecting some current feasible solutions as the retained individuals according to the selection possibility includes:

[0108] Divide the selection possibility of each current feasible solution by the sum of the selection possibilities of all current feasible solutions to obtain the selection probability of each current feasible solution;

[0109] Use the selection probability of each current feasible solution as the extraction probability, and randomly extract a preset third number of current feasible solutions from all current feasible solutions as the retained individuals.

[0110] This embodiment is described by taking the preset third quantity as 10 as an example. Other values can be taken in other embodiments, and this embodiment is not specifically limited.

[0111] So far, this embodiment is completed.

[0112] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for optimizing 3D printing process parameters of a protective face mask, characterized in that Including: Obtain the feasible value range of each process parameter for 3D printing of the protective face shield, and use the spatial range composed of the feasible value ranges of all process parameters as the solution space; Based on the solution space, use the genetic algorithm to iteratively optimize until the stopping condition is met, obtain the optimal solution, and use each process parameter in the optimal solution as the optimized process parameter; In one optimization process, use the quality index of the protective face shield corresponding to each current feasible solution as the fitness, divide the solution space into several sub-spaces, record the positions of the feasible solutions that have been traversed in the previous iteration process as the traversed positions, evenly divide each sub-space into a preset second number of small spaces as the analysis space, record the number of traversed positions included in each analysis space, and use the reciprocal of the variance of the number of traversed positions in all analysis spaces of each sub-space as the exploration uniformity degree of each sub-space; multiply the number of traversed positions included in each sub-space by the exploration uniformity degree as the exploration degree of each sub-space; use the traversed feasible solutions and the corresponding fitness values in the solution space to fit a polynomial, denoted as the overall relationship; divide the fitting model of the overall change relationship at the sub-space into several independent peaks, obtain the kurtosis and skewness of each independent peak, multiply the kurtosis difference between every two independent peaks by the skewness difference to obtain the law difference degree between the two independent peaks, and use the mean value of the law difference degrees of all two independent peaks as the law difference degree of the sub-space; use the mean value of the interval distances between each independent peak and the surrounding independent peaks as the interval distance of each independent peak, and use the variance of the interval distances of all independent peaks in the sub-space as the distribution difference degree of the sub-space; multiply the law difference degree of the sub-space by the distribution difference degree to obtain the change difference degree of the sub-space; calculate the selection possibility of the current feasible solution according to the exploration degree, the change difference degree, the fitness value of the current feasible solution, and the fitness predicted value at the peak position of the peak where the current feasible solution is located, where the selection possibility is positively correlated with the change difference degree, the fitness value of the current feasible solution, and the fitness predicted value at the peak position of the peak where the current feasible solution is located, and negatively correlated with the exploration degree; Select some current feasible solutions as the retained individuals according to the selection possibility.

2. The optimization method of 3D printing process parameters for a protective face mask according to claim 1, characterized in that, The selection possibility of the current feasible solution satisfies the relationship: ； Among them, T represents the exploration degree of the subspace where the current feasible solution is located, and C represents the variation degree in the subspace where the current feasible solution is located. It represents the predicted fitness value at the peak position of the peak where the current feasible solution is located. It represents the prediction accuracy of the predicted fitness value. It represents the fitness value of the current feasible solution. It represents the selection possibility of the current feasible solution.

3. The method for optimizing the 3D printing process parameters of a protective face mask according to claim 1, characterized in that The dividing the solution space into several sub-spaces includes: Evenly divide the solution space into a preset first number of sub-spaces.

4. The optimization method of 3D printing process parameters for a protective face mask according to claim 1, characterized in that, The dividing the fitting model of the overall change relationship at the sub-space into several independent peaks includes: Obtain the closed area surrounded by the positions corresponding to the minimum value points of the overall change relationship as the independent area, and use the part of the relationship model corresponding to the overall change relationship within the independent area as the independent peak.

5. The optimization method of 3D printing process parameters for a protective face mask according to claim 2, characterized in that, The prediction accuracy of the fitness predicted value includes: Obtain the fitness predicted values of the feasible solutions that have been traversed within the independent peak where the current feasible solution is located, use the absolute value of the difference between the fitness of each traversed feasible solution and the fitness predicted value as the fitness value prediction deviation of each traversed feasible solution, and use the reciprocal of the fitness prediction deviation of all traversed feasible solutions within the peak where the current feasible solution is located as the prediction accuracy of the fitness predicted value.

6. The method for optimizing the 3D printing process parameters of a protective face mask according to claim 1, wherein, The selecting some current feasible solutions as the retained individuals according to the selection possibility includes: Dividing the selection possibility of each current feasible solution by the sum of the selection possibilities of all current feasible solutions to obtain the selection probability of each current feasible solution; Using the selection probability of each current feasible solution as the extraction probability, randomly extracting a preset third number of current feasible solutions from all current feasible solutions as the retained individuals.

7. A method for optimizing the 3D printing process parameters of a protective face mask according to claim 1, characterized in that, Taking the number of iterations reaching the preset cut-off number as the stop condition.

8. The optimization method of 3D printing process parameters for a protective face mask according to claim 1, characterized in that, The method for obtaining the quality index of the protective mask includes: On the simulation software, using the printing parameters corresponding to each process parameter in the current feasible solution to control the printing of the protective mask and obtaining the processed protective mask; Performing various performance simulation tests on the protective mask, obtaining the performance scores of each performance test, and taking the mean of the performance scores of all performance tests as the quality index of the protective mask.

Images