A multi-objective optimization method and system for laser cladding process parameters

Abstract

The application belongs to the technical field of laser cladding process, and provides a multi-target optimization method and system for laser cladding process parameters, wherein, based on a support vector regression algorithm, laser cladding process parameters and quality characteristic parameters of a composite coating, a target LSSVM model is generated, the quality characteristic parameters of the composite coating are optimized based on the target LSSVM model and a DCNSGA-III algorithm, a target process parameter combination is determined, calculation is carried out based on the target process parameter combination, a comprehensive score corresponding to each optimization scheme in the target process parameter combination is obtained, and the optimization schemes are sorted according to a preset rule based on the comprehensive score, so that efficient multi-target optimization of process parameters is realized, and the optimization efficiency and quality of the performance of the ceramic composite coating are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of laser cladding process, and particularly relates to a laser cladding process parameter multi-objective optimization method and system. BACKGROUND

[0002] The laser cladding process technology is an efficient surface modification technology, which melts metal powder by laser and forms metallurgical bonding with the substrate, has the advantages of small heat-affected zone, low dilution rate and dense coating grain structure. In order to improve the performance of the material, ceramic particles and other two-phase materials are usually added to the iron-based alloy powder. Due to the significant differences in expansion coefficient, thermal conductivity and melting point between the metal powder and the ceramic particles in the composite powder, the laser cladding process parameters and the composite coating quality characteristics present a relatively complex nonlinear coupling relationship. At the same time, when the laser cladding process parameters are unreasonable, the composite coating will appear excessive dilution, cracks, pores and other defects, thereby shortening the service life of mechanical parts. Therefore, it is necessary to optimize the laser cladding process parameters to prepare high-quality composite coatings.

[0003] The optimization of laser cladding process parameters is a complex multivariate and multi-objective problem. When dealing with high-dimensional and nonlinear data, it is often difficult to capture the complex interaction effects between parameters by using traditional mathematical statistical methods (such as linear regression, grey relation analysis and empirical formula, etc.), thereby resulting in insufficient modeling accuracy. And using traditional optimization methods such as Taguchi method and response surface method will be limited by experimental design and data size, so that there is a difference between the optimized process parameters and the optimal process parameters.

[0004] Therefore, how to solve the optimization of laser cladding process parameters to prepare high-quality composite coatings is a problem that those skilled in the art need to solve. SUMMARY

[0005] Therefore, the embodiments of the present application provide a laser cladding process parameter multi-objective optimization method and system to solve the problem of how to optimize the laser cladding process parameters to prepare high-quality composite coatings in the prior art; that is, the embodiments of the present application can improve the quality characteristics of the composite coating.

[0006] According to an aspect of the present application, a laser cladding process parameter multi-objective optimization method is provided, which comprises: generating a target LSSVM model based on a support vector regression algorithm, laser cladding process parameters and quality characteristic parameters of a composite coating; optimizing the quality characteristic parameters of the composite coating based on the target LSSVM model and a DCNSGA-III algorithm, and determining a target process parameter combination; calculating a comprehensive score corresponding to each optimization scheme in the target process parameter combination, and sorting the optimization schemes based on the comprehensive score according to a preset rule.

[0007] In one embodiment, the generating of the target LSSVM model based on the support vector regression algorithm, the laser cladding process parameters and the quality characteristic parameters of the composite coating comprises: generating a sample data set based on the laser cladding process parameters and the quality characteristic parameters of the composite coating; and generating the target LSSVM model based on the support vector regression algorithm and the sample data set.

[0008] In one embodiment, the optimizing of the quality characteristic parameters of the composite coating based on the target LSSVM model and the DCNSGA-III algorithm to determine the target process parameter combination comprises: obtaining parameters of the DCNSGA-III algorithm; initializing dynamic constraint boundaries of the laser cladding process parameters and the quality characteristic parameters of the composite coating to obtain constraint conditions of the laser cladding process parameters and the quality characteristic parameters of the composite coating, respectively; generating a parent population based on the quality characteristic parameters of the composite coating, the target LSSVM model and the constraint conditions; processing the parent population to obtain a child population; processing the parent population and the child population to generate a next-generation parent population; iterating the above process until a preset iteration number is reached, and outputting the target process parameter combination.

[0009] In one of the embodiments, the calculating a comprehensive score corresponding to each optimization scheme in the target process parameter combination and ranking the optimization schemes based on the comprehensive score according to a preset rule comprises: constructing an initial evaluation matrix based on optimization targets in the target process parameter combination; preprocessing the optimization targets in the initial evaluation matrix based on a preset manner to obtain preprocessed optimization targets; calculating the preprocessed optimization targets based on an entropy weight method to obtain first weights; calculating the preprocessed optimization targets based on a coefficient of variation method to obtain second weights; constructing a weight matrix based on the first weights and the second weights; constructing a correlation matrix based on the first weights and the second weights; calculating a combination weight based on the correlation matrix and the weight matrix; and performing weighted summation on the preprocessed optimization targets based on the combination weight to obtain a comprehensive score corresponding to each optimization scheme, and ranking the optimization schemes based on the comprehensive score according to a preset rule.

[0010] In one of the embodiments, the calculating the preprocessed optimization targets based on the entropy weight method to obtain the first weights comprises: calculating an index value ratio based on the preprocessed optimization targets; calculating an information entropy based on the index value ratio; calculating the first weights based on the information entropy, and calculating a first score based on the first weights and the preprocessed optimization targets.

[0011] In one of the embodiments, the calculating the preprocessed optimization targets based on the coefficient of variation method to obtain the second weights comprises: calculating a mean value based on the preprocessed optimization targets, and calculating a standard deviation based on the mean value and the preprocessed optimization targets; calculating a coefficient of variation based on the mean value and the standard deviation; calculating the second weights based on the coefficient of variation, and calculating a second score based on the second weights and the preprocessed optimization targets.

[0012] In one of the embodiments, the calculating the combination weight based on the correlation matrix and the weight matrix comprises: calculating a target correlation coefficient based on the correlation matrix, and comparing the target correlation coefficient with a preset threshold to obtain a comparison result; and calculating the combination weight based on the weight matrix based on the comparison result.

[0013] In one of the embodiments, the calculating the combination weight based on the comparison result and the weight matrix comprises: when the target correlation coefficient is less than the preset threshold, obtaining the combination weight through weighted average calculation, and the combination weight is:

[0014]

[0015] wherein ω ij is the weight of the pre-processed optimization target calculated according to different weight calculation methods, l is the weight number of columns in the weight matrix, and n is the number of optimization targets in each optimization scheme in the target process parameter combination.

[0016] In one of the embodiments, the calculating the combined weight based on the weight matrix according to the comparison result further includes: when the target correlation coefficient is greater than the preset threshold, calculating a standard deviation of the weight, a contradictoriness measure of elements in the correlation matrix, and an information carrying capacity, wherein the standard deviation, the contradictoriness measure, and the information carrying capacity are:

[0017]

[0018] wherein λ i is the standard deviation of the weight, f i is the contradictoriness measure of the elements in the correlation matrix, c i is the information carrying capacity, ω ij is the weight of the pre-processed optimization target calculated according to different weight calculation methods, is the mean value of the jth column weight in the weight matrix, r ij is the Kendall rank correlation coefficient between the ith pre-processed optimization target in the first weight set and the jth pre-processed optimization target in the second weight set; the information carrying capacity is normalized to generate a weight coefficient, and the weight is weighted and averaged based on the weight coefficient to obtain a combined weight, wherein the combined weight is:

[0019]

[0020] wherein c' i is the weight coefficient.

[0021] According to another aspect of the present application, a laser cladding process parameter multi-objective optimization system is provided, which comprises a target model generation module, a target parameter combination determination module, and an optimization scheme selection module. The target model generation module is configured to generate a target LSSVM model based on a support vector regression algorithm, laser cladding process parameters, and quality characteristic parameters of a composite coating. The target parameter combination determination module is configured to optimize the quality characteristic parameters of the composite coating based on the target LSSVM model and a DCNSGA-III algorithm, and determine a target process parameter combination. The optimization scheme selection module is configured to calculate a comprehensive score corresponding to each optimization scheme in the target process parameter combination, and sort the optimization schemes according to a preset rule based on the comprehensive score.

[0022] To sum up, in the embodiment of the application, by generating a target LSSVM model based on a support vector regression algorithm, laser cladding process parameters and quality characteristic parameters of a composite coating, optimizing the quality characteristic parameters of the composite coating based on the target LSSVM model and a DCNSGA-III algorithm, determining a target process parameter combination, calculating a comprehensive score corresponding to each optimization scheme in the target process parameter combination, and sorting the optimization schemes based on the comprehensive score according to a preset rule, efficient multi-objective optimization of process parameters is realized, and the optimization efficiency and quality of the performance of the ceramic composite coating are improved. BRIEF DESCRIPTION OF DRAWINGS

[0023] In the following description of exemplary embodiments in conjunction with the accompanying drawings, more details, features and advantages of the application are disclosed, in which:

[0024] Figure 1 A flowchart of a laser cladding process parameter multi-objective optimization method disclosed in an embodiment of the application is shown;

[0025] Figure 2 A flowchart of step S110 is shown; Figure 1 A flowchart of step S120 is shown;

[0026] Figure 3 A flowchart of step S130 is shown; Figure 1 A flowchart of step S130 is shown;

[0027] Figure 4 A flowchart of step S133 is shown; Figure 1 A flowchart of step S133 is shown;

[0028] Figure 5 A flowchart of step S134 is shown; Figure 4 A flowchart of step S134 is shown;

[0029] Figure 6 A flowchart of step S137 is shown; Figure 4 A flowchart of step S137 is shown;

[0030] Figure 7 A flowchart of step S137 is shown; Figure 4 A flowchart of step S137 is shown;

[0031] Figure 8 A flowchart of step S137 is shown;

[0032] Figure 9 A flowchart of step S137 is shown;

[0033] Figure 10 A flowchart of step S137 is shown;

[0034] Figure 11 Fig. 3 shows a schematic diagram of a multi-objective optimization process of the DCNSGA-III;

[0035] Figure 12 Fig. 4 shows a schematic diagram of a multi-objective optimization system for laser cladding process parameters according to an embodiment of the present application. DETAILED DESCRIPTION

[0036] Embodiments of the present application will be described in more detail with reference to the drawings. While several embodiments of the application are shown in the drawings, it is understood, however, that the application can be practiced by using various forms other than those illustrated in the drawings. There is no intent that the application should necessarily be limited to the embodiments illustrated in the drawings. For instance, it is generally contemplated that the application can be implemented with various types of electronic circuits and components, and therefore the drawings are not intended to limit the scope of the application. It is also contemplated that one or more features of one embodiment can be used with other embodiments of the application. It should be understood that the drawings and detailed description thereto are not intended to limit the scope of the present application.

[0037] It should be understood that the steps of the method embodiments of the present application can be performed in different orders and / or in parallel. In addition, the method embodiments can include additional steps and / or omit performing the steps shown. The scope of the present application is not limited in this respect.

[0038] The term "comprising" and variations thereof as used herein are used inclusively, i.e., "comprising but not limited to." The term "based on" is "based at least in part on." The term "one embodiment" means "at least one embodiment." The term "another embodiment" means "at least one additional embodiment." The term "some embodiments" means "at least some embodiments." Related terms are defined in the following description. It should be noted that reference to a "first," "second," etc. concept does not imply that the concepts so designated must be different or that the concepts so designated are necessarily related or dependent upon one another.

[0039] It should be noted that the terms "a" or "an" as used herein mean "one or more" unless otherwise explicitly provided. It should be noted that the terms "first," "second," etc. are used herein to describe various elements, and are not used to designate a particular order or a particular order of performing the steps of the method.

[0040] The names of the messages or information exchanged between the various apparatuses in the embodiments of the present application are used for illustrative purposes only, and are not intended to limit the scope of the messages or information.

[0041] It should be noted that the execution subject of the laser cladding process parameter multi-objective optimization method provided in the embodiments of the present application can be one or more electronic devices, and the present application does not make any limitation in this regard; wherein the electronic device can be a terminal (i.e. client) or a server, and when the execution subject includes multiple electronic devices and at least one terminal and at least one server are included in the multiple electronic devices, the laser cladding process parameter multi-objective optimization method provided in the embodiments of the present application can be executed by the terminal and the server together. Correspondingly, the terminal mentioned here can include but is not limited to: a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart watch, a smart voice interaction device, a smart home appliance, a vehicle-mounted terminal, an aircraft, etc. The server mentioned here can be a stand-alone physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and basic cloud computing services such as big data and artificial intelligence platforms, etc.

[0042] Based on the above description, the embodiments of the present application propose a laser cladding process parameter multi-objective optimization method, which can be executed by the electronic device (terminal or server) mentioned above; or the laser cladding process parameter multi-objective optimization method can be executed by the terminal and the server together. For the convenience of description, the laser cladding process parameter multi-objective optimization method executed by the electronic device is taken as an example for description hereinafter.

[0043] Please refer to Figure 1 , which is a flowchart of a laser cladding process parameter multi-objective optimization method disclosed in the embodiments of the present application. The laser cladding process parameter multi-objective optimization method solves the problem of how to optimize the laser cladding process parameters to prepare a high-quality composite coating, thereby improving the quality characteristics of the composite coating. It should be noted that the laser cladding process parameter multi-objective optimization method of the present application is not limited to the steps and order shown in the flowchart. Figure 1 According to different needs, the steps in the flowchart shown can be added, removed, or the order changed. In the embodiments of the present application, as shown in Figure 1 , the flow of a laser cladding process parameter multi-objective optimization method includes at least the following steps.

[0044] S110, generating a target LSSVM model based on a support vector regression algorithm, laser cladding process parameters and quality characteristic parameters of the composite coating.

[0045] As Figure 2As shown in the embodiments of the present application, Figure 2 The step S110 at least includes the following steps:

[0046] S111, generating a sample data set based on the laser cladding process parameters and the quality characteristic parameters of the composite coating.

[0047] In the embodiments of the present application, the laser cladding process parameters x k At least can include: laser power P, scanning speed V, powder feeding rate F and overlap rate The quality characteristics y of the composite coating k At least can include: dilution rate η, surface flatness θ and microhardness HV, wherein x k =(x k1 ,x k2 ,…,x kd )∈R n , d is the dimension of the input vector. Based on the laser cladding process parameters x k and the quality characteristic parameters y k of the composite coating, a sample data set {x k ,y k} is generated, wherein k = 1, 2, …, l, and l is the sample number.

[0048] S112, generating a target LSSVM model based on the support vector regression algorithm and the sample data set.

[0049] In the embodiments of the present application, the model is constructed by support vector regression algorithm (Support Vector Regression, SVR). Based on the laser cladding process parameters x k Nonlinear mapping The laser cladding process parameters x k are mapped to a high-dimensional feature space, so that the data which is nonlinearly separable or difficult to fit becomes more manageable. The relationship between the laser cladding process parameters and the quality characteristic parameters of the composite coating is as follows:

[0050]

[0051] Wherein w is the weight vector, and b is the bias.

[0052] In order to construct the target LSSVM model, the optimal values of w and b are needed, so the penalty factor γ and the relaxation variable ξ are introduced, and based on the regularization theory and the least square function, the optimization problem and the equal constraint are as follows:

[0053]

[0054] Wherein w Tw is used to control the complexity of the decision function, γ is used to determine the proportion between model complexity and accuracy, ξ k is the regression error of the kth sample (i.e., the difference between the actual value and the model predicted value).

[0055] By constructing a Lagrange function and combining the Karush-Kuhn-Tucker (KKT) condition, the optimal values of w and b are derived, and the dual form of the Lagrange multiplier is as follows:

[0056]

[0057] where a k and a j are Lagrange multipliers, K(x k ,x j ) is a kernel function of Mercer's theorem, K(x k ,x j ) is used to represent the inner product of sample x k and sample x j in a high-dimensional feature space, and K(x k ,x j ) is used to balance the similarity between samples.

[0058] Further, the target LSSVM model can be obtained, and the target LSSVM model is as follows:

[0059]

[0060] where a k is a Lagrange multiplier, K(x k ,x j ) is a kernel function of Mercer's theorem, and b is a bias.

[0061] S120, based on the target LSSVM model and the DCNSGA-III algorithm, the quality characteristic parameters of the composite coating are optimized, and a target process parameter combination is determined.

[0062] As shown in FIG. Figure 3 in the embodiment of the present application, Figure 3 the step S120 at least includes the following steps:

[0063] S121, obtaining the parameters of the DCNSGA-III algorithm.

[0064] In the embodiment of the present application, the constraint range of the laser cladding process parameters is initialized, and the third generation dynamic constraint non-dominated sorting genetic algorithm (DCNSGA-III) is parameterized. The parameters of the DCNSGA-III algorithm include that the maximum iteration number is 100, the population size is 100, the dynamic constraint boundary reduction trend is 5, the crossover probability is 0.9, and the mutation probability is 0.01.

[0065] In S122, the dynamic constraint boundary of the laser cladding process parameters and the quality characteristic parameters of the composite coating is initialized, and the constraint conditions of the laser cladding process parameters and the quality characteristic parameters of the composite coating are obtained respectively.

[0066] In the embodiment of the present application, the initial population P0 is generated by randomly selecting from the quality characteristic parameters of the composite coating, the degree of constraint violation of individuals in the initial population P0 is calculated, and the solution with the most constraint violations is selected as the original ε-constraint boundary. The contraction of the ε-dynamic constraint boundary adopts an exponential function of the simulated annealing algorithm, and the exponential function of the simulated annealing algorithm is as follows:

[0067]

[0068] Wherein, A i and B i are constants, A i represents the initial state of the ε-constraint boundary, B i represents the final state of the ε-constraint boundary, t is the iteration number, ct is a constant for controlling the temperature decay trend, and δ is a value close to 0, and δ is used to ensure that all individuals are ε-feasible in the initial state.

[0069] In S123, the parent population is generated based on the quality characteristic parameters of the composite coating, the target LSSVM model and the constraint conditions.

[0070] In the embodiment of the present application, the parent population P t is generated based on the quality characteristic parameters of the composite coating, the target LSSVM model and the constraint conditions.

[0071] In S124, the child population is obtained by processing the parent population.

[0072] In the embodiment of the present application, two individuals x1 and x2 are randomly selected from the parent population P t for comparison, when x1 is an ε-feasible solution and x2 is an ε-infeasible solution, x1 is retained, and when x1 and x2 are both ε-infeasible solutions, the ε-infeasible solution with fewer constraint violations is retained. The parent population P t is processed through operations such as crossover and mutation to generate the child population Q.t .

[0073] S125, performing processing based on the parent population and the offspring population to generate a next generation parent population.

[0074] In the embodiment of the present application, the parent population P t and the offspring population Q t are merged to generate a new population U t , and solutions in the new population U t are divided into an ε-feasible solution set S1 and an ε-infeasible solution set S2. When the number of solutions in S1 is less than a preset population size N, solutions with the least constraint violation are selected from S2 to supplement S1; when the number of solutions in S1 is equal to the preset population size N, N solutions are selected from S1 based on an elite selection strategy of a reference point to generate a next generation parent population P t+1 .

[0075] S126, iterating the above process until a preset number of iterations is reached, and outputting the target process parameter combination.

[0076] S130, calculating a comprehensive score corresponding to each optimization scheme in the target process parameter combination, and sorting the optimization schemes based on the comprehensive score according to a preset rule.

[0077] As shown in FIG. Figure 4 , in the embodiment of the present application, Figure 4 the step S130 at least includes the following steps:

[0078] S131, constructing an initial evaluation matrix based on optimization objectives in the target process parameter combination.

[0079] In the embodiment of the present application, the target process parameter combination contains m optimization schemes for laser cladding process parameters, and each optimization scheme is composed of n optimization objectives (i.e., quality characteristic parameters of the target composite coating). An initial evaluation matrix X is constructed based on the optimization objectives. The predicted value of the jth optimization objective in the ith process parameter optimization scheme is denoted as x ij by the target LSSVM model. The initial evaluation matrix X is as follows:

[0080]

[0081] S132, preprocessing the optimization objectives in the initial evaluation matrix based on a preset manner to obtain preprocessed optimization objectives.

[0082] In the embodiment of the present application, since different types of optimization objectives have different evaluation criteria, it is necessary to determine the index type to divide the types of optimization objectives, and then normalize or moderate according to the index type, and then standardize, so as to ensure that the optimization objectives are under the same evaluation criteria.

[0083] S133, calculating the pre-processed optimization objectives based on the entropy weight method to obtain a first weight.

[0084] As shown in Figure 5 , in the embodiment of the present application, Figure 5 the step S133 at least includes the following steps:

[0085] S1331, calculating the pre-processed optimization objectives to obtain an index value ratio.

[0086] In the embodiment of the present application, the index value ratio P ij of the jth optimization objective in the ith process parameter optimization scheme is taken as an example for description, the index value ratio P ij is as follows:

[0087]

[0088] Wherein, s ij is the jth optimization objective in the pre-processed ith process parameter optimization scheme, and m is the number of optimization schemes for the laser cladding process parameters.

[0089] S1332, calculating the index value ratio to obtain an information entropy.

[0090] In the embodiment of the present application, the information entropy E j is as follows:

[0091]

[0092] Wherein, E j ≥ 0, when P ij = 0, then P ij lnP ij = 0, that is, E j = 0.

[0093] S1333, calculating the first weight based on the information entropy, and calculating a first score based on the first weight and the pre-processed optimization objectives.

[0094] In the embodiment of the present application, the first weight ω 1j and the first score δ 1j are as follows:

[0095]

[0096] S134, calculating the pre-processed optimization target based on a coefficient of variation method to obtain a second weight.

[0097] As shown in the embodiments of the present application, Figure 6 the step S134 at least includes the following steps: Figure 6

[0098] S1341, calculating a mean value based on the pre-processed optimization target, and calculating a standard deviation based on the mean value and the pre-processed optimization target.

[0099] In the embodiments of the present application, the mean value A j and the standard deviation B j are calculated according to the following formula:

[0100]

[0101] S1342, calculating a coefficient of variation based on the mean value and the standard deviation.

[0102] In the embodiments of the present application, the coefficient of variation V j is calculated according to the following formula:

[0103]

[0104] S1343, calculating the second weight based on the coefficient of variation, and calculating a second score based on the second weight and the pre-processed optimization target.

[0105] In the embodiments of the present application, the second weight ω 2j and the second score δ 2j are calculated according to the following formula:

[0106]

[0107] S135, constructing a weight matrix based on the first weight and the second weight.

[0108] In the embodiments of the present application, the first weight and the second weight of each pre-processed optimization target are calculated, and a first weight set W1=[ω 11 ,ω 12 ,…,ω 1j ] and a second weight set W2=[ω 21 ,ω 22 ,…,ω 2j ] are obtained, and a weight matrix W is constructed based on the first weight set and the second weight set.

[0109] ​

[0110] wherein j is the number of pre-processed optimization targets.

[0111] S136, constructing a correlation matrix based on the first weight and the second weight.

[0112] In the embodiment of the present application, the first weight and the second weight of each pre-processed optimization target are calculated, and the first weight set W1=[ω 11 ,ω 12 ,…,ω 1j ] and the second weight set W2=[ω 21 ,ω 22 ,…,ω 2j ] are obtained respectively. In order to evaluate the correlation between W1 and W2, a correlation matrix R is established, and the correlation matrix R is as follows:

[0113]

[0114] wherein R is a j×j symmetric matrix, and r ij in the matrix is the Kendall rank correlation coefficient of the i-th pre-processed optimization target in the first weight set W1 and the j-th pre-processed optimization target in the second weight set W2. When i=j, the correlation coefficient is 1, indicating the complete consistency of the same optimization target in the two weight calculation methods.

[0115] S137, calculating a combined weight based on the correlation matrix and the weight matrix.

[0116] As shown in FIG. 13, in the embodiment of the present application, Figure 7 the step S137 at least includes the following steps: Figure 7

[0117] S1371, calculating a target correlation coefficient based on the correlation matrix, and comparing the target correlation coefficient with a preset threshold to obtain a comparison result.

[0118] In the embodiment of the present application, the sum of the correlation coefficients of the non-diagonal lines in the correlation matrix R is calculated to obtain a target correlation coefficient, and the target correlation coefficient is compared with a preset threshold to obtain a comparison result. The preset threshold can be 0.05*n, and n is the number of quality characteristic parameters of the target composite coating.

[0119] S1372, calculating the combined weight based on the weight matrix according to the comparison result.

[0120] ​In this embodiment of the invention, when the target correlation coefficient is less than a preset threshold, it indicates that the entropy weight method and the coefficient of variation method lack consistency in weight ranking. A preliminary simple weighted average calculation is then performed to obtain the combined weight WC. j Combined weights WC j As shown in the following formula:

[0121]

[0122] Where, ω ij The weight values ​​are calculated for the preprocessed optimization objectives using different weight calculation methods. l is the number of weight values ​​in the column of the weight matrix W, and n is the number of optimization objectives for each scheme.

[0123] When the target correlation coefficient is greater than the preset threshold, it indicates that the entropy weight method and the coefficient of variation method have high consistency in weight ranking. The standard deviation λ of the weight vector is then calculated. i The measure of contradiction f of elements in the correlation matrix R i and information carrying capacity c i Standard deviation λ i Contradiction measurement f i and information carrying capacity c i As shown in the following formula:

[0124]

[0125] in, Let r be the mean of the weights in the j-th column of the weight matrix W. ij Let be the Kendall rank correlation coefficient between the i-th preprocessed optimization objective in the first weight set W1 and the j-th preprocessed optimization objective in the second weight set W2.

[0126] Information carrying capacity c i Standardize the data to generate weight coefficients c' i And based on the weighting coefficient c' i The combined weight WC is obtained by performing a weighted average calculation on the weight vector. j Combined weights WC j As shown in the following formula:

[0127]

[0128] S138. Based on the combined weights, the preprocessed optimization objectives are weighted and summed to obtain a comprehensive score for each optimization scheme, and the optimization schemes are sorted according to preset rules based on the comprehensive scores.

[0129] In the embodiment of the present application, the optimization target and the combined weight in the optimization scheme of the i laser cladding process parameters are weighted and summed to obtain the comprehensive score Score corresponding to each optimization scheme i The optimization schemes are sorted according to the comprehensive score Score corresponding to each scheme according to a preset rule, so as to determine the order of the optimization schemes. The comprehensive score Score corresponding to each optimization scheme is as follows: i

[0130]

[0131] Wherein, s ij is the jth optimization target in the ith process parameter optimization scheme after processing, and n is the number of quality characteristic parameters of the target composite coating.

[0132] As can be seen from the above, in the laser cladding process parameter multi-objective optimization method of the present application, the target LSSVM model is generated based on the support vector regression algorithm, the laser cladding process parameters and the quality characteristic parameters of the composite coating, the quality characteristic parameters of the composite coating are optimized based on the target LSSVM model and the DCNSGA-III algorithm, the target process parameter combination is determined, the comprehensive score corresponding to each optimization scheme in the target process parameter combination is calculated, and the optimization schemes are sorted according to the comprehensive score according to a preset rule, so as to realize efficient multi-objective optimization of the process parameters and improve the optimization efficiency and quality of the ceramic composite coating performance.

[0133] In a specific embodiment, referring to Figure 8 S10, determine the ceramic composite powder and the substrate, perform laser cladding experiments by trial and error method, ensure that the surface of the composite coating has no obvious defects, and determine the laser cladding process parameter range; S20, based on the laser cladding process parameter range, perform central composite experimental design and laser cladding experiments by response surface method, determine the quality characteristic parameters of the composite coating, collect the laser cladding process parameters and the quality characteristic parameters of the composite coating corresponding to the laser cladding process parameters; S30, use the laser cladding process parameters (i.e. laser power P, scanning speed V, powder feeding rate F and overlap rate ​) is input, and the quality characteristic parameters (i.e., dilution rate η, surface flatness θ and microhardness HV) of the composite coating are output, a least squares support vector machine (LS-SVM) prediction model is constructed; S40, the least squares support vector machine prediction model is solved by using the third generation dynamic constraint non-dominated sorting genetic algorithm, the quality characteristic parameters of the composite coating are balanced, and an optimal Pareto front optimal solution set (including four laser cladding process parameters and three corresponding predicted quality characteristic parameters of the composite coating) is obtained; S50, the optimal Pareto front optimal solution set is calculated, scored and sorted by using the "entropy weight method-variation coefficient method" comprehensive evaluation decision method, and an optimal set of laser cladding process parameters and corresponding predicted quality characteristic parameters of the composite coating are obtained.

[0134] Embodiment one

[0135] Exemplarily, 15% Cr3C2 / 15-5PH composite powder is used, wherein the preparation process of the 15% Cr3C2 / 15-5PH composite powder is as follows: first, Cr3C2 ceramic particles are put into a planetary ball mill, the planetary ball mill is used to grind the Cr3C2 ceramic particles at a speed of 600 r / min for 1 hour, then the ground Cr3C2 ceramic particles are added to 15-5PH powder to obtain a mixed powder, and the mixed powder is put into the planetary ball mill to mix the mixed powder at a speed of 360 r / min for 1 hour to obtain the Cr3C2 / 15-5PH composite powder. A central composite design experiment with four factors and five levels is adopted, and the laser cladding process parameters x k include laser power P, scanning speed V, powder feeding rate F and overlap rate The laser cladding process parameters and levels are shown in Table 1, and the central composite design experiment scheme is shown in Table 2.

[0136] Table 1: Laser cladding process parameters and levels

[0137]

[0138] Table 2: Central composite design experiment scheme

[0139]

[0140] As Figure 9 shown, the quality characteristics of the composite coating after laser cladding of the 15% Cr3C2 / 15-5PH composite powder mainly include dilution rate (η), surface flatness (θ) and microhardness (HV), wherein the calculation formulas of the dilution rate (η) and the surface flatness (θ) are as follows:

[0141]

[0142] Where η is the dilution rate, A c The area of ​​the composite coating is in mm. 2 A m The area of ​​the molten pool is in mm. 2 .

[0143]

[0144] Where θ is the surface flatness, and A c The area of ​​the composite coating is in mm. 2 W represents the width of the composite coating in mm, and H represents the height of the composite coating in mm.

[0145] The measurement results of the central composite design experiment are shown in Table 3.

[0146] Table 3: Measurement Results of the Central Composite Design Experiment

[0147]

[0148] like Figure 10 As shown, the laser cladding process parameters x k (That is, laser power P, scanning speed V, powder feeding rate F, and overlap rate) ) is the input, representing the quality characteristic parameter y of the composite coating. k (That is, the dilution rate η, surface smoothness θ, and microhardness HV) are used as outputs to construct the target LSSVM model.

[0149] The dilution rate η, surface smoothness θ, and microhardness HV of the composite coating were selected as multi-objective optimization indicators. Based on response surface methodology, the optimization objective for the dilution rate was set at 30% to ensure a good metallurgical bond between the composite coating and the substrate. Improved surface smoothness facilitates subsequent processing operations, increases processing efficiency, and reduces related costs. Increasing microhardness ensures the wear resistance of the coating, meeting the practical application requirements of 20CrMnTi parts. The optimization objective function is shown in the following formula:

[0150]

[0151] Based on preliminary experimental data, engineering experience, and the upper and lower limits of the safety and performance of related equipment, the laser cladding process parameters (P, V, F, ...) were determined. The constraints are as follows:

[0152]

[0153] like Figure 11As shown, based on the target LSSVM model, the optimization objective function and the constraint conditions, the DCNSGA-III algorithm is used for multi-objective optimization of laser cladding process parameters. Through the strategies of "dynamic constraint boundary" and "binary competition selection program", efficient multi-objective optimization under complex constraint conditions is realized.

[0154] Based on the different characteristics and requirements of the optimization objectives (i.e. dilution rate η, surface flatness θ and microhardness HV) in the optimal Pareto front optimal solution set obtained by the DCNSGA-III algorithm, the optimization objectives are adapted and normalized to ensure that they can be effectively compared in the same evaluation system. The adaptation and normalization are as follows:

[0155]

[0156] Wherein, x ij is the predicted value of the jth optimization objective in the ith process parameter optimization scheme through the target LSSVM model.

[0157] After the adaptation and normalization, the matrix X = [x ij ′] mn is obtained. Since there are dimensional differences between different optimization objectives, in order to ensure the fairness of the evaluation results, the optimization objectives are standardized, and the standardization is as follows:

[0158]

[0159] Wherein, Mean(x j ) is the average value of the jth column optimization objective, and Std(x j ) is the standard deviation of the jth column optimization objective.

[0160] Based on the EMW and CV objective weighting methods, the combined weight is calculated, and the combined weight is shown in Table 4.

[0161] Table 4: EWM-CV comprehensive weight calculation value

[0162]

[0163] Through the combined weight and the optimization objective predicted value, the comprehensive score of each process parameter optimization scheme is calculated, and then the scores are sorted according to the size. The top five process parameter optimization schemes are shown in Table 5.

[0164] Table 5: EWM-CV ranking results

[0165]

[0166] The first ranked process parameter optimization scheme in Table 5 is selected for verification experiment, and the experimental results are shown in Table 6. As shown in Table 6, the relative errors of the predicted values of dilution rate η, surface flatness θ and microhardness HV and the actual values are 3.38%, 1.52% and 2.08% respectively. Compared with the actual values, the relative errors of the predicted values and the engineering experience values are obviously increased, in which the relative errors of dilution rate η, surface flatness θ and microhardness HV are 4.90%, 12.39% and 3.32% respectively. This shows that the laser cladding process parameter optimization by using DCNSGA-III and EWM-CV can effectively obtain the optimal laser cladding process parameters.

[0167] Table 6: Comparison of results before and after optimization

[0168]

[0169] Referring to Figure 12 , which is a structural schematic diagram of a laser cladding process parameter multi-objective optimization system disclosed in an embodiment of the present application. In an embodiment, as shown in Figure 12 , the present application provides a laser cladding process parameter multi-objective optimization system 100, which can at least include: a target model generation module 110, a target parameter combination determination module 130 and an optimization scheme selection module 150. Among them, there is information interaction between the target model generation module 110 and the target parameter combination determination module 130, and there is information interaction between the parameter combination determination module 130 and the optimization scheme selection module 150.

[0170] The target model generation module 110 is used to generate a target LSSVM model based on a support vector regression algorithm, laser cladding process parameters and quality characteristic parameters of a composite coating. The target model generation module 110 can at least include a first generation module 111 and a second generation module 113.

[0171] The first generation module 111 is used to generate a sample data set based on the laser cladding process parameters and the quality characteristic parameters of the composite coating. Specifically, the laser cladding process parameters x k can at least include laser power P, scanning speed V, powder feeding rate F and overlap rate The quality characteristic parameters y k of the composite coating can at least include dilution rate η, surface flatness θ and microhardness HV, wherein x k = (x k1 , x k2 ,…, x kd ) ∈ R n , d is the dimension of the input vector. Based on the laser cladding process parameters x kand the quality characteristic parameter y of the composite coating k Generate sample dataset {x k ,y k}, where k = 1, 2, ..., l, and l is the sample size.

[0172] The second generation module 113 is used to generate a target LSSVM model based on the support vector regression algorithm and the sample dataset. Specifically, the model is constructed using the support vector regression (SVR) algorithm. This is based on the laser cladding process parameters x. k Perform nonlinear mapping laser cladding process parameters x k Mapping to a high-dimensional feature space makes nonlinearly separable or difficult-to-fit data easier to process. The relationship between laser cladding process parameters and the quality characteristic parameters of the composite coating is shown in the following formula:

[0173]

[0174] Where w is the weight vector and b is the bias.

[0175] To construct the target LSSVM model, optimal values ​​for w and b are required. Therefore, a penalty factor γ and a slack variable ξ are introduced. Based on regularization theory and the least squares function, the optimization problem and equality constraints are as follows:

[0176]

[0177] Among them, w T w is used to control the complexity of the decision function, γ is used to determine the weighting between model complexity and accuracy, and ξ is used to control the complexity of the decision function. k This represents the regression error for the k-th sample (i.e., the difference between the actual value and the model's predicted value).

[0178] By constructing the Lagrange function and combining it with the Karush-Kuhn-Tucker (KKT) conditions, the optimal values ​​of w and b are derived, resulting in the dual form of the Lagrange multiplier as shown in the following formula:

[0179]

[0180] Among them, a k and a j All are Lagrange multipliers, K(x) k ,x j K(x) is the kernel function of Mercer's theorem. k ,x j ) is used to represent sample x k and sample xj inner product in high dimensional feature space, K(x k , x j ) is used to balance the similarity between samples.

[0181] Further, a target LSSVM model can be obtained, and the target LSSVM model is as follows:

[0182]

[0183] where a k is a Lagrange multiplier, K(x k , x j ) is a kernel function of Mercer's theorem, and b is a bias.

[0184] The target parameter combination determination module 130 is configured to optimize the quality characteristic parameters of the composite coating based on the target LSSVM model and the DCNSGA-III algorithm, and determine a target process parameter combination. The target parameter combination determination module 130 can at least include a first acquisition module 131, a second acquisition module 133, a first population generation module 135, a second population generation module 136, a third population generation module 137, and an output module 139.

[0185] The first acquisition module 131 is configured to acquire parameters of the DCNSGA-III algorithm. Specifically, the constraint range of the laser cladding process parameters is initialized, and the third generation dynamic constraint non-dominated sorting genetic algorithm (DCNSGA-III) is parameterized. The parameters of the DCNSGA-III algorithm include a maximum iteration number of 100, a population size of 100, a dynamic constraint boundary reduction trend of 5, a crossover probability of 0.9, and a mutation probability of 0.01.

[0186] The second acquisition module 133 is configured to initialize the dynamic constraint boundary of the laser cladding process parameters and the quality characteristic parameters of the composite coating, and obtain the constraint conditions of the laser cladding process parameters and the quality characteristic parameters of the composite coating, respectively. Specifically, an initial population P0 is generated by randomly selecting from the quality characteristic parameters of the composite coating, the degree of constraint violation of individuals in the initial population P0 is calculated, and the solution with the most constraint violations is selected as the original ε-constraint boundary. The ε-dynamic constraint boundary is contracted by using an exponential function of the simulated annealing algorithm, and the exponential function of the simulated annealing algorithm is as follows:

[0187]

[0188] where A i and B i are constants, A i represents the initial state of the ε-constraint boundary, and B iThe final state of the ε-constrained boundary is represented by t, the number of iterations is ct, the constant controlling the temperature decay trend is ct, and δ is a value close to 0. δ is used to ensure that all individuals are ε-feasible in the initial state.

[0189] The first population generation module 135 is used to generate a parent population based on the quality characteristic parameters of the composite coating, the target LSSVM model, and the constraints. Specifically, it generates a parent population P based on the quality characteristic parameters of the composite coating, the target LSSVM model, and the constraints. t .

[0190] The second population generation module 136 is used to process the parent population to obtain the offspring population. Specifically, it generates the offspring population from the parent population P. t Two individuals, x1 and x2, are randomly selected for comparison. If x1 is an ε-feasible solution and x2 is an ε-infeasible solution, x1 is retained. If both x1 and x2 are ε-infeasible solutions, the ε-infeasible solution that violates fewer constraints is retained. Crossover and mutation operations are used to adjust the parent population P. t Process the data to generate a progeny population Q. t .

[0191] The third population generation module 137 is used to process the parent population and the offspring population to generate the next generation parent population. Specifically, it processes the parent population P... t and offspring population Q t Merge to generate a new population U t The new population U t The solutions are divided into an ε-feasible solution set S1 and an ε-infeasible solution set S2. When the number of solutions in S1 is less than the preset population size N, the solution with the fewest constraints violations is selected from S2 and added to S1. When the number of solutions in S1 is equal to the preset population size N, based on the elite selection strategy of the reference point, N solutions are selected from S1 to generate the next generation parent population P. t+1 .

[0192] The above process is iterated until the preset number of iterations is reached, and the output module 139 is used to output the target process parameter combination.

[0193] The optimization scheme selection module 150 is used to calculate the comprehensive score corresponding to each optimization scheme in the target process parameter combination, and to sort the optimization schemes according to a preset rule based on the comprehensive score. The optimization scheme selection module 150 may include at least: a first matrix generation module 151, a preprocessing module 152, a first calculation module 153, a second calculation module 154, a second matrix generation module 155, a third matrix generation module 156, a third calculation module 157, and a selection module 158.

[0194] The first matrix generation module 151 is configured to construct an initial evaluation matrix based on the optimization objectives in the target process parameter combination. Specifically, the target process parameter combination includes m optimization schemes for laser cladding process parameters, each optimization scheme includes n optimization objectives (i.e., quality characteristic parameters of the target composite coating), and the initial evaluation matrix X is constructed based on the optimization objectives. The predicted value of the jth optimization objective in the ith process parameter optimization scheme is denoted as x ij , by the target LSSVM model. The initial evaluation matrix X is as follows:

[0195]

[0196] The preprocessing module 152 is configured to preprocess the optimization objectives in the initial evaluation matrix based on a preset manner to obtain preprocessed optimization objectives. Specifically, since different types of optimization objectives have different evaluation standards, the types of optimization objectives need to be divided according to the index types, and then normalized or moderated according to the index types, and then standardized to ensure that the optimization objectives are under the same evaluation standard.

[0197] The first calculation module 153 is configured to calculate the preprocessed optimization objectives based on an entropy weight method to obtain first weights. The first calculation module 153 can at least include a first calculation unit 1531, a second calculation unit 1533, and a third calculation unit 1535.

[0198] The first calculation unit 1531 is configured to calculate the preprocessed optimization objectives to obtain index value ratios. Specifically, the index value ratio P ij of the jth optimization objective in the ith process parameter optimization scheme is taken as an example for illustration. The index value ratio P ij is as follows:

[0199]

[0200] wherein s ij is the jth optimization objective in the ith process parameter optimization scheme after processing, and m is the number of optimization schemes for laser cladding process parameters.

[0201] The second calculation unit 1533 is configured to calculate the index value ratios to obtain information entropy. Specifically, the information entropy E j is as follows:

[0202]

[0203] wherein E j ≥ 0, when P ij = 0, then P ij lnPij = 0, i.e. E j = 0.

[0204] The third calculation unit 1535 is configured to calculate the first weight based on the information entropy, and calculate a first score based on the first weight and the preprocessed optimization target. Specifically, the first weight ω 1j and the first score δ 1j are calculated according to the following formula:

[0205]

[0206] The second calculation module 154 is configured to calculate the second weight based on the coefficient of variation method. The second calculation module 154 at least includes a fourth calculation unit 1541, a fifth calculation unit 1542 and a sixth calculation unit 1543.

[0207] The fourth calculation unit 1541 is configured to calculate a mean value based on the preprocessed optimization target, and calculate a standard deviation based on the mean value and the preprocessed optimization target. Specifically, in the embodiment of the present application, the mean value A j and the standard deviation B j are calculated according to the following formula:

[0208]

[0209] The fifth calculation unit 1542 is configured to calculate a coefficient of variation based on the mean value and the standard deviation. Specifically, the coefficient of variation V j is calculated according to the following formula:

[0210]

[0211] The sixth calculation unit 1543 is configured to calculate the second weight based on the coefficient of variation, and calculate a second score based on the second weight and the preprocessed optimization target. Specifically, the second weight ω 2j and the second score δ 2j are calculated according to the following formula:

[0212]

[0213] The second matrix generation module 155 is configured to construct a weight matrix based on the first weight and the second weight. Specifically, the first weight and the second weight of each preprocessed optimization target are calculated to obtain a first weight set W1 = [ω 11 , ω 12 , …, ω 1j ] and a second weight set W2 = [ω 21 , ω 22,…,ω 2j A weight matrix W is constructed based on the first weight set and the second weight set.

[0214]

[0215] Where j is the number of optimization objectives after preprocessing.

[0216] The third matrix generation module 156 is used to construct a correlation matrix based on the first weight and the second weight. Specifically, it calculates the first weight and the second weight for each preprocessed optimization objective, and obtains the first weight set W1 = [ω 11 ,ω 12 ,…,ω 1j ] and the second weight set W2=[ω 21 ,ω 22 ,…,ω 2j To assess the correlation between W1 and W2, a correlation matrix R is constructed, as shown in the following formula:

[0217]

[0218] Where R is a j×j symmetric matrix, and r in the matrix ij Let be the Kendall rank correlation coefficient between the i-th preprocessed optimization objective in the first weight set W1 and the j-th preprocessed optimization objective in the second weight set W2. When i = j, the correlation coefficient is 1, indicating complete consistency of the same optimization objective in the two weight calculation methods.

[0219] The third calculation module 157 is used to calculate the combined weights based on the correlation matrix and the weight matrix. The third calculation module 157 may include at least a comparison unit 1571 and a seventh calculation unit 1573.

[0220] The comparison unit 1571 is used to calculate the target correlation coefficient based on the correlation matrix and compare the target correlation coefficient with a preset threshold to obtain a comparison result. Specifically, it calculates the sum of the correlation coefficients of the off-diagonal lines in the correlation matrix R to obtain the target correlation coefficient, and compares the target correlation coefficient with the preset threshold to obtain a comparison result. The preset threshold can be 0.05*n, where n is the number of quality characteristic parameters of the target composite coating.

[0221] The seventh calculation unit 1573 is used to calculate the combined weights based on the comparison results of the weight matrix. Specifically, when the target correlation coefficient is less than a preset threshold, it indicates that the entropy weight method and the coefficient of variation method lack consistency in weight ranking. A preliminary simple weighted average calculation is then performed to obtain the combined weights WC. jcombination weight WC j As the following formula:

[0222]

[0223] Wherein, ω ij is the weight value calculated according to different weight calculation methods, l is the number of weight values in the column of the weight matrix W, and n is the number of optimization objectives of each scheme.

[0224] When the target correlation coefficient is greater than the preset threshold, it indicates that the entropy weight method and the coefficient of variation method have high consistency in weight ordering, and the standard deviation λ i , the contradiction measure f i and the information carrying capacity c i of the elements in the correlation matrix R. i , the contradiction measure f i and the information carrying capacity c i As the following formula:

[0225]

[0226] Wherein, is the mean value of the jth column weight in the weight matrix W, r ij is the Kendall rank correlation coefficient of the ith preprocessed optimization objective in the first weight set W1 and the jth preprocessed optimization objective in the second weight set W2.

[0227] The information carrying capacity c i is standardized to generate a weight coefficient c' i , and the weight vector is weighted and averaged based on the weight coefficient c' i to obtain the combination weight WC j , and the combination weight WC j As the following formula:

[0228]

[0229] The selection module 158 is used for weighting and summing the preprocessed optimization objectives based on the combination weight, obtaining the comprehensive score corresponding to each optimization scheme, and sorting the optimization schemes according to the comprehensive score corresponding to each scheme according to a preset rule. Specifically, by weighting and summing the optimization objectives in the optimization scheme of i laser cladding process parameters and the combination weight, the comprehensive score Score i corresponding to each optimization scheme is obtained, and the optimization schemes are sorted according to the comprehensive score corresponding to each scheme according to a preset rule, so as to determine the order of the schemes. The comprehensive score Score iAs the following formula:

[0230]

[0231] Wherein, s ij is the jth optimization objective in the ith process parameter optimization scheme after processing, and n is the number of quality characteristic parameters of the target composite coating.

[0232] As can be seen from the above, in the laser cladding process parameter multi-objective optimization system of the present application, the target model generation module 110 generates a target LSSVM model based on the support vector regression algorithm, the laser cladding process parameters and the quality characteristic parameters of the composite coating, the target parameter combination determination module 130 optimizes the quality characteristic parameters of the composite coating based on the target LSSVM model and the DCNSGA-III algorithm, determines the target process parameter combination, and the optimization scheme selection module 150 calculates the comprehensive score corresponding to each optimization scheme in the target process parameter combination, and sorts the optimization schemes based on the comprehensive score according to the preset rule, thereby realizing efficient multi-objective optimization of process parameters and improving the optimization efficiency and quality of the performance of the ceramic composite coating.

[0233] In the description of the present specification, the description of the terms "one embodiment", "some embodiments", "example", "specific example", "one implementation", "one preferred implementation" or "some examples" and the like means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are contained in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0234] Although the embodiments of the present application have been shown and described, those skilled in the art can understand that various changes, modifications, replacements and variations can be made to the embodiments without departing from the principles and purposes of the present application, and the scope of the present application is defined by the claims and their equivalents.

Claims

1. A multi-objective optimization method for laser cladding process parameters, characterized in that, The multi-objective optimization method for laser cladding process parameters includes: A target LSSVM model is generated based on the support vector regression algorithm, laser cladding process parameters, and quality characteristic parameters of the composite coating. The quality characteristic parameters of the composite coating are optimized based on the target LSSVM model and the DCNSGA-Ⅲ algorithm to determine the target process parameter combination, including: Obtain the parameters of the DCNSGA-Ⅲ algorithm; The dynamic constraint boundaries of the laser cladding process parameters and the quality characteristic parameters of the composite coating are initialized to obtain the constraint conditions of the laser cladding process parameters and the quality characteristic parameters of the composite coating, respectively. Based on the quality characteristic parameters of the composite coating, the target LSSVM model, and the constraints, a parent population is generated. The offspring population is obtained by processing the parent population. Based on the parent population and the offspring population, the next generation parent population is generated by processing them. The above process is iterated until the preset number of iterations is reached, and the target combination of process parameters is output. Calculate the comprehensive score corresponding to each optimization scheme in the target process parameter combination, and sort the optimization schemes according to a preset rule based on the comprehensive score, including: An initial evaluation matrix is ​​constructed based on the optimization objectives in the target process parameter combination. The optimization objective in the initial evaluation matrix is ​​preprocessed based on a preset method to obtain the preprocessed optimization objective; The first weight is obtained by calculating the preprocessed optimization objective based on the entropy weight method; The second weight is obtained by calculating the preprocessed optimization objective based on the coefficient of variation method; Construct a weight matrix based on the first weight and the second weight; Construct a correlation matrix based on the first weight and the second weight; The combined weights are calculated based on the correlation matrix and the weight matrix. Based on the combined weights, the preprocessed optimization objectives are weighted and summed to obtain a comprehensive score for each optimization scheme, and the optimization schemes are sorted according to the preset rules based on the comprehensive scores.

2. The multi-objective optimization method for laser cladding process parameters according to claim 1, characterized in that, The generation of the target LSSVM model based on the support vector regression algorithm, laser cladding process parameters, and quality characteristic parameters of the composite coating includes: Based on the laser cladding process parameters and the quality characteristic parameters of the composite coating, a sample dataset is generated; The target LSSVM model is generated based on the support vector regression algorithm and the sample dataset.

3. The multi-objective optimization method for laser cladding process parameters according to claim 1, characterized in that, The calculation of the preprocessed optimization objective based on the entropy weight method to obtain the first weight includes: The ratio of index values ​​is calculated based on the preprocessed optimization objective. The information entropy is calculated based on the ratio of the aforementioned index values. The first weight is calculated based on the information entropy, and the first score is calculated based on the first weight and the preprocessed optimization objective.

4. The multi-objective optimization method for laser cladding process parameters according to claim 1, characterized in that, The calculation of the preprocessed optimization objective based on the coefficient of variation method to obtain the second weight includes: The mean is calculated based on the preprocessed optimization objective, and the standard deviation is calculated based on the mean and the preprocessed optimization objective. The coefficient of variation is calculated based on the mean and the standard deviation. The second weight is calculated based on the coefficient of variation, and the second score is calculated based on the second weight and the preprocessed optimization objective.

5. The multi-objective optimization method for laser cladding process parameters according to claim 1, characterized in that, The calculation of the combined weights based on the correlation matrix and the weight matrix includes: The target correlation coefficient is calculated based on the correlation matrix, and the target correlation coefficient is compared with a preset threshold to obtain a comparison result; Based on the comparison results, the combined weights are calculated from the weight matrix.

6. The multi-objective optimization method for laser cladding process parameters according to claim 5, characterized in that, The step of calculating the combined weights based on the comparison results includes: When the target correlation coefficient is less than the preset threshold, the combined weight is obtained by weighted average calculation, and the combined weight is: in, The weights are calculated for the preprocessed optimization objective using different weighting methods. The number of weights in the columns of the weight matrix. n This refers to the number of optimization objectives in each optimization scheme within the target process parameter combination.

7. The multi-objective optimization method for laser cladding process parameters according to claim 6, characterized in that, The step of calculating the combined weights based on the comparison results of the weight matrix further includes: When the target correlation coefficient is greater than the preset threshold, the standard deviation of the weights, the contradictory measure of the elements in the correlation matrix, and the information carrying capacity are calculated. The standard deviation, the contradictory measure, and the information carrying capacity are: in, λ i The standard deviation of the weights f i This serves as a measure of the inconsistency of the elements in the correlation matrix. c i For information carrying capacity, The weights are calculated for the preprocessed optimization objective using different weighting methods. The weight matrix is ​​the first... j The mean of the column weights, For the first weight set, the th i The preprocessed optimization objective and the second weight set j Kendall rank correlation coefficients of the preprocessed optimization objectives; The information carrying capacity is standardized to generate weight coefficients, and a weighted average is calculated based on these weight coefficients to obtain a combined weight, which is: in, The weighting coefficients are... WC j For combined weights.

8. A multi-objective optimization system for laser cladding process parameters, characterized in that, The multi-objective optimization system for laser cladding process parameters includes: a target model generation module, a target parameter combination determination module, and an optimization scheme selection module. The target model generation module is used to generate a target LSSVM model based on the support vector regression algorithm, laser cladding process parameters, and quality characteristic parameters of the composite coating. The target parameter combination determination module is used to optimize the quality characteristic parameters of the composite coating based on the target LSSVM model and the DCNSGA-Ⅲ algorithm, and to determine the target process parameter combination, including: Obtain the parameters of the DCNSGA-Ⅲ algorithm; The dynamic constraint boundaries of the laser cladding process parameters and the quality characteristic parameters of the composite coating are initialized to obtain the constraint conditions of the laser cladding process parameters and the quality characteristic parameters of the composite coating, respectively. Based on the quality characteristic parameters of the composite coating, the target LSSVM model, and the constraints, a parent population is generated. The offspring population is obtained by processing the parent population. Based on the parent population and the offspring population, the next generation parent population is generated by processing them. The above process is iterated until the preset number of iterations is reached, and the target combination of process parameters is output. The optimization scheme selection module is used to calculate the comprehensive score corresponding to each optimization scheme in the target process parameter combination, and to sort the optimization schemes according to a preset rule based on the comprehensive score, including: An initial evaluation matrix is ​​constructed based on the optimization objectives in the target process parameter combination. The optimization objective in the initial evaluation matrix is ​​preprocessed based on a preset method to obtain the preprocessed optimization objective; The first weight is obtained by calculating the preprocessed optimization objective based on the entropy weight method; The second weight is obtained by calculating the preprocessed optimization objective based on the coefficient of variation method; Construct a weight matrix based on the first weight and the second weight; Construct a correlation matrix based on the first weight and the second weight; The combined weights are calculated based on the correlation matrix and the weight matrix. Based on the combined weights, the preprocessed optimization objectives are weighted and summed to obtain a comprehensive score for each optimization scheme, and the optimization schemes are sorted according to the preset rules based on the comprehensive scores.

Images