Parameter optimization method for micro-milling size precision control of micro-thin-wall structure

Through parameter optimization methods, a multi-objective optimization model is established and combined with finite element verification, the problem of difficult control of dimensional accuracy in micro-milling machining of micro-milling of micro-milling of micro-wall structures is solved, high-quality and efficient production is achieved, and parts scrap rate and processing losses are reduced.

CN119065320APending Publication Date: 2024-12-03SHANDONG JIANZHU UNIV
View PDF 0 Cites 5 Cited by

Patent Information

Application Number
CN202411281973.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-13
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

In micro-milling processing, thin-wall structures are weak and easy to deform, and the processing dimensional accuracy is difficult to control, resulting in difficult assembly of parts and hidden dangers of service stability. The existing technology relies on experience or trial cutting tests, which is cost-effective and inefficient.

Method used

The parameter optimization method is used to obtain the discrete relationship between input and output parameters through experiments, establish the fitted equation, and build a multi-objective optimization model. The improved NSGA-II algorithm and entropy weight method are used to solve the problem and the TOPSIS method is combined with finite element verification. Optimal processing parameter combination.

Benefits of technology

The dimensional accuracy control of micro-milling machining of micro-milling in thin-wall structures is realized, reducing the scrap rate of parts, improving production efficiency, and reducing processing losses.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119065320A_ABST
    Figure CN119065320A_ABST
Patent Text Reader

Abstract

The invention discloses a parameter optimization method for micro-milling size precision control of a micro-thin-wall structure, and belongs to the field of cutting machining. According to the method, the relation between input variables (axial cutting depth, radial cutting depth, cutter rotating speed and feed per tooth) and output variables (micro-milling force, residual stress, top surface deformation and side surface deformation) in the micro-milling machining process is subjected to test and data processing, and a response surface method is utilized to establish a multi-objective optimization model; the model is solved through an improved NSGA-II algorithm, finally, decision making is conducted on a Pareto solution set through an entropy weight method, an optimal solution is selected, and auxiliary optimization verification is conducted on the optimal solution through a finite element model. The method can solve the problem of size precision control in the machining process of the micro-thin-wall structure, so that the machining quality is improved, the part rejection rate is reduced, and the production efficiency is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the field of cutting processing, and in particular to a method for optimizing processing parameters of micro-milling of a micro-thin-wall structure. Background Art

[0002] Micro-thin-wall structures have the characteristics of high thickness ratio and low rigidity and easy deformation during processing, which is especially serious for micro-milling of typical difficult-to-process metal materials. At the micro-machining scale, the machine tool stability, tool runout, size effect, machining thermal coupling effect and other uncontrollable factors are magnified during the processing of thin-walled microstructures. Unlike macro-scale processing, the above factors can no longer be ignored, and the cost of quantifying and controlling them is extremely high. The existence of machining dimensional accuracy deviation makes it difficult to assemble parts and components, which poses a hidden danger to the service stability and safety of parts and components as the core of mobility.

[0003] Normally, micro-milling of thin-walled structures requires processes such as blank preparation and parameter selection. The performance of the processing evaluation indicators of processing dimensional accuracy depends largely on the selection of a reasonable combination of processing parameters. The selection of processing parameters is mostly based on experience or certain trial cutting tests. By calibrating the optimization direction of processing indicators, clarifying the processing dimensional accuracy target, and performing numerical approximation based on intelligent optimization algorithms, a better processing parameter combination directly facing the target quantity is determined. This process does not require complicated theoretical analysis and calculations, and the data itself incorporates the influence of uncontrollable factors in micro-milling processing, which can effectively reduce the cost of trial and error. This method is rarely used in actual processing scenarios. Summary of the invention

[0004] In order to solve the shortcomings of the prior art based on actual needs, the present invention provides a parameter optimization method for dimensional accuracy control of micro-milling of thin-walled structures. On the basis of ensuring the processing quality of thin-walled parts, the test parameters during processing are optimized, thereby optimizing the micro-milling processing goals of thin-walled structures, so as to achieve high-quality and efficient production during the processing process, reduce the scrap rate of parts, and significantly improve production efficiency and reduce processing losses.

[0005] To achieve the above object, the technical solution adopted by the present invention is a parameter optimization method for dimensional accuracy control of micro-milling of thin-walled structures, comprising the steps of: S1. Obtain the discrete relationship between input parameters and output parameters through several experiments; S2. Solve the normal equations to determine the coefficients of the test values ​​and establish the fitting equation; S3. Based on the set parameter range and the established process requirements, a target-oriented multi-process parameter optimization model is constructed; S4. Use the improved NSGA-II algorithm to solve and obtain the Pareto solution set; S5. Use the entropy weight method and TOPSIS method to make decisions on the Pareto solution set and obtain the only optimal solution in the solution set; S6. Use finite element method to perform auxiliary optimization verification on the only optimal solution in the solution set to verify its rationality.

[0006] Preferably, in step S1, the input parameters include one or more of axial cutting depth, radial cutting depth, tool speed and feed per tooth; the output parameters include one or more of micro-milling force, residual stress, top surface deformation and side surface deformation.

[0007] Preferably, in step S3, the potential factors affecting the optimization target include axial cutting depth, radial cutting depth, tool speed and feed per tooth; the micro-milling force, residual stress, top surface deformation and side deformation of the micro-milled workpiece are taken as optimization targets, and the potential factors affecting the optimization target are analyzed through experiments to determine the potential factors that significantly affect the objective function; the functional relationship between the significant potential factors and the optimization target is fitted using a second-order response surface polynomial, the normal equations are solved to determine the coefficients in the second-order response surface polynomial, and a response surface equation is established.

[0008] Preferably, in step S4, within the allowable range of the parameters, the NSGA-II algorithm is used to perform multi-objective optimization solution of the response surface multi-objective optimization model for the micro-thin-wall structure milling process, and the step of determining the optimal input variables includes: taking axial cutting depth, radial cutting depth, tool speed, and feed per tooth as input variables, and taking micro-milling force, residual stress, top surface deformation, and side deformation as objective functions to construct a multi-objective optimization model; and performing the multi-objective optimization solution of the micro-thin-wall structure milling process through the multi-objective optimization model.

[0009] Preferably, in step S4, the second parameter of the improved NSGA-II algorithm includes the number of groups, the reference vector set, the maximum number of evaluations, the parent population, the fitness value, and the degree of crowding; the steps of multi-objective optimization of the parameter selection for dimensional accuracy control of micro-milling of micro-thin-walled structures using the improved RVEA algorithm include: S4-1. Initialize the population; then, choose whether to generate a primary population containing N individuals ; If yes, perform non-dominated sorting: disassemble the initial population and perform non-dominated sorting to obtain a non-dominated set: , , ,......, ; S4-2. Introducing the tournament selection mechanism for the initial population Perform crossover, mutation, and recombination to obtain a new set of N individuals , at this time the population generation number +1; S4-3. The primary population With the new collection Merge to get a population of 2N individuals ; S4-4. Gather Disassemble and perform non-dominated sorting to obtain a new generation of non-dominated sets: , , ,......, ; S4-5. Merge the above-mentioned new generation non-dominated sets into a second-generation population containing N individuals. The merging method is: put the individuals in the above non-dominated sets into the population one by one. If the remaining space is not enough to accommodate all the individuals in a non-dominated set, calculate the crowding degree of the individuals in the set. ; S4-6. The individuals are merged into the second-generation population in order of crowding degree, with priority given to those with greater crowding degree. At this time, the population generation number is +1. If the population generation number is still less than the set generation number, the fitness of the individuals in the above second-generation population is calculated, and the population is updated according to the individual fitness using the roulette wheel selection method.

[0010] 6. Preferably, the roulette wheel screening method includes: calculating the fitness of each individual in the population, calculating the probability of each individual being inherited to the next generation population, and calculating the cumulative probability of each individual And generate a uniformly distributed pseudo-random number in the interval [0, 1] r ,like r < Then select individual 1, otherwise, select individual So that: < r ≤ Established; then, the updated population is compared with the new set The merging continues to iterate until the population generation number reaches the set generation number, the Pareto solution set is output, and the NSGA-II algorithm is terminated.

[0011] Preferably, in step S5, the steps are included: S5-1 initializes each sample, regards each Pareto solution set as a sample, and regards the solution of each input variable in the sample as an indicator of the sample. For n samples and m indicators, Defined as: i The sample j indicators; S5-2 Normalize the indicators of each sample: For positive indicators , negative indicator , perform the following operations respectively: ; ; S5-3. Calculate the factors to be evaluated for each sample, including: j Under the indicator i The proportion of samples , No. j The entropy value of the index , information entropy redundancy ; in: ; ; ; Finally, the weight of each indicator is obtained ,Right now: ; S5-4. Use the TOPSIS method to calculate the comprehensive score of each sample , the process is as follows; S5-4-1. Standardize each indicator, namely: ; in is the standardized indicator; S5-4-2. Take the maximum and minimum values ​​of each column, that is: ; ; in, is the maximum value of the index after standardization of each column, is the minimum value of the standardized index of each column; S5-4-3. Calculate the i The distance between a sample and the maximum and minimum values ​​is: ; ; in, For the i The distance between samples and the maximum value; S5-4-4. Calculate the comprehensive score of each sample, namely: ; In the formula To calculate the composite score for each sample; S5-4-5. Based on the comprehensive scores of each sample, the user selects the sample with the highest score as the optimal solution. The advantage of this claim is that the user can obtain the optimal sample conveniently and accurately.

[0012] Preferably, step S6 includes, firstly, inputting the design points to construct the response surface, building the finite element model and verifying the accuracy, if the accuracy meets the requirements, setting the error threshold between the model prediction value and the finite element simulation value, otherwise adjusting the finite element parameters and re-calibrating until the accuracy meets the requirements; then, comparing the prediction value with the simulation value to calculate the error, if the error is greater than the set threshold, using the finite element model calculation result and the corresponding processing parameter combination as the verification point to correct the response surface, re-establishing and calculating the model, otherwise outputting the current processing parameter combination for actual processing.

[0013] It can be seen from the above technical scheme that the advantage of the present invention is that, when limited and discrete experimental data are obtained, the optimal solution for multiple objective functions can be obtained within the range; in the field of mechanical processing, the superiority of initial processing data is improved, the scrap rate of workpieces during processing is reduced, costs are saved, and production efficiency is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Attached Figure 1 A flow chart of the parameter optimization method for dimensional accuracy control of micro-milling of thin-walled structures proposed by the present invention; Attached Figure 2 This is the flow chart of the improved NSGA-II algorithm; Attached Figure 3 Flowchart for entropy weight method decision making; Attached Figure 4 Flowchart for finite element assisted optimization validation; DETAILED DESCRIPTION

[0015] Micro-thin-wall structures have the characteristics of high thickness ratio and low rigidity and easy deformation during processing, which is more serious for micro-milling of typical difficult-to-process metal materials. At the micro-machining scale, machine tool stability, tool runout, size effect, machining thermal coupling effect and other uncontrollable factors are magnified during the processing of micro-thin-wall structures. Different from the influence on machining dimensional accuracy at the macro scale, the above factors can no longer be ignored, but their quantification and control costs are extremely high. The existence of machining dimensional accuracy deviation makes it difficult to assemble parts and components, which poses a hidden danger to the service stability and safety of parts and components as the core of mobility.

[0016] In order to solve the above-mentioned defects of the prior art, the present invention provides a parameter optimization method for dimensional accuracy control of micro-milling of thin-walled structures. On the basis of ensuring the processing quality of thin-walled parts, the test parameters during processing are optimized, thereby optimizing the processing process of micro-milling of thin-walled parts, so as to achieve high-quality and efficient production during the processing process, reduce the scrap rate of parts, and significantly improve production efficiency and reduce processing losses.

[0017] Please refer to Figure 1 , Figure 1 A flow chart of a parameter optimization method for dimensional accuracy control in micro-milling of micro-thin-wall structures provided according to some embodiments of the present invention is shown.

[0018] like Figure 1 As shown, the parameter optimization method 100 for dimensional accuracy control of micro-milling of micro-thin-wall structures may include step S1: obtaining a discrete relationship between input (test parameters) and output (response parameters) through several tests.

[0019] Specifically, the discrete relationship between the input variables and the output variables can be obtained through a data acquisition device according to the actual situation of the five-axis high-speed machining center during micro-milling; the second-order response surface polynomial is used for fitting, and the polynomial is as follows: ; in, is the output variable, , is the regression coefficient, is the input variable, To evaluate the error.

[0020] like Figure 1 As shown, the parameter optimization method 100 for dimensional accuracy control of micro-milling of micro-thin-walled structures may include step S2: fitting the above discrete relationship with a quadratic response surface equation, and solving the corresponding normal equations to determine the regression coefficients, and establishing a fitting equation; Specifically, after obtaining the discrete relationship between the input variable and the output variable, in order to determine the input variable coefficients in the second-order response surface polynomial, the normal equations corresponding to the response surface polynomial are solved to obtain the input variable coefficients. The polynomial response surface model based on the least squares principle minimizes the estimated value Compared with the measured value The square of the difference is used to estimate the regression coefficient, that is: ; in, The minimized estimate of the output quantity, is the measured value of the output.

[0021] Solving by fitting method: According to the requirements of the quadratic fitting polynomial, the normal equations for the polynomial are as follows: ; ; ; Solving the above normal equation, we can get the regression coefficient in the fitting polynomial , , , and finally obtain the response surface equation of the continuous relationship between the input variables (i.e., the various process parameters that need to be set before processing) and the output variables (i.e., the dependent variables affected by the input parameters).

[0022] like Figure 1 As shown, the parameter optimization method 100 for dimensional accuracy control of micro-milling of thin-walled structures may include step S3: determining the range of each response value and establishing a multi-objective optimization model based on the response surface.

[0023] Specifically, in a non-limiting example, the axial cutting depth, radial cutting depth, tool speed and feed per tooth can be used as input variables, and the micro-milling force, residual stress, top surface deformation and side deformation can be used as output variables; in order to improve the processing quality of side milling of micro-thin walls and extend the tool life, the periodic term coefficient in the milling force expression was corrected in the study; since the milling force is taken as the average amplitude of the measured signal, the milling force can be stabilized by minimizing the corrected expression; the residual stress is the result of the redistribution of internal stress caused by the combined action of mechanical load and thermal load; under the action of hot and cold aging, the residual stress on the surface of the micro-thin wall will experience attenuation and re-emergence. The new balance process may cause secondary deformation or cracks in the thin-walled structure, so the residual stress needs to be strictly controlled; the top surface deformation and the side surface deformation are key indicators for measuring dimensional accuracy, so the minimization of these deformations is taken as the optimization direction; in addition, in order to facilitate layered milling, constraints are added to each input variable: the overall height of the thin wall must be divisible by the axial cutting depth; based on the above factors, a multi-objective optimization model is formed, which aims to simultaneously optimize the stability of the milling force, control the residual stress, minimize the top and side surface deformation, and meet specific layered milling constraints; in summary, the optimization model formed by each input variable and output variable is: ; in, , , , is the function expression fitted between each input and output. , , , are the minimum values ​​of each input variable, , , , are the maximum values ​​of each input variable.

[0024] like Figure 1 As shown, the parameter optimization method 100 for dimensional accuracy control of micro-milling of thin-walled structures may include step S4: using the improved NSGA-II algorithm to solve and obtain a Pareto solution set.

[0025] For the specific implementation of this algorithm, please refer to Figure 2 , which is a flow chart of an improved NSGA-II algorithm proposed according to some embodiments of the present method.

[0026] like Figure 2 As shown, first, the population is initialized; then, it is decided whether to generate an initial population containing N individuals. , for the initial population , whose size is N, and each individual is represented by , and its objective function value is ; If yes, perform non-dominated sorting: disassemble the initial population and perform non-dominated sorting to obtain a non-dominated set: , , ,......, , then the initial population is subjected to adaptive crossover mutation and local search from extensive to detailed to regenerate the initial population ; Then, the tournament selection mechanism was introduced to the initial population Perform crossover, mutation, and recombination to obtain a new set of N individuals , at this time the population generation +1; then, the initial population With the new collection Merge to get a population of 2N individuals ; Then, collect Disassemble and perform non-dominated sorting to obtain a new generation of non-dominated sets: , , ,......, ; At this time, the above-mentioned new generation non-dominated sets are merged into a second-generation population containing N individuals. The merging method is: the individuals in the above non-dominated sets are put in one by one. If the remaining space is not enough to accommodate all the individuals in a non-dominated set, the crowding degree of the individuals in the set is calculated. , in, ; Afterwards, they are merged into the second-generation population in order of crowding degree, with the ones with greater crowding degree given priority. At this time, the population generation number is +1; afterwards, if the population generation number is still less than the set generation number, the fitness of the individuals in the second-generation population is calculated, and the population is updated according to the individual fitness using the roulette method. The specific method is: calculate the fitness of each individual in the population, calculate the probability of each individual being inherited to the next generation population, calculate the cumulative probability of each individual and Generates a uniformly distributed pseudo-random number in the interval [0, 1] r ,like r < , then select individual 1, otherwise, select individual , so that: < r ≤ Established; then, the updated population is compared with the new set The merging continues to iterate until the population generation number reaches the set generation number, the Pareto solution set is output, and the improved NSGA-II algorithm is terminated.

[0027] like Figure 1 As shown, the parameter optimization method 100 for dimensional accuracy control of micro-milling of thin-walled structures may include step S5: using the entropy weight method and the TOPSIS method to make decisions on the Pareto solution set, and obtaining a unique optimal solution according to the requirements.

[0028] For the specific implementation of this decision-making method, please refer to Figure 3 , which is a decision flow chart of the entropy weight method proposed according to some embodiments of the present method.

[0029] like Figure 3 As shown in the figure, the entropy weight method is first used to calculate the weight of each indicator of each sample, and each sample is initialized: each solution in the Pareto solution set is regarded as a sample, and each input variable in the sample is regarded as an indicator of the sample. For n samples and m indicators, No. i The sample j indicators.

[0030] After that, normalize the indicators of each sample: , negative indicator , perform the following operations respectively: ; ; Then, the factors to be evaluated for each sample are calculated separately, including: j Under the indicator i The proportion of samples , No.j The entropy value of the index , information entropy redundancy .

[0031] in: ; ; ; Finally, the weight of each indicator is obtained ,Right now: ; Then, the TOPSIS method was used to calculate the comprehensive score of each sample. , the process is as follows: First, standardize each indicator, namely: ; in Standardized indicators; Then, take the maximum and minimum values ​​of each column, that is: ; ; Next, calculate the i The distance between a sample and the maximum and minimum values ​​is: ; ; in, For the i The distance between samples and the maximum value, For the i The distance between samples and the minimum value; Then, the comprehensive score of each sample is calculated, namely: ; In the formula To calculate the composite score for each sample; Finally, the user selects the sample with the highest score as the optimal solution based on the comprehensive scores of each sample.

[0032] like Figure 1 As shown, the parameter optimization method 100 for dimensional accuracy control of micro-milling of thin-walled structures may include step S6: using finite elements to perform auxiliary optimization verification.

[0033] For the specific implementation of this decision-making method, please refer to Figure 4 , which is a finite element assisted verification flow chart proposed according to some embodiments of the present method.

[0034] like Figure 4 As shown, firstly, the design points are input to construct the response surface, the finite element model is built and the accuracy is verified. If the accuracy meets the requirements, the error threshold between the model prediction value and the finite element simulation value is set, otherwise the finite element parameters are adjusted and re-calibrated until it meets the requirements; then, the prediction value is compared with the simulation value to calculate the error. If the error is greater than the set threshold, the finite element model calculation result and the corresponding processing parameter combination are used as the verification point to correct the response surface, and the model is re-established and calculated, otherwise the current processing parameter combination is output for actual processing.

[0035] This description provides sufficient details to enable those skilled in the relevant art to implement the present invention. Although some embodiments have been described, it is obvious that those skilled in the art can make various modifications thereto and apply these general principles to other variations without departing from the spirit or scope of the present invention. Therefore, the present invention should not be limited to the specific examples and designs described herein, but should enjoy the broadest scope of protection consistent with the principles and novel features disclosed in this specification.

[0036] For those skilled in the art, according to the teachings of the present invention, without departing from the principles and spirit of the present invention, changes, modifications, substitutions and variations made to the implementation methods are still within the protection scope of the present invention.

Claims

1. A parameter optimization method for dimensional accuracy control of micro-milling of thin-walled structures, characterized in that: Includes steps: S1. Obtain the discrete relationship between input parameters and output parameters through several experiments; S2. Solve the normal equations to determine the coefficients of the test values ​​and establish the fitting equation; S3. Based on the set parameter range and the established process requirements, a target-oriented multi-process parameter optimization model is constructed; S4. Use the improved NSGA-II algorithm to solve and obtain the Pareto solution set; S5. Use the entropy weight method and TOPSIS method to make decisions on the Pareto solution set and obtain the only optimal solution in the solution set; S6. The only optimal solution in the solution set is verified by auxiliary optimization using finite elements to verify its rationality. The advantage of this claim is that a set of experimental data can be optimized to obtain the optimal processing data and reduce the scrap rate of workpieces.

2. The parameter optimization method for dimensional accuracy control of micro-milling of micro-thin-wall structures according to claim 1 is characterized in that: In step S1, the input parameters include one or more of axial cutting depth, radial cutting depth, tool speed and feed per tooth; the output parameters include one or more of micro-milling force, residual stress, top surface deformation and side surface deformation. The advantage of this claim is that the input parameters and output parameters can be determined for subsequent optimization.

3. The parameter optimization method for dimensional accuracy control of micro-milling of micro-thin-wall structures according to claim 1 is characterized in that: In step S3, the potential factors affecting the optimization target include axial cutting depth, radial cutting depth, tool speed and feed per tooth; the micro-milling force, residual stress, top surface deformation and side deformation of the micro-milling workpiece are taken as optimization targets, and the potential factors affecting the optimization target are analyzed through experiments to determine the potential factors that significantly affect the objective function; the functional relationship between the significant potential factors and the optimization target is fitted using a second-order response surface polynomial, the coefficients in the second-order response surface polynomial are determined by solving the normal equations, and the response surface equation is established. The advantage of this claim is that a continuous relationship between input parameters and output parameters is established for algorithm calculation.

4. The parameter optimization method for dimensional accuracy control of micro-milling of micro-thin-wall structures according to claim 1 is characterized in that: In step S4, within the allowable range of the parameters, the improved NSGA-II algorithm is used to perform multi-objective optimization solution of the response surface multi-objective optimization model for the micro-thin-wall structure milling process, and the step of determining the optimal input variables includes: taking axial cutting depth, radial cutting depth, tool speed, and feed per tooth as input variables, and taking micro-milling force, residual stress, top surface deformation, and side deformation as objective functions to construct a multi-objective optimization model; and performing the multi-objective optimization solution of the micro-thin-wall structure milling process through the multi-objective optimization model. The advantage of this claim is that the improved NSGA-II algorithm can solve the model more accurately and quickly.

5. The parameter optimization method for dimensional accuracy control of micro-milling of micro-thin-wall structures according to claim 1 is characterized in that: In step S4, the second parameters of the improved NSGA-II algorithm include the number of groups, the reference vector set, the maximum number of evaluations, the parent population, the fitness value, and the degree of crowding; The steps of multi-objective optimization of parameter selection for dimensional accuracy control of micro-milling of micro-thin-walled structures using the improved NSGA-II algorithm include: S4-1. Initialize the population; Then, choose whether to generate a first-generation population containing N individuals If yes, perform non-dominated sorting: disassemble the initial population and perform non-dominated sorting to obtain a non-dominated set: , , ,......, Otherwise, the initial population is subjected to adaptive crossover mutation and local search from extensive to detailed to regenerate the initial population. ; S4-2. Introducing the tournament selection mechanism for the initial population Perform crossover, mutation, and recombination to obtain a new set of N individuals At this time, the population generation number is +1; S4-3. The primary population With the new collection Merge to get a population of 2N individuals ; S4-4. Gather Disassemble and perform non-dominated sorting to obtain a new generation of non-dominated sets: , , ,......, ; S4-5. Merge the above-mentioned new generation non-dominated sets into a second-generation population containing N individuals. The merging method is: put the individuals in the above non-dominated sets into the population one by one. If the remaining space is not enough to accommodate all the individuals in a non-dominated set, calculate the crowding degree of the individuals in the set. ; S4-6. The individuals are merged into the second-generation population in order of crowding degree, with priority given to those with greater crowding degree. At this time, the population generation number is +1. If the population generation number is still less than the set generation number, the fitness of the individuals in the second-generation population is calculated, and the population is updated according to the individual fitness by the roulette wheel screening method. The advantage of this claim is that the roulette wheel screening and other links are added to the improved NSGA-II algorithm to ensure the quality of the solution set.

6. The parameter optimization method for dimensional accuracy control of micro-milling of micro-thin-wall structures according to claim 4 is characterized in that: The roulette wheel screening method comprises: calculating the fitness of each individual in the group, calculating the probability of each individual being inherited to the next generation group, calculating the cumulative probability of each individual and Generates a uniformly distributed pseudo-random number in the interval [0, 1]. r ,like r < , then select individual 1, otherwise, select individual , so that: < r ≤ Established; then, the updated population is compared with the new set The merging continues to iterate until the population generation number reaches the set generation number, the Pareto solution set is output, and the improved NSGA-II algorithm is terminated. The advantage of this claim is that a higher quality solution set is obtained through the roulette screening method.

7. The parameter optimization method for dimensional accuracy control of micro-milling of thin-walled structures according to claim 1 is characterized in that: In step S5, the steps are included: S5-1. Initialize each sample, regard each Pareto solution set as a sample, and regard the solution of each input variable in the sample as an indicator of the sample. For n samples and m indicators, Defined as: i The sample j indicators; S5-2. Normalize the indicators of each sample: For positive indicators , negative indicator , perform the following operations respectively: ; ; S5-3. Calculate the factors to be evaluated for each sample, including: j Under the indicator i The proportion of samples No. j The entropy value of the index , information entropy redundancy ; in: ; ; ; Finally, the weight of each indicator is obtained ,Right now: ; S5-4. Use the TOPSIS method to calculate the comprehensive score of each sample , the process is as follows; S5-4-1. Standardize each indicator, namely: ; in is the standardized indicator; S5-4-2. Take the maximum and minimum values ​​of each column, that is: ; ; in, is the maximum value of the index after standardization of each column, It is the minimum value of the index after standardization of each column; S5-4-3. Calculate the i The distance between a sample and the maximum and minimum values ​​is: ; ; in, For the i The distance between samples and the maximum value; S5-4-4. Calculate the comprehensive score of each sample, namely: ; In the formula To calculate the composite score for each sample; S5-4-5. Based on the comprehensive scores of each sample, the user selects the sample with the highest score as the optimal solution. The advantage of this claim is that the user can obtain the optimal sample conveniently and accurately.

8. The parameter optimization method for dimensional accuracy control of micro-milling of micro-thin-wall structures according to claim 1 is characterized in that: Step S6 includes, first inputting the design point to construct the response surface, building the finite element model and verifying the accuracy, if the accuracy meets the requirements, setting the error threshold between the model prediction value and the finite element simulation value, otherwise adjusting the finite element parameters and re-calibrating until it meets the requirements; then, comparing the prediction value with the simulation value to calculate the error, if the error is greater than the set threshold, using the finite element model calculation result and the corresponding processing parameter combination as the verification point to correct the response surface, re-establishing and calculating the model, otherwise outputting the current processing parameter combination for actual processing. The advantage of this claim is that it can verify and improve the rationality and feasibility of the optimal solution described in claim 7.

Citation Information

Cited By

  • Control method and device for filler-free welding of stainless steel thin-wall tank

    CN120421722A

  • Redundant angle and process parameter joint optimization method and equipment for complex curved surface robot machining

    CN120572524A

  • Multi-target cutting parameter optimization method

    CN122284508A

  • Five-axis turn-milling blade parameter optimization method based on multi-objective genetic algorithm

    CN122310999A

  • A method and system for multi-objective optimization of cutting parameters by fusing six-dimensional point cloud and NSGA-III

    CN122818975A