A method and device for optimizing parameters of a low-stress milling cutter for an aero-engine casing

Abstract

The application relates to the technical field of aircraft engine manufacturing, and discloses an aircraft engine casing low-stress milling cutter parameter optimization method and device, which comprises the following steps: different cutters are used to mill the casing under the same preset machining process condition, surface residual stress parameters are collected, the collected surface residual stress parameters are standardized, a target function based on the weighted mean of the surface residual stress parameters corresponding to the cutter parameters is constructed based on the standardization result, a multi-objective low-stress milling optimization model is established with the minimum target function as the optimization target, a neural network optimized by the pelican optimization algorithm is used to establish the mapping relationship between the cutter parameters and the target function value, and the genetic algorithm is used to optimize the cutter parameters based on the mapping relationship to obtain optimal cutter parameters. The application solves the technical problems that the synergistic influence of various residual stress parameters is not comprehensively considered in the existing aircraft engine casing milling cutter parameter optimization.

Description

Technical Field

[0001] This invention relates to the field of aero-engine manufacturing technology, and discloses a method and apparatus for optimizing the parameters of low-stress milling tools for aero-engine casings. Background Technology

[0002] As a core load-bearing component of the engine, the machining quality of the aircraft engine casing directly affects the engine's operational safety and reliability. Milling is a critical process in casing manufacturing, and the surface residual stresses generated during machining (such as the outermost residual stress, peak tensile stress, and peak compressive stress) significantly affect the casing's fatigue strength, corrosion resistance, and dimensional stability. If the residual stress distribution is unreasonable (e.g., excessive tensile stress), it may lead to crack propagation, deformation, and other failures during service, seriously threatening flight safety.

[0003] Currently, tool parameters (such as blunt radius, rake angle, and clearance angle) are key factors affecting residual stress on milled surfaces. Existing technologies often rely on accumulated experience or single residual stress parameters for tool parameter optimization, which has the following shortcomings: 1. They fail to comprehensively consider the synergistic effects of multiple residual stress parameters, resulting in a one-sided optimization objective; 2. They lack standardized processing of residual stress parameters, and directly involving stress parameters of different magnitudes in optimization can easily lead to result deviations.

[0004] Therefore, this invention proposes a method and apparatus for optimizing the parameters of low-stress milling tools for aero-engine casings to solve the above-mentioned technical problems. Summary of the Invention

[0005] The purpose of this invention is to provide a method and apparatus for optimizing the parameters of low-stress milling tools for aero-engine casings, thereby solving the technical problems in the existing optimization of milling tool parameters for aero-engine casings, such as the failure to comprehensively consider the synergistic effects of multiple residual stress parameters, the one-sided optimization objectives, the lack of standardized processing of residual stress parameters, and the tendency for deviations in results to occur when stress indicators of different magnitudes are directly involved in the optimization.

[0006] To achieve the above-mentioned technical effects, the technical solution adopted by the present invention is as follows:

[0007] A method for optimizing the parameters of a low-stress milling tool for an aero-engine casing includes:

[0008] The casing is milled using different cutting tools under the same preset machining process conditions. The residual stress parameters of the milled surface are collected. The cutting tool parameters include the tool blunt radius, rake angle and clearance angle. The residual stress parameters of the surface include the outermost residual stress, peak tensile stress and peak compressive stress.

[0009] By standardizing the collected surface residual stress parameters, and based on the standardization results, an objective function is constructed based on the weighted mean of the surface residual stress parameters corresponding to the tool parameters.

[0010] With the objective function minimization as the optimization objective, a multi-objective low-stress milling optimization model is established, and a neural network optimized by the Pelican optimization algorithm is used to establish the mapping relationship between tool parameters and objective function values;

[0011] The genetic algorithm is applied to optimize the tool parameters based on the mapping relationship to obtain the optimal tool parameters.

[0012] As a preferred implementation, the standardization formula is:

[0013]

[0014] Where A is the standardized outermost residual stress, B is the standardized peak tensile stress, and C is the standardized peak compressive stress. The original value of the surface residual stress in the milling of the housing by the q-th tool. The original value of the peak tensile stress during the q-th milling operation of the tool on the housing. The original value of the peak compressive stress during the q-th milling operation of the housing by the tool. for maximum value for minimum value for maximum value for minimum value for maximum value for The minimum value.

[0015] As a preferred embodiment, the objective function is:

[0016]

[0017] Where Y is the objective function value, a is the A weight coefficient determined by the analytic hierarchy process (AHP), b is the B weight coefficient determined by the AHP, and c is the C weight coefficient determined by the AHP, satisfying a+b+c=1 and a>b>c>0.

[0018] As a preferred embodiment, the application of the genetic algorithm to optimize tool parameters based on the mapping relationship includes: initializing a tool parameter population, wherein each individual in the initial tool parameter population is a combination of tool parameters, the fitness is set by the objective function value, and the initial tool parameter population is updated by selection, crossover and mutation operations; iterating until a preset number of iterations is reached, and outputting the tool parameters corresponding to the optimal individual.

[0019] Based on the above solution, the present invention also proposes a device for optimizing the parameters of low-stress milling tools for aero-engine housings, comprising:

[0020] The machining acquisition module is used to perform milling on the casing using different tools under the same preset machining process conditions, and to acquire residual stress parameters on the milled surface. The tool parameters include the tool blunt radius, rake angle and clearance angle, and the surface residual stress parameters include the outermost residual stress, peak tensile stress and peak compressive stress.

[0021] The standardization and objective function module is used to standardize the collected surface residual stress parameters and, based on the standardization results, construct an objective function based on the weighted mean of the surface residual stress parameters corresponding to the tool parameters.

[0022] The mapping module is used to establish a multi-objective low-stress milling optimization model with the objective function as the optimization objective, and to establish the mapping relationship between tool parameters and objective function values ​​using a neural network optimized by the Pelican optimization algorithm.

[0023] The parameter optimization module is used to apply a genetic algorithm to optimize tool parameters based on the mapping relationship and obtain the optimal tool parameters.

[0024] As a preferred implementation, the standardization processing formula performed by the standardization and objective function module is as follows:

[0025]

[0026] Where A is the standardized outermost residual stress, B is the standardized peak tensile stress, and C is the standardized peak compressive stress. The original value of the surface residual stress in the milling of the housing by the q-th tool. The original value of the peak tensile stress during the q-th milling operation of the tool on the housing. The original value of the peak compressive stress during the q-th milling operation of the housing by the tool. for maximum value for minimum value for maximum value for minimum value for maximum value for The minimum value.

[0027] As a preferred implementation, the objective function constructed by the standardization and objective function module is:

[0028]

[0029] Where Y is the objective function value, a is the A weight coefficient determined by the analytic hierarchy process (AHP), b is the B weight coefficient determined by the AHP, and c is the C weight coefficient determined by the AHP, satisfying a+b+c=1 and a>b>c>0.

[0030] As a preferred embodiment, the genetic algorithm executed by the parameter optimization module includes: initializing a tool parameter population with a population size of 30-60; using roulette wheel selection for the selection operation, having a crossover probability of 0.6-0.8 for the crossover operation, and a mutation probability of 0.01-0.05 for the mutation operation; and setting a preset number of iterations of 50-100, outputting the optimal tool parameters after the iteration terminates.

[0031] Compared with traditional methods, the present invention has the following advantages:

[0032] I. Comprehensive optimization to reduce residual stress: Simultaneously considering the residual stress of the outermost layer, the peak tensile stress, and the peak compressive stress, the weighted objective function achieves multi-index synergistic optimization, effectively reducing the residual stress on the casing surface and improving its fatigue strength and service life.

[0033] II. High-precision mapping: The pelican optimization algorithm is used to optimize the neural network, which solves the problem of low prediction accuracy of traditional neural networks and ensures a more accurate mapping relationship between tool parameters and target function values.

[0034] III. High-efficiency optimization: Genetic algorithms converge quickly through selection, crossover, and mutation operations, and can find the globally optimal tool parameters in a large parameter space, significantly improving optimization efficiency.

[0035] IV. High Applicability: The method and apparatus can be extended to low-stress milling of other high-precision mechanical parts, and have broad engineering application value. Attached Figure Description

[0036] Figure 1 This is a flowchart of the method for optimizing the parameters of low-stress milling tools for aero-engine casings according to the present invention.

[0037] Figure 2 This is a structural block diagram of the low-stress milling tool parameter optimization device for aero-engine casing according to the present invention.

[0038] Figure 3 This is a graph showing the average relative error of neural network training under different numbers of hidden layer nodes according to the present invention.

[0039] Figure 4 This is a schematic diagram of the neural network structure optimized by the Pelican optimization algorithm in an embodiment of the present invention.

[0040] Figure 5 This is a flowchart illustrating the Pelican optimization algorithm in an embodiment of the present invention.

[0041] Figure 6 This is one of the comparison charts between the objective function value and the actual value of the multi-objective low-stress milling optimization model predicted by the neural network training set optimized by the Pelican optimization algorithm in an embodiment of the present invention.

[0042] Figure 7 This is the second comparison chart between the objective function value and the actual value of the multi-objective low-stress milling optimization model predicted by the neural network test set optimized by the Pelican optimization algorithm in this embodiment of the invention.

[0043] Figure 8 This is a graph showing the evolutionary curve of the fitness of individuals optimized by the genetic algorithm in an embodiment of the present invention.

[0044] 1. Processing and acquisition module; 2. Standardization and objective function module; 3. Mapping relationship module; 4. Parameter optimization module. Detailed Implementation

[0045] The present invention will now be described in further detail with reference to the embodiments and accompanying drawings. However, this should not be construed as limiting the scope of the above-described subject matter of the present invention to the following embodiments; all technologies implemented based on the content of the present invention fall within the scope of the present invention.

[0046] like Figure 1 As shown in Example 1, a method for optimizing tool parameters in low-stress milling of aero-engine casings is described. This method achieves tool parameter optimization through multi-step collaborative processes, specifically including:

[0047] Step 1: Machining and Parameter Acquisition: Mill the casing using different tools under the same preset machining process conditions, and collect surface residual stress parameters; Controlled variables: Only change the tool parameters (including tool blunt radius, rake angle, and clearance angle), keep other machining process conditions (such as cutting speed, feed rate, depth of cut, cooling method, etc.) consistent, and ensure that the difference in residual stress is only caused by changes in tool parameters; Acquisition objects: Record the surface residual stress parameters after each machining, including the outermost residual stress ( Peak tensile stress () Peak compressive stress ( ).

[0048] Step 2: Standardization and Objective Function Construction: The collected residual stress parameters are standardized to eliminate dimensional differences, and then an objective function is constructed based on the standardization results. Standardization: The original stress parameters are converted into dimensionless standardized values ​​(A, B, C) using the following formulas:

[0049]

[0050] Where A is the standardized outermost residual stress, B is the standardized peak tensile stress, and C is the standardized peak compressive stress. The original value of the surface residual stress in the milling of the housing by the q-th tool. The original value of the peak tensile stress during the q-th milling operation of the tool on the housing. The original value of the peak compressive stress during the q-th milling operation of the housing by the tool. for maximum value for minimum value for maximum value for minimum value for maximum value for The minimum value.

[0051] Step 3: Constructing the Objective Function: The weighted mean of the standardized values ​​is used as the objective function, which is:

[0052]

[0053] Where Y is the objective function value (which needs to be minimized), a is the A weight coefficient determined by the analytic hierarchy process (AHP), b is the B weight coefficient determined by the AHP, and c is the C weight coefficient determined by the AHP. The condition a+b+c=1 and a>b>c>0 is met. The condition a>b>c>0 reflects that the residual stress of the outermost layer has the highest priority, followed by the peak compressive stress, and the peak tensile stress has the lowest priority.

[0054] Step 4: Taking "minimizing the objective function Y" from Step 3 as the optimization objective, establish a multi-objective low-stress milling optimization model, and use the Pelican optimization algorithm to optimize the neural network to construct the mapping relationship between tool parameters and objective function values.

[0055] Neural network structure: The input layer is the tool parameters (blunt radius, front angle, back angle), the output layer is the target function value Y, and the hidden layers are set according to the sample size (e.g., 2-3 layers, 10-20 neurons per layer).

[0056] Pelican optimization algorithm optimization: By simulating the "dive hunting" behavior of pelicans, the initial weights and biases of the neural network are optimized to reduce prediction errors caused by random initialization and improve the accuracy of the mapping relationship (that is, after inputting tool parameters, the corresponding Y value can be accurately output).

[0057] Step 5: Tool parameter optimization. Apply a genetic algorithm to optimize the tool parameters based on the above mapping relationship to obtain the optimal solution.

[0058] Population initialization: Generate 30-60 initial individuals, each individual being a set of tool parameter combinations (reasonable range of values ​​for blunt radius, rake angle, and clearance angle).

[0059] Fitness function: The fitness is the objective function value Y from step three (the smaller Y is, the higher the fitness).

[0060] Genetic manipulation:

[0061] Selection: A roulette wheel selection method is used, selecting high-quality individuals to enter the next generation based on fitness probability;

[0062] Crossover: The crossover probability is set to 0.6-0.8, and the parameters of the selected individuals are randomly recombined;

[0063] Mutation: The mutation probability is set to 0.01-0.05, and individual parameters are randomly changed to increase diversity.

[0064] Iteration Termination: Repeat the above operation until the preset number of iterations (50-100 times), and output the tool parameters corresponding to the individual with the highest fitness, which are the optimal tool parameters.

[0065] When using,

[0066] S1: Preset machining process conditions: cutting speed 100m / min, feed rate 0.1mm / r, cutting depth 2mm, cooling method is oil mist cooling. In the milling machine casing cutting experiment, a three-factor five-level orthogonal experimental design is adopted, that is, the tool blunt circle radius r, the tool rake angle γ and the tool clearance angle α are used as three factors, and five levels are designed under each factor. The orthogonal experimental design is shown in Table 1.

[0067] Table 1

[0068]

[0069] Measure the residual stress parameters of the milled surfaces separately: residual stress of the outermost layer Peak tensile stress and peak compressive stress Establish L 25The orthogonal array, as shown in Table 2, uses the results of orthogonal experiments based on range analysis to obtain the degree of influence of tool parameters on the residual stress parameters of the milling surface: tool blunt radius r > tool rake angle γ > tool clearance angle α. The importance of each residual stress parameter of the milling surface to the effect of low-stress milling is determined: outermost residual stress > peak tensile stress > peak compressive stress. The goal of optimizing tool parameters for low-stress milling is clarified: to obtain a milling surface with large compressive stress, large peak compressive stress and small peak tensile stress, that is, to minimize the value of each residual stress parameter of the milling surface.

[0070] Table 2.L 25 Orthogonal array

[0071]

[0072] S2: The standardization formula is:

[0073]

[0074] Where A is the standardized outermost residual stress, B is the standardized peak tensile stress, and C is the standardized peak compressive stress. The original value of the surface residual stress in the milling of the housing by the q-th tool. The original value of the peak tensile stress during the q-th milling operation of the tool on the housing. The original value of the peak compressive stress during the q-th milling operation of the housing by the tool. for maximum value for minimum value for maximum value for minimum value for maximum value for The minimum value.

[0075] S2.1: The specific form of the multi-objective low-stress milling optimization model is as follows:

[0076]

[0077] Where Y is the objective function value, a is the A weight coefficient determined by the analytic hierarchy process, b is the B weight coefficient determined by the analytic hierarchy process, and c is the C weight coefficient determined by the analytic hierarchy process, satisfying a+b+c=1 and a>b>c>0. The setting of the weight coefficients reflects the priority of the influence of each residual stress parameter on the milling quality.

[0078] Based on experimental data and range analysis results, the numerical values ​​of various aspects of the multi-objective low-stress milling optimization model were obtained using the analytic hierarchy process (AHP). The multi-objective low-stress milling optimization model is as follows:

[0079] ;

[0080] In S2.1: The weight coefficients are determined by the Analytic Hierarchy Process (AHP).

[0081] To determine reasonable weighting coefficients , , The importance of the three residual stress indices was compared and scored pairwise using the analytic hierarchy process (AHP).

[0082] (1) Construct the judgment matrix

[0083]

[0084] (2) Calculate the eigenvectors and normalize them.

[0085] The judgment matrix is ​​normalized and the average value of each row is calculated to obtain the weight vector:

[0086] [a,b,c] = Normalized eigenvectors

[0087] First, calculate the sum of each column:

[0088] Column 1: 1 + 1 / 3 + 1 / 6 = 1.5, Column 2: 3 + 1 + 1 / 3 ≈ 4.333, Column 3: 6 + 3 + 1 = 10.

[0089] The normalized matrix is:

[0090]

[0091] Row average (eigenvector):

[0092] a = (0.667 + 0.692 + 0.600) / 3 = 0.653

[0093] b = (0.222 + 0.231 + 0.300) / 3 = 0.251

[0094] c = (0.111 + 0.077 + 0.100) / 3 = 0.096

[0095] The sum = 1.000, and the normalized result is approximately: ≈0.653, ≈0.251, ≈0.096

[0096] Further fine-tuning yields a final, consistent result:

[0097] If the judgment matrix contains... The comparison value was slightly adjusted from 6 to 5.5, and after recalculating, the exact result can be obtained:

[0098] a≈0.636, b≈0.258, c≈0.105

[0099] (3) Consistency check

[0100] Verify whether the judgment matrix is ​​consistent and meets the consistency ratio. It can be considered effective.

[0101] ,

[0102] Where: CI: Consistency Index. : Determine the largest eigenvalue of the matrix, n=3: Compare the number of dimensions. Random consistency index (obtained from a table).

[0103] In this example, the final judgment matrix is:

[0104]

[0105] Calculate the eigenvector:

[0106] We have calculated the weight vector using the normalization method as follows:

[0107]

[0108] Calculate the consistency index (CI):

[0109] First, calculate the product of matrix A and weight vector W:

[0110]

[0111] The numerical calculations are as follows:

[0112] Line 1: 0.636 + 0.774 + 0.5775 = 1.9875, Line 2: 0.212 + 0.258 + 0.315 = 0.785, Line 3: 0.1156 + 0.086 + 0.105 = 0.3066. A·W gives:

[0113]

[0114] Element-by-element division (to obtain an approximate eigenvalue ratio):

[0115]

[0116] Calculate the average value to obtain the largest eigenvalue:

[0117] Calculate the consistency index (CI):

[0118] Calculate the consistency ratio CR

[0119] According to the table, when n=3, the random consistency index RI=0.58.

[0120]

[0121] Conclusion: The consistency requirement is met, because Therefore, the matrix satisfies the consistency condition and is acceptable. Weights a=0.63, b=0.258, and c=0.105 can be used in the objective function, ultimately yielding:

[0122]

[0123] S3: Based on orthogonal experimental data, the neural network inputs are determined to be the tool blunt radius r, tool rake angle γ, and tool clearance angle α, with 3 nodes in the input layer. The neural network output is the objective function value Y of the multi-objective low-stress milling optimization model, with 1 node in the output layer. The training set and test set of the neural network are determined, and the training set and test set are normalized. The range of the number of hidden layer nodes is solved and the number of hidden layer nodes is selected. The neural network is established, and the weights and thresholds of the neural network are initialized. The input and output data of the neural network are shown in Table 3.

[0124] The range of the number of hidden layer nodes is obtained by solving the following formula:

[0125] ;

[0126] Where m is the number of nodes in the input layer, m=3; n is the number of nodes in the output layer, n=1; t is a constant between 1 and 10; and k is the number of nodes in the hidden layer. ;

[0127] The selection of the number of hidden layer nodes is achieved by substituting values ​​within the range of hidden layer node numbers into the neural network, i.e. The average relative error between the simulation results and the orthogonal experimental results corresponding to each number of hidden layer nodes in the test set was obtained by performing three training sessions. The number of hidden layer nodes corresponding to the minimum average relative error was selected as the optimal number of hidden layer nodes.

[0128] The average relative error of neural network training under different numbers of hidden layer nodes in this embodiment is shown in the figure below. Figure 3 As shown.

[0129] Table 3. Initialization weights and thresholds of the neural network; neural network input and output data table.

[0130]

[0131] The average relative error of neural network training under different numbers of hidden layer nodes in this embodiment is shown in the figure below. Figure 3 As shown.

[0132] Depend on Figure 3 It can be seen that the average relative error is the smallest when the number of hidden layer nodes is 9, which is 0.0309. Therefore, 9 is chosen as the optimal number of hidden layer nodes, and the structure of the neural network is 3-9-1.

[0133] This embodiment illustrates the neural network structure optimized by the Pelican optimization algorithm, as shown in the diagram. Figure 4 As shown.

[0134] S3.1: Reference Figure 5 Pelican population initialization: initial population size is 60, maximum number of iterations is 70, and minimum training error is 1×10⁻⁶. -5 Dimension number = number of input layer nodes × number of hidden layer nodes + number of hidden layer nodes × number of output layer nodes + number of hidden layer nodes + number of output layer nodes. In this embodiment, the dimension number is 46. The pelican population initialization is performed using the following formula:

[0135] ;

[0136] in, Let j be the position of the i-th individual pelican. ; , , which represent the upper and lower boundaries of the j-th position, respectively; rand is a random number in [0,1]. These represent the population size and latitude of the pelican, respectively.

[0137] Prey is randomly generated globally, and its fitness function value is calculated. The position of the individual pelicans is updated according to the following formula:

[0138]

[0139] ;

[0140] in, Before the first phase update The j-th dimension position of an individual pelican. This represents the j-th dimension position of the i-th pelican individual after the first phase update; rand is a random number within [0,1]. A random integer that is either 1 or 2; Let be the fitness function value of the i-th candidate solution; Let j be the position of the prey in the j-th dimension; This represents the effective update position of the i-th individual pelican after the first phase update. This represents the position of the i-th individual pelican after the first phase update. This represents the fitness function value of the position point of the i-th individual pelican after the first phase update. This refers to the position of the first individual Pelican before the first phase update. , where m is the dimension of the pelican.

[0141] Calculate the fitness function value of points near the updated pelican individual's location. The position of the individual pelicans is updated again using the following formula:

[0142]

[0143] ;

[0144] in, Let $\mathbf{i}$ be the $j$-th position of the $i$-th pelican individual after the second-stage update, $R$ be a random integer ∈ {0, 2}, $t$ be the current iteration number, and $T$ be the maximum iteration number. This represents the position of the i-th individual pelican after the second phase update. Let i be the fitness function value of the position of the i-th pelican individual after the second phase update. For the second phase update The effective update position of each individual pelican.

[0145] The algorithm iterates by repeatedly updating the position of individual pelicans until the maximum number of iterations or the required computational accuracy is reached. The resulting global optimal solution is the optimal initial weights and thresholds of the neural network. These optimal initial weights and thresholds are then assigned to the neural network to train the neural network optimized by the Pelican Optimization Algorithm (POA-NN).

[0146] The comparison between the objective function value and the actual value of the multi-objective low-stress milling optimization model predicted by the training set of the neural network (POA-NN) optimized by the Pelican Optimization Algorithm in this invention is shown in the figure. Figure 6 As shown.

[0147] The comparison between the objective function value and the actual value of the multi-objective low-stress milling optimization model predicted by the Pelican Optimization Algorithm-optimized neural network (POA-NN) test set in this application is shown in the figure below. Figure 7 As shown.

[0148] In this embodiment, the evolutionary curve of individual fitness optimized by the genetic algorithm is shown in the figure below. Figure 8 As shown.

[0149] In this embodiment, the fitness reached its minimum and stabilized after 11 iterations, with an individual fitness value of -0.1261, which corresponds to the objective function value Y = -0.1261 of the multi-objective low-stress milling optimization model. The corresponding optimal individuals are [0.0109, 6.2869, 12.6797], representing the optimal tool parameters: tool blunt radius. The tool rake angle is 0.0109mm. The tool clearance angle is 6.2869°. It is 12.6797°.

[0150] Example 2: A device for optimizing the parameters of a low-stress milling tool for an aero-engine casing, comprising:

[0151] The processing acquisition module 1 is used to perform milling on the casing using different tools under the same preset processing conditions, and to acquire residual stress parameters on the milled surface. The tool parameters include the tool blunt radius, rake angle and clearance angle, and the surface residual stress parameters include the outermost residual stress, peak tensile stress and peak compressive stress.

[0152] Standardization and Objective Function Module 2 is used to standardize the collected surface residual stress parameters and, based on the standardization results, construct an objective function based on the weighted average of the surface residual stress parameters corresponding to the tool parameters.

[0153] The mapping module 3 is used to establish a multi-objective low-stress milling optimization model with the minimum objective function as the optimization objective, and to establish the mapping relationship between tool parameters and objective function values ​​using a neural network optimized by the Pelican optimization algorithm.

[0154] The parameter optimization module 4 is used to apply a genetic algorithm to optimize the tool parameters based on the mapping relationship and obtain the optimal tool parameters.

[0155] In summary, the present invention discloses a method and apparatus for optimizing the parameters of a low-stress milling tool for a casing. This method provides a more efficient and accurate approach to optimizing the parameters of a low-stress milling tool, which can quantify the relationship between tool parameters and residual stress on the milled surface, quickly and reliably determine the optimal tool parameters, achieve low-stress milling in the manufacturing of the casing, improve the surface quality of the milled surface, and thus achieve fatigue-resistant manufacturing.

[0156] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for optimizing the parameters of a low-stress milling tool for an aero-engine casing, characterized in that, include: The casing is milled using different cutting tools under the same preset machining process conditions. The residual stress parameters of the milled surface are collected. The cutting tool parameters include the tool blunt radius, rake angle and clearance angle. The residual stress parameters of the surface include the outermost residual stress, peak tensile stress and peak compressive stress. By standardizing the collected surface residual stress parameters, and based on the standardization results, an objective function is constructed based on the weighted mean of the surface residual stress parameters corresponding to the tool parameters. With the objective function minimization as the optimization objective, a multi-objective low-stress milling optimization model is established, and a neural network optimized by the Pelican optimization algorithm is used to establish the mapping relationship between tool parameters and objective function values; The genetic algorithm is applied to optimize the tool parameters based on the mapping relationship to obtain the optimal tool parameters.

2. The method for optimizing the parameters of low-stress milling tools for aero-engine casings according to claim 1, characterized in that, The standardization formula is as follows: Where A is the standardized outermost residual stress, B is the standardized peak tensile stress, and C is the standardized peak compressive stress. The original value of the surface residual stress in the milling of the housing by the q-th tool. The original value of the peak tensile stress during the q-th milling operation of the tool on the housing. The original value of the peak compressive stress during the q-th milling operation of the housing by the tool. for maximum value for minimum value for maximum value for minimum value for maximum value for The minimum value.

3. The method for optimizing the parameters of low-stress milling tools for aero-engine casings according to claim 2, characterized in that, The objective function is: Where Y is the objective function value, a is the A weight coefficient determined by the analytic hierarchy process (AHP), b is the B weight coefficient determined by the AHP, and c is the C weight coefficient determined by the AHP, satisfying a+b+c=1 and a>b>c>0.

4. The method for optimizing the parameters of low-stress milling tools for aero-engine casings according to claim 1, characterized in that, The application of the genetic algorithm to optimize tool parameters based on the mapping relationship includes: initializing a tool parameter population, where each individual in the initial tool parameter population is a combination of tool parameters, using the objective function value as the fitness, and updating the initial tool parameter population through selection, crossover, and mutation operations; iterating until a preset number of iterations is reached, and outputting the tool parameters corresponding to the optimal individual.

5. A device for optimizing the parameters of low-stress milling tools for aero-engine casings, characterized in that, include: The machining acquisition module is used to perform milling on the casing using different tools under the same preset machining process conditions, and to acquire residual stress parameters on the milled surface. The tool parameters include the tool blunt radius, rake angle and clearance angle, and the surface residual stress parameters include the outermost residual stress, peak tensile stress and peak compressive stress. The standardization and objective function module is used to standardize the collected surface residual stress parameters and, based on the standardization results, construct an objective function based on the weighted mean of the surface residual stress parameters corresponding to the tool parameters. The mapping module is used to establish a multi-objective low-stress milling optimization model with the objective function as the optimization objective, and to establish the mapping relationship between tool parameters and objective function values ​​using a neural network optimized by the Pelican optimization algorithm. The parameter optimization module is used to apply a genetic algorithm to optimize tool parameters based on the mapping relationship and obtain the optimal tool parameters.

6. The device for optimizing the parameters of low-stress milling tools for aero-engine casings according to claim 5, characterized in that, The standardization process performed by the standardization and objective function module is as follows: Where A is the standardized outermost residual stress, B is the standardized peak tensile stress, and C is the standardized peak compressive stress. The original value of the surface residual stress in the milling of the housing by the q-th tool. The original value of the peak tensile stress during the q-th milling operation of the tool on the housing. The original value of the peak compressive stress during the q-th milling operation of the housing by the tool. for maximum value for minimum value for maximum value for minimum value for maximum value for The minimum value.

7. The device for optimizing the parameters of low-stress milling tools for aero-engine casings according to claim 6, characterized in that, The objective function constructed by the standardization and objective function module is: Where Y is the objective function value, a is the A weight coefficient determined by the analytic hierarchy process (AHP), b is the B weight coefficient determined by the AHP, and c is the C weight coefficient determined by the AHP, satisfying a+b+c=1 and a>b>c>0.

8. The device for optimizing the parameters of low-stress milling tools for aero-engine casings according to claim 7, characterized in that, The genetic algorithm executed by the parameter optimization module includes: initializing a tool parameter population with a size of 30-60; updating the initial tool parameter population with the objective function value as fitness through selection, crossover, and mutation operations, iterating until a preset number of iterations is reached; the selection operation uses roulette wheel selection; the crossover probability of the crossover operation is 0.6-0.8; and the mutation probability of the mutation operation is 0.01-0.05; the preset number of iterations is 50-100; and outputting the optimal tool parameters after the iteration terminates.

Images