CF / PEEK milling process parameter optimization method based on improved lssvr-ooa algorithm

Abstract

The application relates to a CF / PEEK milling process parameter optimization method based on an improved LSSVR-OOA algorithm, for the modeling of a machining process and the acquisition of process parameter combinations, factors such as process parameters and surface roughness need to be considered, and multiple evaluation indexes need to be comprehensively evaluated, therefore, the method adopts a comprehensive method based on LSSVR and an improved fish-eagle optimization algorithm, more parameter combinations that can exceed the predetermined experimental parameter range can be acquired, a weight method is used to establish a multi-objective optimization function, the function is used as an adaptive function of the fish-eagle optimization algorithm, the model is solved, and the optimized milling process parameter combination is acquired, and the method has better applicability and effectiveness.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of production and manufacturing, and particularly relates to a CF / PEEK milling process parameter optimization method based on an improved LSSVR-OOA algorithm. BACKGROUND

[0002] In recent years, carbon fiber reinforced thermoplastic composites (CFRTPs) have great potential in the fields of aerospace, national defense and military, and medical devices due to their high strength-to-weight ratio and corrosion resistance. In recent years, as the demand for low-cost and high-performance materials has become increasingly high, CFRTPs have accounted for an increasingly large proportion of composite material research and application. Carbon fiber reinforced polyether ether ketone (CF / PEEK) composite material is a typical material in CFRTPs, which is a high-grade composite material composed of carbon fiber as a reinforcing phase and polyether ether ketone as a matrix through a special molding process. Compared with traditional metal materials, CF / PEEK composite material exhibits lighter weight, higher strength, higher modulus, high temperature resistance, corrosion resistance, impact and wear resistance, and excellent processing performance. Due to the above characteristics, CF / PEEK material is recognized as a high-quality and high-performance material, which can be applied to various fields including aerospace, wind turbine blades, military and new energy vehicles.

[0003] CF / PEEK has good design capability, and its components are usually manufactured by a near-net-shape method, which still needs secondary processing after primary molding. Milling is the main processing method for CF / PEEK material, which can meet good geometric precision, tolerance level and edge quality. During the processing, due to the anisotropy and non-uniformity of the composite material, CF / PEEK is considered as a difficult-to-machine material, which is prone to typical surface defects. Therefore, effectively controlling the surface quality of CF / PEEK processing has become a widely concerned problem, which has important practical significance.

[0004] In the process parameter optimization problem, most scholars mainly study the target of metal materials, but similar composite materials such as CF / PEEK belong to non-metallic materials, and are difficult-to-machine materials. After milling, the surface quality is poor and the energy consumption is high. The research on such materials is not perfect enough. For modeling, the academia has established models based on mathematical methods or finite element methods, but the research on using machine learning methods is not sufficient enough. For parameter optimization, the original optimization algorithm is easy to fall into local optimum and is not suitable for process parameter optimization scenarios, which needs to be improved. SUMMARY

[0005] The application aims to provide a CF / PEEK milling process parameter optimization method based on an improved LSSVR-OOA algorithm, which can improve the effectiveness of CF / PEEK milling process parameter optimization and has strong applicability.

[0006] To achieve the above-mentioned purpose, the technical scheme adopted by the application is as follows: a CF / PEEK milling process parameter optimization method based on an improved LSSVR-OOA algorithm, comprising the following steps:

[0007] Step S1: taking the three-dimensional surface roughness, three-dimensional fractal dimension and milling energy consumption of the milled carbon fiber reinforced polyether ether ketone material as the optimization target, taking the cutting speed, feed per tooth, cutting depth and fiber direction angle as the optimization variable, constructing an experimental scheme and performing experiments;

[0008] Step S2: according to the experimental results, processing the experimental data and forming a data set, dividing the data set into a training set D t and a test set D v at a set ratio; improving the LSSVR algorithm based on multi-core learning to obtain a model of evaluation indexes;

[0009] Step S3: improving the fish eagle optimization algorithm by using Tent mapping and convex lens reverse learning strategy to increase population diversity and avoid falling into local optimum;

[0010] Step S4: using the improved fish eagle optimization algorithm OOA, taking the test set and root mean square error RMSE as the fitness function of the OOA algorithm, optimizing the hyperparameter combination of LSSVR to obtain the optimal hyperparameter combination, training to obtain the LSSVR model; using the entropy weight method to calculate the information entropy value to allocate the weights of the three-dimensional surface roughness, milling energy consumption and three-dimensional fractal dimension three targets, and constructing a multi-objective comprehensive evaluation function;

[0011] Step S5: taking the multi-objective comprehensive evaluation function for the improved OOA algorithm to calculate the fitness value, and using the improved OOA algorithm to optimize the process parameters to obtain the milling parameters corresponding to the optimal carbon fiber polyether ether ketone milling quality, thereby realizing the optimization of the CF / PEEK milling process parameters.

[0012] Further, the step S1 comprises the following steps:

[0013] S1.1, selecting the cutting speed V c , the cutting depth α e and the feed per tooth f z as experimental factors, setting the levels of each factor, and designing orthogonal experiments; setting the constraints of each factor from the CF / PEEK milling conditions:

[0014] cutting speed V min≤V c ≤V max ;

[0015] Depth of cut α min <α e <α max ;

[0016] Feed per tooth f min ≤f z ≤f max ;

[0017] Fiber direction angle θ1 = 90°, θ2 = 0°;

[0018] In the formula, V min , V max is the minimum and maximum value of the cutting speed V c ; α min , α max is the minimum and maximum value of the depth of cut α e ; f min , f max is the minimum and maximum value of the feed f z ; θ1, θ2 is the fiber direction angle;

[0019] S1.2, milling experiments are carried out, and after each milling experiment, the surface quality of the material after milling, i.e. three-dimensional surface roughness and three-dimensional fractal dimension, is measured to characterize the milling surface quality; and the electric parameter signals obtained by the electric signal collector are collected to calculate the milling energy consumption.

[0020] Further, the step S2 comprises the following steps:

[0021] S2.1, after the experiment is completed, according to the experimental results, the experimental data is processed, and the data set is set as D = {(x i , y i )|i = 1, 2,... n}, x i is the input variable, and y i is the output variable;

[0022] S2.2, the data is preprocessed to remove outliers in the original data;

[0023] S2.3, the data is normalized to make the output value of each node change within a certain range; the normalization equation is:

[0024]

[0025] In the formula, x i represents the original variable; represents the normalized variable; x maxis the maximum value in the original variable; x min is the minimum value in the original variable;

[0026] The transformed data set is obtained as D={(x i ,y i )|i=1,2,...n};

[0027] S2.4, the processed data is divided into training set and test set according to the proportion of 8:2; three-dimensional surface roughness, three-dimensional fractal dimension and energy consumption model are established respectively;

[0028] S2.5, the least square support vector regression LSSVR is used to model each training set data segment to obtain the prediction model of the whole data, and the LSSVR algorithm maps the input space to a high-dimensional feature space through nonlinear mapping , and the optimal linear function is obtained in the feature space; the improved LSSVR model based on multi-kernel learning is constructed, the radial basis kernel function and the polynomial kernel function are fused, the local and global features are cooperatively modeled through dynamic optimization of kernel weight, so as to improve the robustness and adaptability of the model;

[0029] Firstly, the prediction function y(x) of LSSVR is represented as:

[0030]

[0031] In the formula, w represents the weight matrix; represents the nonlinear mapping of the original data to the high-dimensional space; b represents the bias;

[0032] S2.6, according to the principle of structural risk minimization, the problem of mapping the original data to the high-dimensional space is converted into solving a quadratic programming problem:

[0033]

[0034] In the formula, ‖w‖ represents the norm of the weight matrix; γ represents the penalty parameter; e i represents the error between the true value y i and the predicted value y(x), i.e. the bias variable;

[0035] S2.7, the quadratic programming problem is solved by Lagrange multiplier method:

[0036] The Lagrange function is defined as:

[0037]

[0038] In the formula, α i represents the Lagrange multiplier;

[0039] S2.8, w, b and αi ,e i Taking partial derivative and setting it to zero, we have:

[0040]

[0041] The above equation can be rearranged as:

[0042]

[0043] where 1 is a unit column vector with appropriate dimension, K is a kernel matrix, y = (y1, y2, …, y n ) T is a vector composed of target values of training samples, and I is an identity matrix with appropriate dimension;

[0044] S2.9, introduce a multi-kernel learning method, combine local kernel function and global kernel function, so that the model can maintain good prediction ability when facing different types of data;

[0045] Radial basis kernel function is a local kernel function, and its mathematical expression is:

[0046]

[0047] where γ1>0 is a kernel parameter, which controls the width of the kernel function, x i and are two sample vectors in the input data set;

[0048] Polynomial kernel function is a global kernel function, and its mathematical expression is:

[0049]

[0050] where α p >0, c p and d are kernel parameters, representing scaling factor, bias and polynomial order respectively;

[0051] In order to combine the local characteristics of radial basis kernel and the global characteristics of polynomial kernel, a linear combination strategy is adopted to construct a multi-kernel function:

[0052]

[0053] where β1, β2≥0 are weight coefficients, and satisfy β1+β2=1;

[0054] Thus the improvement of LSSVR algorithm is completed, and its prediction function is represented as:

[0055]

[0056] The optimization problem solved by the least squares support vector regression machine algorithm is a convex quadratic programming problem with only one equality constraint, which is suitable for modeling small sample data sets.

[0057] Further, the step S3 comprises the following steps:

[0058] S3.1, Tent chaos mapping strategy is applied to the fish eagle population initialization process to enhance the global search ability, and a convex lens reverse learning strategy is added in the development stage to avoid the algorithm falling into local optimum, so as to complete the improvement of the fish eagle optimization algorithm;

[0059] S3.2, the improved fish eagle optimization algorithm OOA is used for parameter optimization; there are N fish eagle population numbers, and the iteration number is set to T, the optimized parameter constraint is set, the fish eagle population is initialized as {X i ,i=1,2,...,N}, N represents the fish eagle population size, and the individual fish eagle is X i ={x i,1 ,x i,2 ,x i,3 ,....,x i,j}, j represents the dimension of the fish eagle position vector;

[0060] The population is initialized by Tent mapping. First, a series of pseudo-random numbers between [0, 1] are generated by Tent mapping, and then chaos sequence calculation is performed. The Tent mapping formula is as follows:

[0061] Z k =(x z1 ,x z2 ,x z3 ...x zN )

[0062]

[0063] In the formula, Z k+1 represents the chaos sequence of Tent mapping; N represents the number of individuals in the fish eagle population; a z is a set coefficient;

[0064] Map the generated chaos sequence into the search space of the fish eagle optimization algorithm solution:

[0065] x i,j =ld j +(ud j -ld j )×Z k+1

[0066] In the formula, x i,j represents the value of the i-th fish eagle in the j-th dimension; lb jFor the jth dimension of the lower bound search, ub j For the jth dimension of the upper bound search;

[0067] S3.3, the fishing eagle predation process mainly contains two stages, the first stage is that the fishing eagle hunts the fish after detecting its position, for each fishing eagle, the position of other fishing eagles with better fitness value in the search space is regarded as "underwater fish school", if not empty, the center position of "underwater fish school" is calculated; otherwise, the current individual position is reserved; the position update formula is set as:

[0068]

[0069] In the formula, P represents the individual set of the underwater fish school; Fk represents the fitness value of the kth fishing eagle; F e j Xj represents the fitness value of the jth fishing eagle; X k Xk represents the individual position of the kth fishing eagle with better fitness; X best Xk represents the individual position of the kth fishing eagle with better fitness; X

[0070] S3.4, the fishing eagle randomly detects the position of one fish and hunts it, and updates the new position of the fishing eagle, the formula is as follows:

[0071]

[0072]

[0073]

[0074] In the formula, Xk represents the new position of the ith fishing eagle in the exploration stage Xk represents the new position of the ith fishing eagle in the exploration stage i,j Xk represents the new position of the ith fishing eagle in the exploration stage i Xk represents the new position of the ith fishing eagle in the exploration stage i,j Xk represents the new position of the ith fishing eagle in the exploration stage i Xk represents the new position of the ith fishing eagle in the exploration stage Xk represents the new position of the ith fishing eagle in the exploration stage ei Xk represents the new position of the ith fishing eagle in the exploration stage Xk represents the new position of the ith fishing eagle in the exploration stage

[0075] S3.5, After the fish eagle kills a fish, it will take it to a suitable position to eat it; the behavior of the fish eagle makes the position of the fish eagle change slightly, thereby enhancing the local development ability of the algorithm; a new random position is calculated as a "suitable position to eat fish" using the following formula, and if the fitness value of the new position can be improved, the previous position of the corresponding fish eagle is replaced;

[0076]

[0077]

[0078] where r is a random number in the interval [0,1]; t is the current iteration number; T represents the maximum number of iterations required; is the new position of the i-th fish eagle in the exploration phase in the j-th dimension; X i represents the individual position of the i-th fish eagle, represents the fish eagle individual in the exploration phase; is the fitness value of the i-th fish eagle in the development phase;

[0079] S3.6, In the development phase of the fish eagle optimization algorithm, the reverse learning strategy based on the principle of convex lens imaging is used to improve the fish eagle optimization algorithm, which is an improved strategy to expand the search range by calculating the reverse solution of the current position; the reverse solution of the convex lens imaging reverse learning strategy is calculated by the following formula:

[0080]

[0081] where represents the reverse solution generated by the principle of convex lens imaging; k represents the ratio of the entity height to the imaging height, i.e. the scaling factor;

[0082] The formula for setting the adjustment scaling factor k is:

[0083] n = (1 + (t / T) 1 / 2 ) 10

[0084] The reverse solution generated by a smaller k has a larger range, while a larger k produces a reverse solution within a small range; as the number of iterations increases, the value of k will increase, and the range of its reverse solution will decrease;

[0085] S3.7, Considering that the reverse solution is not necessarily better than the original solution, a greedy mechanism is introduced to select the individual after reverse learning to obtain the best individual, and the mathematical model formula of the greedy mechanism is as follows:

[0086]

[0087] where X new is the selected individual;

[0088] S3.8. Iterate and output the improved osprey optimization algorithm. When the number of iterations is less than the maximum number of iterations T, iterate according to steps S3.2 to S3.7. When the iteration termination condition is met, output the osprey individual with the best fitness and its fitness value, and obtain the value in the position coordinates of the osprey individual.

[0089] Furthermore, step S4 includes the following steps:

[0090] S4.1. The improved Osprey optimization algorithm is used to optimize the LSSVR hyperparameters to ensure that reasonable hyperparameters are selected to improve the performance of the LSSVR algorithm. Model 1, Model 2 and Model 3 are respectively trained and fitted using the milling process parameter combination as input and the evaluation indicators of three-dimensional surface roughness, milling energy consumption and three-dimensional fractal dimension as output. The data set is divided into training set D t1 、D t2 and D t3 With the test set D v1 、D v2 and D v3 ;

[0091] S4.2. Complete the setting of osprey population N1, N2, number of iterations T1, T2, and least squares support vector regression hyperparameter constraints. The individual osprey position vector is determined by the penalty coefficient γ, the kernel parameter γ1 of the radial basis kernel function, and the scaling factor α of the polynomial kernel function. p Composed of test set D v1 、D v2 The root mean square error RMSE is used as the fitness function of the improved Osprey algorithm. The root mean square error RMSE calculation formula is:

[0092]

[0093] Where y i is the measured data; y i (x) is the prediction result of the least squares support vector regression algorithm; N is the number of prediction samples;

[0094] S4.3. Based on the settings of steps S4.1 and S4.2, solve the hyperparameters of LSSVR through steps S3.2 to S3.8; then use the hyperparameter combination optimized by the improved Osprey algorithm to train LSSVR models F1, F2 and F3 through steps S2.2 to S2.9.

[0095] Furthermore, step S5 includes the following steps:

[0096] S5.1, the model weight ratio calculation, the introduction of entropy weight method, using the method to establish a multi-objective comprehensive evaluation function; for multi-objective optimization, first, the optimization index value is preprocessed; for the construction of comprehensive evaluation function, three kinds of indexes are set to the same attribute, so the three indexes are all converted into cost type indexes, as follows:

[0097]

[0098]

[0099]

[0100] In the formula, Sa min and Sa max are the minimum and maximum values of three-dimensional surface roughness; P min and P max are the minimum and maximum values of milling energy consumption; D min and D max are the minimum and maximum values of three-dimensional fractal dimension; x represents the optimization variable, i.e. the milling process parameters; and respectively represent the numerical value of the three-dimensional surface roughness, energy consumption and three-dimensional fractal dimension of CF / PEEK milling under the current milling process parameters; f1(x), f2(x) and f3(x) represent the preprocessed values respectively;

[0101] S5.2, the preprocessed values are substituted into the matrix to calculate the prior probability, i.e. the proportion of each set of milling process parameters:

[0102]

[0103] In the formula, r ij represents the value of the preprocessed index, and m represents the number of values; R represents the original data matrix formed after standardization, R=(r ij ) m×n ;

[0104] The proportion of the index value of the i-th item under the j-th index, i.e. the prior probability, is calculated:

[0105]

[0106] In the formula, p ij represents the proportion of the index value, and represents the proportion of the number of samples with value i in the total number of samples in index j;

[0107] Thus, the proportion matrix P={p ij} m×n of the data is established;

[0108] S5.3, the calculation formula of entropy weight is:

[0109]

[0110] wherein, represents the entropy weight, k = 1 / ln(n) > 0, to meet

[0111] S5.4, calculate the information utility value based on the entropy weight:

[0112]

[0113] wherein, d j represents the information utility value of the jth index;

[0114] S5.5, calculate the size of the weight coefficient of each target through the information utility value calculation;

[0115]

[0116] wherein, w * j represents the weight coefficient size of the jth target calculated by the entropy weight method;

[0117] S5.6, weight each target, and establish a comprehensive evaluation function as shown in the following formula:

[0118]

[0119] wherein, F e represents the comprehensive evaluation function established based on the entropy weight method; and respectively represent the weight coefficient of the three-dimensional surface roughness of CF / PEEK milling, the milling energy consumption and the three-dimensional fractal dimension;

[0120] S5.7, after weighting the three models, a multi-objective comprehensive evaluation function is obtained, which is used as the fitness function of the improved fish eagle optimization algorithm:

[0121] minf(x) = F e [f1, f2, f3]

[0122] wherein, f1, f2, f3 respectively represent the three-dimensional surface roughness, the milling energy consumption and the three-dimensional fractal dimension;

[0123] S5.8, the improved OOA algorithm is used to optimize the three-dimensional surface roughness, the three-dimensional fractal dimension and the milling energy consumption of CF / PEEK after milling, and the milling process parameters corresponding to the individual solution at the optimal milling quality are obtained, so as to realize the optimization of the milling process parameters; the size of the first dimension is the cutting speed V c , and the second dimension is the cutting depth αe a size of the third dimension is a feed per tooth f z a size of the fourth dimension is a value of the fiber direction angle θ

[0124]

[0125] In the formula, x i,1 , x i,2 , x i,3 , x i,4 respectively represent the values of the spindle speed, the cutting depth, the feed per tooth and the fiber direction angle of the i-th fish eagle; x i,1min , x i,1max respectively represent the minimum and maximum values of the spindle speed; x i,2min , x i,2max respectively represent the minimum and maximum values of the cutting depth; x i,3min , x i,3max respectively represent the minimum and maximum values of the feed per tooth; u1 and v1 are constants.

[0126] S5.9, iteratively processing according to steps S3.2 to S3.8; when the iteration termination is met, outputting the fish eagle individual with the best fitness and the fitness value thereof, and obtaining the optimal solution of the multi-objective optimization; in the position coordinates of the fish eagle individual, each dimension is respectively a cutting speed V c , a cutting depth α e , a feed per tooth f z and a fiber direction angle, and obtaining the optimal process parameter combination of the CF / PEEK milling multi-objective optimization.

[0127] Compared with the prior art, the CF / PEEK milling process parameter optimization method based on the improved LSSVR-OOA algorithm has the following beneficial effects: for the modeling of the machining process and the acquisition of the process parameter combination, not only the factors such as the process parameters and the surface roughness need to be considered, but also the comprehensive evaluation of multiple evaluation indexes needs to be considered, therefore, the method adopts the comprehensive method based on the LSSVR and the improved fish eagle optimization algorithm, can acquire more parameter combinations which can be optimized beyond the predetermined experimental parameter range, and uses the weight method to establish a multi-objective optimization function as an adaptive function of the fish eagle optimization algorithm, solves the model to acquire the optimized milling process parameter combination, and the method has better applicability and effectiveness. BRIEF DESCRIPTION OF DRAWINGS

[0128] Figure 1 is a method implementation flowchart of the embodiment of the application. DETAILED DESCRIPTION

[0129] The application will be further described below in combination with the drawings and embodiments.

[0130] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present application belongs.

[0131] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0132] like Figure 1 As shown, this embodiment provides a CF / PEEK milling process parameter optimization method based on the improved LSSVR-OOA algorithm, and its specific implementation steps are as follows.

[0133] Step S1: Taking the three-dimensional surface roughness, three-dimensional fractal dimension, and milling energy consumption of milling carbon fiber reinforced polyetheretherketone material as optimization targets, and cutting speed, feed per tooth, cutting depth, and fiber orientation angle as optimization variables, an experimental plan is constructed and experiments are conducted.

[0134] Step S2: According to the experimental results, the experimental data is processed and a data set is formed. The data set is divided into a training set D according to the set ratio (8:2) t and the test set D v ; The LSSVR algorithm is improved based on multi-kernel learning to obtain a model of evaluation indicators.

[0135] Step S3: Tent mapping and convex lens reverse learning strategies are used to improve the Osprey optimization algorithm to increase population diversity and avoid falling into local optimality.

[0136] Step S4: Using the improved Osprey optimization algorithm (OOA), with the test set and root mean square error (RMSE) as the OOA algorithm's fitness function, the optimal hyperparameter combination for LSSVR is optimized. The optimal hyperparameter combination is obtained and the LSSVR model is trained. The information entropy value is calculated using the entropy weight method to assign weights to the three objectives of 3D surface roughness, milling energy consumption, and 3D fractal dimension, thus constructing a multi-objective comprehensive evaluation function.

[0137] Step S5: A multi-objective comprehensive evaluation function is used to improve the OOA algorithm to calculate the fitness value, and the improved OOA algorithm is used to optimize the process parameters to obtain the milling parameters corresponding to the optimal carbon fiber polyetheretherketone milling quality, thereby realizing the optimization of CF / PEEK milling process parameters.

[0138] In the embodiment, the step S1 comprises the following steps:

[0139] S1.1, selecting a cutting speed V c , a cutting depth a e and a feed per tooth f z as experimental factors, setting levels of each factor and designing an orthogonal experiment. The CF / PEEK milling conditions are set as constraints of each factor.

[0140] The cutting speed V min ≤V c ≤V max ;

[0141] The cutting depth a min <a e <a max ;

[0142] The feed per tooth f min ≤f z ≤f max ;

[0143] The fiber direction angles are θ1=90° and θ2=0°.

[0144] In the formula, V min , V max are the minimum and maximum values of the cutting speed V c ; a min , a max are the minimum and maximum values of the cutting depth a e ; f min , f max are the minimum and maximum values of the feed f z ; and θ1 and θ2 are the fiber direction angles.

[0145] S1.2, performing a milling experiment, and after each milling experiment, measuring the surface quality of the material after milling, i.e. the three-dimensional surface roughness and the three-dimensional fractal dimension, to characterize the milling surface quality. The electric parameter signals obtained by the electric signal collector are collected to calculate the milling energy consumption.

[0146] In the embodiment, the step S2 comprises the following steps:

[0147] S2.1, after the experiment is completed, processing the experimental data according to the experimental results, setting the data set as D={(x i ,y i )|i=1,2,...n}, x i as the input variable and y i as the output variable.

[0148] S2.2, the collected data is generally expressed in quantities, due to the interference of machine tool processing, there are abnormal values and noises in the data, and the data needs to be preprocessed to remove the abnormal values in the extremely individual original data.

[0149] S2.3, the scales of various variables may differ greatly, and such difference will greatly affect the accuracy and stability of numerical calculation. Therefore, before training and testing with sample data, the data needs to be normalized to make the output values of each node change within a certain range. The normalization equation is:

[0150]

[0151] In the formula, x i represents the original variable; x max represents the normalized variable; x min is the maximum value of the original variable; x i is the minimum value of the original variable.

[0152] The transformed data set is D={(x i ,y i )|i=1,2,...n}.

[0153] S2.4, the processed data is divided into training set and test set according to the ratio of 8:2 to prepare for subsequent modeling. Three-dimensional surface roughness, three-dimensional fractal dimension and energy consumption model are established respectively.

[0154] S2.5, the least square support vector regression (LSSVR) is used to model each training set data segment to obtain the prediction model of the overall data. The LSSVR algorithm maps the input space to a high-dimensional feature space through nonlinear mapping , and finds the optimal linear function in the feature space. The inner product of nonlinear mapping is used instead of direct calculation of nonlinear mapping, which can obviously simplify the calculation. The traditional LSSVR model may face the problem of single kernel function selection, thereby limiting the expression ability of the model. The improved LSSVR model based on multi-kernel learning is proposed, which combines radial basis kernel function and polynomial kernel function, and realizes the collaborative modeling of local and global features through dynamic optimization of kernel weight, so as to improve the robustness and adaptability of the model.

[0155] Firstly, the prediction function y(x) of LSSVR can be expressed as:

[0156]

[0157] In the formula, w represents the weight matrix; represents the nonlinear mapping of the original data to the high-dimensional space; b represents the bias.

[0158] S2.6, according to the principle of structural risk minimization, the problem of mapping the original data to a high-dimensional space is converted into solving a quadratic programming problem:

[0159]

[0160] In the formula, ‖w‖ represents the norm of the weight matrix; γ represents the penalty parameter; e i represents the error between the true value y i and the predicted value y(x), that is, the bias variable.

[0161] S2.7, the quadratic programming problem is solved by the Lagrange multiplier method:

[0162] The Lagrange function is defined as:

[0163]

[0164] In the formula, α i represents the Lagrange multiplier.

[0165] S2.8, partial derivatives of w, b, α i , e i are solved and set to 0:

[0166]

[0167] The above formula is rearranged as:

[0168]

[0169] In the formula, 1 represents a unit column vector with appropriate dimensions, the matrix K is the kernel matrix, y=(y1, y2, …, y n ) T is the target value vector of the training sample, and I is an appropriate unit matrix.

[0170] S2.9, on this basis, a multi-kernel learning method is introduced, which can effectively improve the generalization performance of LSSVR, and combines local kernel functions and global kernel functions, so that the model can maintain good prediction ability when facing different types of data.

[0171] The radial basis kernel function is a local kernel function, which is suitable for problems with complex data distribution and significant local characteristics, and its mathematical expression is:

[0172]

[0173] In the formula, γ1>0 is the kernel parameter, which controls the width of the kernel function, x i and are two sample vectors in the input data set.

[0174] The polynomial kernel function is a global kernel function which can effectively capture the global nonlinear distribution relationship, and its mathematical expression is as follows:

[0175]

[0176] In the formula, α p > 0, c p and d are kernel parameters, respectively representing a scaling factor, a bias and an order of the polynomial.

[0177] In order to combine the local characteristics of the radial basis kernel and the global characteristics of the polynomial kernel, a linear combination strategy is adopted to construct a multi-kernel function:

[0178]

[0179] In the formula, β1, β2≥ 0, as weight coefficients, and satisfy β1+β2=1. It is proposed that β1=β2=0.5 to model.

[0180] The improved LSSVR algorithm is completed by the above method, and its prediction function can be expressed as:

[0181]

[0182] The optimization problem solved by the least squares support vector regression machine algorithm is a convex quadratic programming problem containing only one equality constraint, which is suitable for modeling of small sample data sets.

[0183] In the embodiment, the step S3 comprises the following steps:

[0184] S3.1, the Tent chaotic mapping strategy is applied to the fish eagle population initialization process to enhance the global search ability, and the convex lens reverse learning strategy is added in the development stage to avoid the algorithm falling into local optimum, so as to complete the improvement of the fish eagle optimization algorithm.

[0185] S3.2, the improved fish eagle optimization algorithm (OOA) is used for parameter optimization. There are N fish eagle population numbers, and the iteration number is set to T, the constraint of the optimized parameters is set, the fish eagle population is initialized as {X i ,i=1,2,...,N}, N represents the fish eagle population size, and the individual fish eagle is X i ={x i,1 ,x i,2 ,x i,3 ,....,x i,j}, j represents the dimension of the fish eagle position vector.

[0186] In order to make the initial osprey population evenly distributed in the search space, considering the good ergodicity of the Tent mapping, it can effectively avoid the population from being concentrated in a specific area and enhance the global search capability. The population is initialized using the Tent mapping. First, a series of pseudo-random numbers between the interval [0, 1] are generated using the Tent mapping, and then the chaotic sequence is calculated. The Tent mapping formula is as follows:

[0187] Z k =(x z1 ,x z2 ,x z3 ...x zN )

[0188]

[0189] Where Z k+1 represents the chaotic sequence of Tent mapping; N represents the number of individuals in the osprey population; a z is the setting coefficient.

[0190] According to experience, a is generally taken z is 0.4, and the generated chaotic sequence is mapped into the search space of the Osprey optimization algorithm solution:

[0191] x i,j =ld j +(ud j -ld j )×Z k+1

[0192] Where x i,j represents the value of the i-th osprey in the j-th dimension; lb j is the jth dimension to search for the lower bound, ub j is the j-th dimension of the search upper bound.

[0193] S3.3. The osprey hunting process mainly consists of two stages. The first stage is that the osprey hunts the fish after detecting its location. For each osprey, the location of other ospreys with better fitness values ​​in the search space is regarded as the "underwater fish school". If it is not empty, the center position of the "underwater fish school" is calculated; otherwise, the current individual position is retained. The position update formula is set as:

[0194]

[0195] Where P represents the individual set of underwater fish; represents the fitness value of the kth osprey; represents the fitness value of the jth osprey; X k represents the position of the kth osprey with better fitness; X best It is the location of the best osprey individuals.

[0196] S3.4, the fish eagle randomly detects the position of one fish and hunts it, and the new position of the fish eagle is updated, as follows:

[0197]

[0198]

[0199]

[0200] wherein, is the new position of the ith fish eagle in the exploration phase is the value in the jth dimension; X i,j represents a random number between the interval [0, 1]; SP i is the fish randomly selected by the ith fish eagle from its fish group P, and is a random number between [0, 1]; I i,j is any number in the set {1, 2}; X i represents the individual position of the ith fish eagle, represents the fish eagle individual in the exploration phase; F ei represents the individual fitness value of the ith fish eagle, is the fitness value of the ith fish eagle in the exploration phase.

[0201] S3.5, after the fish eagle hunts a fish, it will take it to a suitable position to eat it. This behavior of the fish eagle causes a slight change in the position of the fish eagle, thereby enhancing the local development ability of the algorithm. A new random position is calculated as the "suitable fish-eating position" using the following formula, and if the fitness value of the new position can be improved, the previous position of the corresponding fish eagle is replaced.

[0202]

[0203]

[0204]

[0205] wherein r is a random number between the interval [0, 1]; t is the current iteration number; T represents the maximum number of iterations required; is the new position of the ith fish eagle in the exploration phase is the value in the jth dimension; X i represents the individual position of the ith fish eagle, represents the fish eagle individual in the exploration phase; is the fitness value of the ith fish eagle in the development phase.

[0206] S3.6, in the fish eagle optimization algorithm development stage, the reverse learning strategy of convex lens imaging principle is used to improve the fish eagle optimization algorithm, which can effectively solve the problem that only using the reverse learning strategy alone cannot make the individual jump out of the local optimum. It is an improved strategy to expand the search range by calculating the reverse solution of the current position. The reverse solution of the convex lens imaging reverse learning strategy can be calculated by the following formula:

[0207]

[0208] In the formula, represents the reverse solution generated by the convex lens imaging principle; k represents the ratio of the entity height to the imaging height, that is, the scaling factor.

[0209] The formula for setting the adjustment scaling factor k value is:

[0210] n = (1 + (t / T) 1 / 2 ) 10

[0211] The reverse solution generated by a smaller k has a larger range, while a larger k produces a smaller range of reverse solutions. As the number of iterations increases, the value of k will increase, and the range of its reverse solution will decrease.

[0212] S3.7, considering that the reverse solution is not necessarily better than the original solution, a greedy mechanism is introduced here to select the individual after reverse learning to obtain the best individual. The mathematical model formula of the greedy mechanism is as follows:

[0213]

[0214] In the formula, X new is the selected individual.

[0215] This mechanism can effectively guide the fish eagle individual to the global optimal point, and thus achieve the goal of reverse solution.

[0216] S3.8, the improved fish eagle optimization algorithm is iterated and output. When the number of iterations is less than the maximum number of iterations T, the iteration process is performed according to steps S3.2 to S3.7; when the iteration termination condition is met, the fish eagle individual with the best fitness and its fitness value are output, and the value of the position coordinates of the fish eagle individual is obtained.

[0217] In this embodiment, the step S4 includes the following steps:

[0218] S4.1, the improved fish eagle optimization algorithm is used to optimize the LSSVR super parameter, and reasonable super parameter is selected to improve the performance of LSSVR algorithm. Model 1, model 2 and model 3 are respectively trained and fitted models with milling process parameter combination as input, and three-dimensional surface roughness, milling energy consumption and three-dimensional fractal dimension as output. The data set is divided into training set D t1 、 D t2 、 D t3 、 D v1 、 D v2 、 D v3 .

[0219] S4.2, the fish eagle population N1, N2, iteration number T1, T2 and least square support vector regression super parameter constraint are set, the fish eagle individual position vector is composed of penalty coefficient γ, kernel parameter γ1 of radial basis kernel function and scaling factor α p 、 D v1 、 D v2 , and root mean square error RMSE are used as the fitness function of improved fish eagle algorithm, and the root mean square error RMSE calculation formula is as follows:

[0220]

[0221] In the formula, y i is the measured data; y i (x) is the prediction result of least square support vector regression algorithm; N is the number of prediction samples;

[0222] S4.3, based on the setting of steps S4.1 and S4.2, the super parameters of LSSVR are solved through steps S3.2 to S3.8; and then the improved fish eagle algorithm is used to optimize the super parameter combination, and LSSVR models F1, model F2 and model F3 are trained through steps S2.2 to S2.9.

[0223] In the embodiment, the step S5 includes the following steps:

[0224] S5.1, the model weight proportion is calculated, the entropy weight method is introduced, and the entropy is a measure of the degree of confusion. The method is used to establish a multi-objective comprehensive evaluation function. For multi-objective optimization, the optimization index value needs to be preprocessed first. Three-dimensional surface roughness and milling energy consumption in the index are cost type indexes, and the smaller the value is, the better; while for three-dimensional fractal dimension, it is an economic index, and the larger the value is, the better. In order to construct the comprehensive evaluation function, the three indexes need to be set to the same attribute, so the three indexes are all converted into cost type indexes, which are shown as follows.

[0225]

[0226]

[0227]

[0228] where Sa min and Sa max are the minimum and maximum of three-dimensional surface roughness; P min and P max are the minimum and maximum of milling energy consumption; D min and D max are the minimum and maximum of three-dimensional fractal dimension; x represents the optimization variable, i.e. the milling process parameters; and represent the values of three-dimensional surface roughness, energy consumption and three-dimensional fractal dimension of CF / PEEK milling under the current milling process parameters; f1(x), f2(x) and f3(x) represent the pre-processed values, respectively.

[0229] S5.2, the pre-processed values are substituted into the matrix to calculate the prior probability, i.e. the proportion of each set of milling process parameters:

[0230]

[0231] where r ij represents the value of the pre-processed index, and m represents the number of values; R represents the original data matrix formed after standardization, R=(r ij ) m×n .

[0232] The proportion of the index value of the i-th item under the j-th index (prior probability) is calculated:

[0233]

[0234] where p ij represents the proportion of the index value (prior probability), and represents the proportion of the number of samples with value i in index j to the total number of samples.

[0235] Thus, the proportion matrix P={p ij} m×n .

[0236] S5.3, the calculation formula of entropy weight is:

[0237]

[0238] where w represents the entropy weight, and k=1 / ln(n)>0, to satisfy

[0239] S5.4, the information utility value is calculated based on the entropy weight:

[0240]

[0241] where d j represents the information utility value of the jth index.

[0242] The information utility value of the index depends on the information entropy e j of the index, and the difference between 1 and e j , the value of which directly affects the size of the weight, the greater the information utility value, the greater the importance of the evaluation, and the greater the weight.

[0243] S5.5, the size of the weight coefficient of each target is calculated by the information utility value calculation.

[0244]

[0245] where w * j represents the weight coefficient size of the jth target calculated by the entropy weight method.

[0246] S5.6, the weight of each target is assigned, and the comprehensive evaluation function is established as shown in the following formula:

[0247]

[0248] where F e represents the comprehensive evaluation function established based on the entropy weight method. and respectively represent the weight coefficients of the three-dimensional surface roughness of CF / PEEK milling, milling energy consumption and three-dimensional fractal dimension.

[0249] S5.7, after weighting the three models, a multi-objective comprehensive evaluation function is obtained, which is used as the fitness function of the improved fish eagle optimization algorithm:

[0250] minf(x)=F e [f1,f2,f3]

[0251] where f1, f2, f3 respectively represent three-dimensional surface roughness, milling energy consumption, and three-dimensional fractal dimension.

[0252] S5.8, the improved OOA algorithm is used to optimize the three-dimensional surface roughness, three-dimensional fractal dimension and milling energy consumption of CF / PEEK after milling, and the milling process parameters corresponding to the individual solution at the optimal milling quality are obtained, so as to realize the optimization of the milling process parameters. Let the first dimension be the cutting speed V c , the second dimension be the cutting depth α e , the third dimension be the feed per tooth f z , and the fourth dimension be the value of the fiber direction angle θ.

[0253]

[0254] wherein x i,1 , x i,2 , x i,3 , x i,4 respectively represent the i-th fish-eagle's spindle speed, cutting depth, feed per tooth and fiber direction angle, the numerical value of the four index dimensions; x i,1min , x i,1max respectively represent the minimum and maximum spindle speed; x i,2min , x i,2max respectively represent the minimum and maximum cutting depth; x i,3min , x i,3max respectively represent the minimum and maximum feed per tooth; u1, v1 are constants.

[0255] S5.9, iterate according to steps S3.2 to S3.8; when the iteration termination is met, output the fish-eagle individual with the best fitness and its fitness value, and obtain the optimal solution of multi-objective optimization; in the position coordinates of the fish-eagle individual, each dimension is cutting speed V c , cutting depth a e , feed per tooth f z and fiber direction angle, and obtain the optimal process parameter combination of CF / PEEK milling multi-objective optimization.

[0256] In this embodiment, a VMC vertical machining center of Shenyang Machine Tool Factory is selected for machining, the maximum speed of which is 10000 rpm, and the positioning accuracy of XYZ axis is 0.01 mm. A 10 mm single blade hard alloy cutter is selected as the milling cutter, the material is fixed fiber orientation, and the size of the CF / PEEK laminate is 100 mm*100 mm. In the milling CF / PEEK material experiment, the key process parameters include cutting speed, cutting depth, feed per tooth and fiber direction angle, and the orthogonal test method is adopted to design the experiment based on the determined factors and levels, and the mixed level orthogonal test is designed, a total of 16 groups, as shown in Tables 1 and 2.

[0257] Table 1 factor and level setting

[0258]

[0259] Table 2 mixed orthogonal experiment planning

[0260]

[0261]

[0262] According to the experimental arrangement in Table 2, the CF / PEEK laminates with a size of 100mm*100mm are milled. After the experiment, the three-dimensional surface roughness, milling energy consumption and three-dimensional fractal dimension of the milled mass evaluation index need to be measured and calculated. First, the three-dimensional surface roughness is calculated using a three-dimensional surface profile measuring instrument; the electric signal in the milling process is collected using an intelligent collector, and the energy consumption required in the processing process is recorded; the three-dimensional fractal dimension is calculated by observing and measuring the plane after the test using a scanning electron microscope. The data acquisition method is: based on a group of process parameters after milling, the three-dimensional surface roughness is measured, the three-dimensional fractal dimension is calculated, and the average value is obtained. Table 3 shows the data of the titanium alloy milling experiment.

[0263] Table 3 CF / PEEK milling experiment data results

[0264]

[0265] (1) Optimize the least squares support vector regression model hyperparameters

[0266] Before training the least squares support vector regression model, the penalty coefficient γ, the kernel parameter γ1 of the radial basis kernel function, and the scaling factor α of the polynomial kernel function p hyperparameters are optimized, and the bias c of the polynomial kernel function p = 1 and the order d = 2 of the polynomial are set as default values and do not participate in optimization. Due to the instability of manual parameter adjustment, the adjustment effect is too dependent on the experience of the adjustment personnel, so the LSSVR hyperparameters of the present application are adjusted using the fish eagle algorithm, and the evaluation index model is trained using the optimized hyperparameters. The fish eagle individual position vector is set to consist of hyperparameters, the fish eagle population size N1, N2, N3 = 100, the iteration number T1, T2, T3 = 50, and the hyperparameter constraint is set to the formula:

[0267]

[0268] After optimization by the OOA algorithm, the optimal hyperparameter combination of the three-dimensional surface roughness, milling energy consumption and three-dimensional fractal dimension fitting model is obtained, and the LSSVR1 model, LSSVR2 model and LSSVR3 model are obtained.

[0269] (2) Multi-objective comprehensive evaluation function establishment

[0270] The weight coefficients of the three-dimensional surface roughness, milling energy consumption and three-dimensional fractal dimension of each target are calculated by steps S5.1 to S5.6. The weight coefficients are 0.4716, 0.2812 and 0.2472, respectively. The multi-objective comprehensive evaluation function formula is constructed by combining the established fitting model:

[0271] F e(f1,f2,f3) = 0.4716f1(x) + 0.2812f2(x) + 0.2472f3(x)

[0272] (3) Improved fish eagle optimization algorithm parameter setting

[0273] Using the constructed multi-objective comprehensive evaluation function as the fitness function of the improved fish eagle optimization algorithm, set the fish eagle population size N x = 100 iteration times T X = 50 =, fish eagle individual position X i = {x i,1 ,x i,2 ,x i,3 ,x i,4}, the fitness function is set as: minf(x) = F e [f1,f2,f3], the milling process parameter constraint is:

[0274]

[0275] Iterate to get the optimal fish eagle individual solution vector.

[0276] (4) Unimproved fish eagle algorithm optimization results

[0277] Table 4 shows the process parameter combination obtained by the unimproved LSSVR-OOA fish eagle optimization algorithm, after the iteration is terminated, the CF / PEEK milling process parameter combination is shown in Table 4,

[0278] Table 4 Milling process parameter multi-objective optimization results after unimproved LSSVR-OOA algorithm optimization

[0279]

[0280] Based on this parameter combination, the three-dimensional surface roughness, milling energy consumption and three-dimensional fractal dimension experimental results obtained by CF / PEEK milling experiment are shown in Table 5.

[0281] Table 5 Milling experiment results of process parameters obtained by unimproved LSSVR-OOA algorithm optimization

[0282]

[0283] (5) Improved fish eagle algorithm optimization results

[0284] Table 6 shows the process parameter combination obtained by the improved LSSVR-OOA fish eagle optimization algorithm, as shown in Table 6

[0285] Table 6 Milling process parameter optimization results

[0286]

[0287] The experimental results of three-dimensional surface roughness, milling energy consumption and three-dimensional fractal dimension obtained from the CF / PEEK milling experiments based on the parameter combination are shown in Table 7.

[0288] Table 7 Milling experimental results of the improved LSSVR-OOA algorithm optimized process parameters

[0289]

[0290] After comparison, the experimental results of the improved LSSVR-OOA algorithm optimized milling process parameters, three-dimensional surface roughness, milling energy consumption and three-dimensional fractal dimension are all significantly improved.

[0291] Those skilled in the art will understand that the embodiments of the present application can be provided as methods, systems or computer program products. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage etc.) containing computer-usable program code.

[0292] The present application is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device that implements the flow Figure 1 The functions specified in a flow or multiple flows and / or blocks Figure 1 The means for performing the functions specified in a flow or multiple flows and / or blocks.

[0293] These computer program instructions can also be stored in a computer-readable memory that can direct the computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including instruction means, which implements the flow Figure 1 The functions specified in a flow or multiple flows and / or blocks Figure 1 The means for performing the functions specified in a flow or multiple flows and / or blocks.

[0294] These computer program instructions can also be loaded into a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flowcharts and / or blocks Figure 1 one or more flowcharts and / or blocks

[0295] The above descriptions are only the preferred embodiments of the present application, not intended to limit the present application to other forms described. Any person skilled in the art may make changes or modifications to the above-described technical contents as equivalent embodiments without departing from the technical solutions of the present application. Any simple modification, equivalent change and modification of the above embodiments according to the technical essence of the present application without departing from the technical solutions of the present application still belongs to the protection scope of the technical solutions of the present application.

Claims

1. A CF / PEEK milling process parameter optimization method based on an improved LSSVR-OOA algorithm, characterized in that, The method comprises the following steps: Step S1: constructing an experimental scheme and performing experiments by taking the three-dimensional surface roughness, three-dimensional fractal dimension and milling energy consumption of the carbon fiber reinforced polyether ether ketone material as the optimization objectives, and taking the cutting speed, feed per tooth, cutting depth and fiber direction angle as the optimization variables; Step S2: according to the experimental results, processing the experimental data and forming a data set, dividing the data set into a training set D t and a test set D v at a set ratio; improving the LSSVR algorithm based on multi-core learning to obtain a model for evaluating indicators; Step S3: improving the fish eagle optimization algorithm by adopting the Tent mapping and convex lens reverse learning strategy, and increasing the population diversity to avoid falling into local optimization; Step S4: obtaining the optimal hyperparameter combination by optimizing the hyperparameter combination of the LSSVR by adopting the improved fish eagle optimization algorithm OOA, taking the test set and the root mean square error RMSE as the fitness function of the OOA algorithm, training the LSSVR model, calculating the information entropy value by adopting the entropy weight method, and distributing the weights of the three-dimensional surface roughness, milling energy consumption and three-dimensional fractal dimension to construct a multi-objective comprehensive evaluation function; Step S5: taking the multi-objective comprehensive evaluation function for the improved OOA algorithm to calculate the fitness value, and optimizing the process parameters by adopting the improved OOA algorithm to obtain the milling parameters corresponding to the optimal carbon fiber polyether ether ketone milling quality, so as to realize the optimization of the CF / PEEK milling process parameters; The step S1 comprises the following steps: S1.1, selecting a cutting speed V c , a cutting depth a e , and a feed per tooth f z The experimental factors were set at various levels, and an orthogonal experiment was designed. The constraints of the CF / PEEK milling conditions were set for each factor: Cutting speed V min ≤ V c ≤ V max ; Depth of cut a min < a e < a max ; feed per tooth f min ≤ f z ≤ f max ; The fiber direction angle θ1 is 90°, and θ2 is 0°; where V min , V max are minimum and maximum values of the cutting speed V c ; α min , α max are minimum and maximum values of the cutting depth α e ; f min , f max are minimum and maximum values of the feed f z ; θ1, θ2 are fiber direction angles; S1.2, performing milling experiments, measuring the surface quality of the material after each milling experiment, i.e. the three-dimensional surface roughness and the three-dimensional fractal dimension, to characterize the milling surface quality; and collecting the obtained electric parameter signals by adopting an electric signal collector, and calculating the milling energy consumption; The step S4 comprises the following steps: S4.1, the improved fish eagle optimization algorithm is used to optimize the LSSVR hyperparameters, so as to select reasonable hyperparameters and improve the performance of the LSSVR algorithm; model 1, model 2 and model 3 are respectively trained and fitted models with the milling process parameter combination as input and the evaluation index three-dimensional surface roughness, milling energy consumption and three-dimensional fractal dimension as output; the data set is divided into training set D t1 , D t2 , and D t3 , test set D v1 , D v2 , and D v3 ; S4.2, complete setting fish eagle population N1, N2, iteration number T1, T2, and least square support vector regression parameter constraint, fish eagle individual position vector is by penalty coefficient γ, kernel parameter γ1 of radial basis kernel function, scaling factor α of polynomial kernel function p Composition, to test set D v1 , D v2 And root mean square error RMSE as the fitness function of improved fish eagle algorithm, the root mean square error RMSE calculation formula is: In the formula, y i is the measured data; y i (x) is the prediction result of the least squares support vector regression algorithm; N is the number of prediction samples. S4.3, based on the settings of steps S4.1 and S4.2, solving the hyperparameters of the LSSVR through steps S3.2 to S3.8; and further adopting the hyperparameter combination optimized by the improved fish eagle algorithm to train the LSSVR model F1, model F2 and model F3 through steps S2.2 to S2.

9.

2. The method for CF / PEEK milling process parameter optimization based on improved LSSVR-OOA algorithm according to claim 1, characterized in that, The step S2 comprises the following steps: S2.1, after the experiment is completed, according to the experimental results, the experimental data is processed, and the data set is set as D={(x i ,y i )|i=1,2,...n}, x i as an input variable, y i as an output variable; S2.2, pre-processing the data to remove abnormal values in the original data; S2.3, normalizing the data to enable the output values of each node to change within a certain range; the normalization equation is: where x i denotes the original variable; denotes the normalized variable; x max is the maximum value in the original variable; x min is the minimum value in the original variable; The transformed dataset is D = {(x i ,y i )|i = 1,2,...n}; S2.4, dividing the processed data into a training set and a test set according to a ratio of 8:2; and establishing three-dimensional surface roughness, three-dimensional fractal dimension and energy consumption models; S2.5, through the least squares support vector regression LSSVR, each training set data is segmented and modeled to obtain the prediction model of the overall data. The LSSVR algorithm uses nonlinear mapping Map the input space to a high-dimensional feature space and find the optimal linear function in the feature space; construct an improved LSSVR model based on multi-kernel learning, integrating radial basis kernel functions and polynomial kernel functions, and achieve collaborative modeling of local and global features through dynamic optimization of kernel weights to improve the robustness and adaptability of the model; Firstly, the prediction function y(x) of the LSSVR is represented as: In the formula, w represents a weight matrix; represents a nonlinear mapping of the original data into a high-dimensional space; b represents a bias; S2.6, according to the principle of structural risk minimization, the problem of mapping the original data to a high-dimensional space is converted into solving a quadratic programming problem: where ||w|| denotes the norm of the weight matrix; γ denotes a penalty parameter; e i denotes the true value y i the error between the predicted value y(x) and the true value y, i.e. the bias variable; S2.7, the quadratic programming problem is solved by the Lagrange multiplier method: The Lagrange function is defined as: In the formula, α i denotes the Lagrange multiplier; S2.8, w, b, a i , e i Taking the partial derivative and setting it to zero gives: The above formula is rearranged as: where 1 is a unit column vector of appropriate dimension, K is a kernel matrix, y = (y1, y2,..., yn)T is a vector of target values for the training samples, and I is an identity matrix of appropriate dimension. n ) T is a vector of target values for the training samples, and I is an identity matrix of appropriate dimension. S2.9, introducing a multi-kernel learning method, combining a local kernel function and a global kernel function to enable the model to maintain good prediction ability when facing different types of data; The radial basis kernel function is a local kernel function, and its mathematical expression is: where γ1> 0 is a kernel parameter that controls the width of the kernel function, x i and are two sample vectors in the input data set; The polynomial kernel function is a global kernel function, and its mathematical expression is: wherein α p > 0, c p and d are core parameters representing a scaling factor, a bias and the order of the polynomial, respectively; In order to combine the local characteristics of the radial basis kernel and the global characteristics of the polynomial kernel, a linear combination strategy is used to construct a multi-kernel function: In the formula, β1, β2≥ 0, as a weight coefficient, and satisfy β1+β2=1; Thus the improvement of LSSVR algorithm is completed, and the prediction function is expressed as: The optimization problem solved by the least squares support vector regression machine algorithm is a convex quadratic programming problem with only one equality constraint, which is suitable for modeling small sample data sets.

3. The method for CF / PEEK milling process parameter optimization based on improved LSSVR-OOA algorithm according to claim 2, characterized in that, The step S3 includes the following steps: S3.1, the Tent chaos mapping strategy is applied to the fish eagle population initialization process to enhance the global search ability, and the convex lens reverse learning strategy is added in the development stage to avoid the algorithm falling into local optimum, so as to complete the improvement of the fish eagle optimization algorithm; S3.2, the improved fish eagle optimization algorithm OOA is used for parameter optimization; there are N fish eagle population numbers, and the iteration number is set to T, the to-be-optimized parameter constraint is set, and the fish eagle population is initialized to {X i , i = 1, 2,..., N}, N represents the fish eagle population size, and the individual fish eagle is X i = {x i,1 , x i,2 , x i,3 ,....,x i,j}, j represents the dimension of the fish eagle position vector; The Tent mapping is used to initialize the population, a series of pseudo-random numbers between 0 and 1 are generated by Tent mapping, and then the chaos sequence is calculated, and the Tent mapping formula is as follows: Z k = (x z1 ,x z2 ,x z3 ...x zN ) wherein Z k+1 represents a chaotic sequence of the Tent map; N represents the number of individuals of the fish-eagle population; a z is a setting coefficient; The generated chaos sequence is mapped into the search space of the fish eagle optimization algorithm solution: x i,j = ld j +(ud j - ld j ) x Z k+1 where x i,j represents the value of the ith fish eagle in the jth dimension; lb j is the jth dimension for the lower bound of the search; ub j is the jth dimension for the upper bound of the search; S3.3, the fish eagle hunting process mainly includes two stages, the first stage is that the fish eagle hunts the fish after detecting its position, for each fish eagle, the position of other fish eagles with better fitness value in the search space is regarded as "underwater fish school", if not empty, the center position of "underwater fish school" is calculated, otherwise, the current individual position is kept; The position update formula is set as: In the formula, P represents a set of individual fish in the underwater fish school; represents the fitness value of the kth fish eagle; represents the fitness value of the jth fish eagle;X k represents the position of the kth fish eagle individual with better fitness;X best is the position of the best fish eagle individual; S3.4, the fish eagle randomly detects the position of one fish and kills it, and updates the new position of the fish eagle, and the formula is as follows: where, is the new position of the ith falcon in the exploration phase value in the jth dimension; r i,j denotes a random number between the interval [0,1]; SP i is the fish randomly selected by the ith falcon from its swarm P, is a random number between [0,1]; I i,j the set {1,2} is any number in the set; X i denotes the individual position of the ith falcon, denotes the falcon individual in the exploration phase; F ei denotes the individual fitness value of the ith falcon, is the fitness value of the ith falcon in the exploration phase; S3.5, after the fish eagle kills a fish, it will take it to a suitable position to eat; The behavior of the fish eagle makes the position of the fish eagle change slightly, thereby enhancing the local development ability of the algorithm; A new random position is calculated as "suitable fish eating position" by using the following formula, if the fitness value of the new position can be improved, the previous position of the corresponding fish eagle is replaced; where r is a random number in the interval [0, 1]; t is the current iteration number; T represents the maximum number of iterations required; is the new position of the i-th fish eagle in the exploration phase is the value in the j-th dimension; X i represents the individual position of the i-th fish eagle, represents the fish eagle individual in the exploration phase; is the fitness value of the i-th fish eagle in the development phase; S3.6, in the development stage of the fish eagle optimization algorithm, the reverse learning strategy of convex lens imaging principle is used to improve the fish eagle optimization algorithm, which is an improved strategy to expand the search range by calculating the reverse solution of the current position; The reverse solution of the convex lens imaging reverse learning strategy is calculated by the following formula: wherein represents the inverse solution generated by the convex lens imaging principle; k represents the ratio of the entity height to the imaging height, i.e. the scaling factor; The formula for setting the adjustment stretching factor k value is: n = (1 + (t / T) 1 / 2 ) 10 The reverse solution generated by smaller k has a larger range, while larger k produces a small range of reverse solutions; With the increase of the number of iterations, the value of k will become larger, and the range of its reverse solution will become smaller; S3.7, considering that the reverse solution is not necessarily better than the original solution, the greedy mechanism is introduced to select the individual after reverse learning to obtain the best individual, and the mathematical model formula of the greedy mechanism is as follows: In the formula, X new is the selected individual; S3.8, the improved fish eagle optimization algorithm is iterated and output, when the number of iterations is less than the maximum number of iterations T, the iteration processing is carried out according to steps S3.2 to S3.7; When the iteration termination condition is met, the fish eagle individual with the best fitness and its fitness value are output, and the value in the position coordinates of the fish eagle individual is obtained.

4. The method for CF / PEEK milling process parameter optimization based on improved LSSVR-OOA algorithm according to claim 1, characterized in that, The step S5 includes the following steps: S5.1, the model weight ratio calculation, the introduction of entropy weight method, using this method to establish multi-objective comprehensive evaluation function; for multi-objective optimization, first, the optimization index value is preprocessed; for the construction of comprehensive evaluation function, three kinds of indexes are set as the same attribute, so the three indexes are all converted into cost type indexes, as follows: where Sa min and Sa max are the minimum and maximum values of the three-dimensional surface roughness; P min and P max are the minimum and maximum values of the milling energy consumption; D min and D max are the minimum and maximum values of the three-dimensional fractal dimension; x represents the optimization variable, i.e., the milling process parameters; and respectively represent the values of the three-dimensional surface roughness, the energy consumption and the three-dimensional fractal dimension of the CF / PEEK milled under the current milling process parameters; f1(x), f2(x) and f3(x) respectively represent the pre-processed values; S5.2, the preprocessed value is substituted into the matrix to calculate the prior probability, that is, the proportion of each set of milling process parameters: In the formula, r ij represents the number of values of the index after pretreatment, m represents the number of values; R represents the standardized original data matrix R=(r ij ) m×n ; The proportion of the index value of the i th item under the j th index, that is, the prior probability: where p ij denotes the proportion of the index value, and denotes the proportion of the number of samples with value i in index j to the total number of samples. Thus, a proportion matrix P = {p ij} m×n ; S5.3, the calculation formula of entropy weight is: wherein denotes the entropy weight, k = 1 / ln(n) > 0, to satisfy S5.4, the information utility value is calculated based on the entropy weight: In the formula, d j information utility value of the jth index; S5.5, the size of the weight coefficient of each target is calculated by the information utility value calculation; In the formula, w * j indicates the weight coefficient size of the jth target calculated by the entropy weight method. S5.6, the weight of each target is assigned, and the comprehensive evaluation function is established as follows: In the formula, F e represents a comprehensive evaluation function established based on an entropy weight method; and respectively represent weight coefficients of the CF / PEEK milling three-dimensional surface roughness, milling energy consumption and three-dimensional fractal dimension. S5.7, after weighting the three models, a multi-objective comprehensive evaluation function is obtained, which is used as the fitness function of the improved fish eagle optimization algorithm: min f(x) = F e [f1,f2,f3] In the formula, f1, f2 and f3 represent three-dimensional surface roughness, milling energy consumption and three-dimensional fractal dimension respectively; S5.8, the improved OOA algorithm is used to optimize the three-dimensional surface roughness, three-dimensional fractal dimension and milling energy consumption of the milled CF / PEEK, the milling process parameters corresponding to the individual solution at the optimal milling quality are obtained, so as to realize the optimization of the milling process parameters; the first dimension is the size of the cutting speed V c , the second dimension is the size of the cutting depth α e , the third dimension is the size of the feed per tooth f z , and the fourth dimension is the value of the fiber direction angle θ. wherein x i,1 , x i,2 , x i,3 , x i,4 respectively represent the i-th fish eagle in the spindle speed, cutting depth, feed per tooth and fiber direction angle, the value of the four index dimensions; x i,1min , x i,1max respectively represent the minimum and maximum spindle speed; x i,2min , x i,2max respectively represent the minimum and maximum cutting depth; x i,3min , x i,3max respectively represent the minimum and maximum feed per tooth; u1, v1 are constants; S5.9, iterate according to steps S3.2 to S3.8; when the iteration termination is met, output the fish eagle individual with the best fitness and its fitness value, and obtain the optimal solution of multi-objective optimization; in the position coordinates of the fish eagle individual, each dimension is respectively the cutting speed V c , the cutting depth α e , the feed per tooth f z , and the fiber direction angle, to obtain the optimal process parameter combination of CF / PEEK milling multi-objective optimization.

Images