Magnesium alloy composite material performance evaluation method and system based on big data
By constructing the microstructure finite element model and state influence factor of magnesium alloy composite materials, combined with deviation analysis and model optimization, the problem of time-consuming and labor-intensive and difficult to fully reflect material performance evaluation methods is solved, and efficient and accurate evaluation of the performance of magnesium alloy composite materials is achieved.
Patent Information
- Application Number
- CN202411776401.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-05
- Publication Date
- 2025-03-28
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Traditional magnesium alloy performance evaluation methods are time-consuming and labor-intensive, and it is difficult to fully reflect material performance. The big data-based method has not yet integrated the microstructure and macro performance of the material.
By obtaining test data and state data of magnesium alloy composite materials, performing pre-processing and finite element simulation, a finite element model of microstructure is constructed to predict macroscopic material properties, combining state impact factors and deviation analysis, a performance evaluation model is constructed and optimized to achieve accurate evaluation.
It improves the efficiency and accuracy of performance evaluation of magnesium alloy composite materials, enhances model adaptability, and provides a scientific basis to improve the research, development, production and application of magnesium alloy composite materials.
Smart Images

Figure CN119724441A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of performance evaluation, and in particular to a method and system for evaluating the performance of a magnesium alloy composite material based on big data. Background Art
[0002] In the field of materials science, magnesium alloy composites are widely used due to their light weight, high strength and corrosion resistance. Accurate performance evaluation of magnesium alloy composites is of great significance for their quality assurance, performance improvement, cost control and technological innovation. At the same time, material performance evaluation based on big data technology combines mathematical statistics and machine learning technology to collect and analyze a large amount of material test data and status data to evaluate the target material performance in a numerical way, which can provide a scientific basis for quality improvement and decision-making.
[0003] However, most traditional magnesium alloy performance evaluation methods rely on experimental testing, which is not only time-consuming and labor-intensive, but also difficult to fully reflect the performance of the material; in addition, the data type of its performance evaluation is complex and the amount of data is huge. How to effectively use these big data to evaluate the performance of magnesium alloy composites is still an urgent problem to be solved. Although there are some material performance evaluation methods based on big data, most of these methods remain in the primary stage of data analysis and mining, lacking in-depth connection with the microstructure and macroscopic properties of the material. Combining big data technology and material science theory, a magnesium alloy composite performance evaluation method and system based on big data is designed to overcome the shortcomings of the existing evaluation model and realize a comprehensive and accurate evaluation of the performance of magnesium alloy composites, which is of great significance to the research and development, production and application of magnesium alloy composites. Summary of the invention
[0004] The purpose of the present invention is to provide a method and system for evaluating the performance of magnesium alloy composite materials based on big data.
[0005] To achieve the above object, the present invention is implemented according to the following technical solutions:
[0006] The present invention comprises the following steps:
[0007] Acquire test data and state data of the magnesium alloy composite material, and preprocess the test data and state data; the test data includes component test data and performance test data;
[0008] constructing a microstructure finite element model based on the component test data, and predicting macroscopic material properties based on the microstructure finite element model;
[0009] Determine a state influence factor according to the state data, correct the macro material property by the state influence factor to obtain a first material property, and perform multi-physics field finite element simulation to obtain performance simulation data;
[0010] Performing deviation analysis on the performance simulation data and the performance test data to obtain a deviation coefficient, and constructing a performance evaluation model for a magnesium alloy composite material;
[0011] The magnesium alloy composite material performance evaluation model is optimized according to the performance error, and the test data and state data of the magnesium alloy composite material to be evaluated are input into the optimized magnesium alloy composite material performance evaluation model to obtain the magnesium alloy performance evaluation result.
[0012] Furthermore, the method for predicting macroscopic material properties based on the microstructure finite element model includes:
[0013] A component test experiment is performed to obtain component test data of the magnesium alloy. The steps for obtaining the component test data include:
[0014] Conduct spectral analysis to determine the constituent elements and impurities of the magnesium alloy and determine the corresponding volume fractions;
[0015] Ultrasonic imaging technology is used to determine the distribution of reinforcement in magnesium alloys, and electron microscope scanning technology and image processing technology are used to determine the grain size, phase distribution and surface morphology of magnesium alloys;
[0016] A finite element model of the microstructure of the magnesium alloy was constructed based on the above component test data, and a finite element simulation of the microstructure of the magnesium alloy was performed. The macroscopic material properties were predicted by applying virtual loads and analyzing the structural response.
[0017] Furthermore, the method of correcting the macroscopic material property by the state influence factor to obtain the first material property includes:
[0018] Input the state data and component test data into the state influence function to obtain the state influence factor, the expression is:
[0019]
[0020] Affect is the state influence factor, w1, w2, w3 are the state influence factor weights, c i is the type of reinforcement in magnesium alloy, v i C in magnesium alloy i Volume content of reinforcement, a i c i The standard volume content of the reinforcement, n1 is the number of reinforcement types in the magnesium alloy, c j is the type of impurities in magnesium alloys, v j C in magnesium alloy jThe volume content of impurities, n2 is the number of impurity categories in the magnesium alloy, ΔH is the difference between the ambient humidity and the standard humidity H0 during the manufacture of the magnesium alloy, ΔT is the difference between the ambient temperature and the standard temperature T0 during the manufacture of the magnesium alloy, n3 is the number of processes experienced in the manufacture of the magnesium alloy, and p k ,q k is the influence index of the kth process, t k is the operation time of the kth process, t k0 is the standard operation time of the kth process, m k The operation score of the operation time of the kth process, m k0 Score the standard operation of the kth process;
[0021] According to the state influence factor, the macro material property is corrected to obtain the first material property, which is expressed as:
[0022]
[0023] Where P 1i is the first property value of the lth material property, P 0l is the macroscopic property value of the lth material property, and w4 is the weight of the first material property.
[0024] Furthermore, the method of performing deviation analysis on the performance simulation data and the performance test data to obtain a deviation coefficient includes:
[0025] A macroscopic finite element model of magnesium alloy is constructed by combining the microstructure finite element model and the first material property. Multi-physics field coupling simulation and RVE analysis are performed on the finite element model of magnesium alloy to obtain simulation data of magnesium alloy performance.
[0026] Conduct mechanical property tests, physical property tests, chemical property tests and damage property tests on magnesium alloys to obtain magnesium alloy performance test data, and perform labeling operations based on the test results;
[0027] Determine the deviation coefficients according to the aluminum alloy data categories: determine the first deviation coefficient according to the mechanical properties and physical properties data, and determine the second deviation coefficient according to the chemical properties and damage properties data;
[0028] The performance simulation data and performance test data corresponding to the mechanical properties and physical properties are input into the first deviation function to obtain the first deviation coefficient, which is expressed as:
[0029]
[0030] Where D1 is the first deviation coefficient, ΔK is the toughness deviation of the aluminum alloy, ΔΨ is the plasticity deviation of the aluminum alloy, ΔH is the hardness deviation of the aluminum alloy, σ totalis the total standard deviation of all mechanical property deviations, Δρ is the density deviation of aluminum alloy, μ1 and μ2 are deviation weight coefficients, Δλ is the thermal conductivity deviation of aluminum alloy, and u Δλ is the mean deviation of thermal conductivity of aluminum alloy, σ Δλ is the standard deviation of thermal conductivity of aluminum alloy, Δζ is the electrical conductivity deviation of aluminum alloy, u Δζ is the mean value of the conductivity deviation of aluminum alloy, σ Δζ is the standard deviation of the conductivity of aluminum alloy;
[0031] The performance simulation data and performance test data corresponding to the chemical properties and damage properties are input into the second deviation function to obtain the second deviation coefficient, which is expressed as:
[0032]
[0033] Where D2 is the second deviation coefficient, μ3, μ4, μ5 are deviation weight coefficients, Δf u-q is the deviation of the difference between the ultimate strength and yield strength of aluminum alloy, Δσ F is the mechanical fatigue property deviation of aluminum alloy, Δσ TF Thermal fatigue performance deviation of aluminum alloy, v CR is the corrosion rate deviation of aluminum alloy, and α is the oxidation resistance deviation of aluminum alloy.
[0034] Furthermore, the method for constructing the magnesium alloy composite material performance evaluation model comprises:
[0035] The deviation coefficient, state influencing factor, performance simulation data and performance test data are combined into an evaluation data set, and the evaluation data set is divided into a training set and a test set;
[0036] Constructing a magnesium alloy composite material performance evaluation model, wherein the magnesium alloy composite material performance evaluation model includes a multimodal prediction model, a strategy layer, and a particle optimization algorithm;
[0037] The multimodal prediction model predicts the performance parameters of magnesium alloys by integrating evaluation data sets, including random forest-based models, support vector machine-based models, and deep neural network-based models;
[0038] The random forest-based model predicts the mechanical and physical properties of magnesium alloys by processing the complex nonlinear relationship between the evaluation data sets; the support vector machine-based model is used to perform regression prediction on the damage performance and chemical properties of magnesium alloys; the deep neural network-based model is used to perform multimodal learning on the dependency relationship between performance simulation data and performance test data, state influencing factors, and deviation coefficients, and predict the performance of aluminum alloy composite materials;
[0039] The strategy layer uses a stacking system to integrate the prediction results of the base model, including multi-head feature fusion and stacking model fusion;
[0040] Multi-head feature fusion effectively captures local features of data through the convolution kernel of CNN, and adds a multi-head attention mechanism for data fusion. The expression is:
[0041]
[0042] MultiHead(Q,K,V)=Conect(head1,head2,…,head h )W 0
[0043] head i =Attention(QW i Q ,KW i K ,VW i V )
[0044] Where Attention(Q,K,V) is the mapping of the query matrix Q in the attention mechanism on the key-value pair, K is the attention content key matrix, V is the attention content value matrix, QK T is the dot multiplication operation of the matrix, which is used to query the matching degree of the matrix. k is the scaling factor, the softmax(·) function is used to generate the attention weight, MultiHead(Q,K,V) is the multi-head attention mechanism representation, which is used to connect multiple attentions, W 0 , W i Q , W i K , W i V is the parameter matrix of the i-th head, head i is the output of the i-th head, and h is the number of attention heads;
[0045] The steps of the stacking strategy are: perform a p-fold cross-check on the fusion results to assign a data set to each base learner, generate data from the data set through the base learner, determine the base learner framework and the meta-learner framework, and integrate the prediction results of the base model through the learner framework;
[0046] The output layer extracts the prediction result labels and finally outputs the performance evaluation results of the magnesium alloy composite material;
[0047] The particle optimization algorithm is used to adjust the hyperparameters of the multimodal prediction model according to the error between the predicted performance and the measured performance.
[0048] Furthermore, the method for optimizing the magnesium alloy composite material performance evaluation model according to the performance error includes:
[0049] Determine the magnesium alloy performance error objective function of the searcher, the expression is:
[0050]
[0051] where F Aim is the objective function of magnesium alloy performance error, and the optimization goal is to minimize the overall error between the predicted value and the true value of magnesium alloy performance, p ij is the true value of the i-th index of the j-th type of performance of magnesium alloy, p i ′ j is the predicted value of the i-th index of the j-th type of performance of magnesium alloy, loss ij is the loss function of the i-th index of the j-th type of performance of magnesium alloy, is the weight of the jth type of performance of magnesium alloy, is the scaling factor, g(·) is the spacing function, cosθ ij For p ij Corresponding category The cosine similarity of n j For p ij The number of indicators in the corresponding category, m is the interval term hyperparameter, and M is the total number of indicators;
[0052] Initialize the particle swarm position, perform chaotic mapping on the population position and determine the chaotic interference factor, the expression is:
[0053]
[0054]
[0055] where x n is the nth element after chaotic mapping in the chaotic sequence, C is the chaotic interference factor, is the control parameter, ρ is the chaotic learning rate;
[0056] The particle speed and position are updated according to the chaos factor. The expression is:
[0057]
[0058] Where V i (t+1) is the particle velocity updated at the t+1th iteration, X i (t+1) is the particle position updated at the t+1th iteration, Φ is the inertia weight, l1 is the learning factor, r1∈(0,1) is a random number, G b (t) is the optimal position of the particle group, X i (t) is the current position of particle i, P b,i (t) is the historical optimal position of particle i, N(·) is a d-dimensional normally distributed random variable, Ed is the expected value of the d-dimensional perturbation, D is the dimension of the search space, α, β, δ, γ are the perturbation intensity parameters, and f b is the best fitness of the particle in history, f d is the fitness of the particle in the current dimension, Cauchy(0,τ) represents the Cauchy distribution with center 0 and scale parameter τ;
[0059] The particle fitness is calculated, and the optimal position of the particle group and the historical optimal position of the particle are updated. The iteration is continued until the magnesium alloy performance error objective function is minimized or the maximum number of iterations is reached. The test data and state data of the magnesium alloy composite material to be evaluated are input into the optimized magnesium alloy composite material performance evaluation model to obtain the magnesium alloy performance evaluation results.
[0060] In the second aspect, a magnesium alloy composite material performance evaluation system based on big data includes:
[0061] Data acquisition module: used to obtain test data and status data of magnesium alloy composite materials, and pre-process the test data and status data;
[0062] Data processing module: used to construct the state influencing factor, to modify the macro material property to obtain the first material property, and to perform deviation analysis on the performance simulation data and the performance test data to obtain a deviation coefficient;
[0063] Finite element module: used for constructing a microstructure finite element model according to the component test data to predict macroscopic material properties, and for performing multi-physics field finite element simulation according to the microstructure finite element model and the first material properties to obtain performance simulation data;
[0064] Model module: used to construct a magnesium alloy composite material performance evaluation model, optimize the magnesium alloy composite material performance evaluation model according to the performance error, and input the test data and state data of the magnesium alloy composite material to be evaluated into the optimized magnesium alloy composite material performance evaluation model to obtain the magnesium alloy performance evaluation result;
[0065] Intelligent supervision module: used for storing, viewing and managing the test data, status data and performance evaluation results of the magnesium alloy composite material, and displaying the performance evaluation results to the user.
[0066] The beneficial effects of the present invention are:
[0067] The present invention is a method and system for evaluating the performance of magnesium alloy composite materials based on big data. Compared with the prior art, the present invention has the following technical effects:
[0068] The present invention can improve the data preprocessing capability and enhance the model adaptability in the performance evaluation of magnesium alloy composite materials through finite element model simulation, state influencing factor determination, macro material property correction, deviation analysis, model construction and model optimization steps, thereby improving the efficiency and accuracy of the performance evaluation of magnesium alloy composite materials. The performance evaluation technology of magnesium alloy composite materials is optimized, which can greatly save resources, improve work efficiency, and realize the performance evaluation of magnesium alloy composite materials. It provides a scientific basis for improving magnesium alloy composite materials, is of great significance to the research and development, production and application of magnesium alloy composite materials, can adapt to different magnesium alloy composite material performance evaluation systems based on big data, and the terminal evaluation needs of magnesium alloy composite material performance evaluation systems based on big data of different users, and has a certain universality. BRIEF DESCRIPTION OF THE DRAWINGS
[0069] Figure 1 The present invention is a flowchart of the steps of a method for evaluating the performance of magnesium alloy composite materials based on big data. DETAILED DESCRIPTION
[0070] The present invention is further described below by means of specific embodiments. The illustrative embodiments and descriptions of the present invention are used to explain the present invention but are not intended to limit the present invention.
[0071] The present invention provides a magnesium alloy composite material performance evaluation method and system based on big data, comprising the following steps:
[0072] like Figure 1 As shown, in this embodiment, the following steps are included:
[0073] Acquire test data and state data of the magnesium alloy composite material, and preprocess the test data and state data; the test data includes component test data and performance test data;
[0074] constructing a microstructure finite element model based on the component test data, and predicting macroscopic material properties based on the microstructure finite element model;
[0075] Determine a state influence factor according to the state data, correct the macro material property by the state influence factor to obtain a first material property, and perform multi-physics field finite element simulation to obtain performance simulation data;
[0076] Performing deviation analysis on the performance simulation data and the performance test data to obtain a deviation coefficient, and constructing a performance evaluation model for a magnesium alloy composite material;
[0077] The magnesium alloy composite material performance evaluation model is optimized according to the performance error, and the test data and state data of the magnesium alloy composite material to be evaluated are input into the optimized magnesium alloy composite material performance evaluation model to obtain the magnesium alloy performance evaluation result.
[0078] In this embodiment, the method for predicting macroscopic material properties according to the microstructure finite element model includes:
[0079] A component test experiment is performed to obtain component test data of the magnesium alloy. The steps for obtaining the component test data include:
[0080] Conduct spectral analysis to determine the constituent elements and impurities of the magnesium alloy and determine the corresponding volume fractions;
[0081] Ultrasonic imaging technology is used to determine the distribution of reinforcement in magnesium alloys, and electron microscope scanning technology and image processing technology are used to determine the grain size, phase distribution and surface morphology of magnesium alloys;
[0082] A microstructure finite element model of the magnesium alloy is constructed based on the above component test data, and a finite element simulation of the microstructure of the magnesium alloy is performed to predict macroscopic material properties by applying virtual loads and analyzing structural responses;
[0083] In the actual evaluation, a magnesium alloy test block was tested to obtain component test data: 1. Main elements and proportions (aluminum 90%, magnesium 6%, zinc 2.5%, copper 1%); 2. Impurities and proportions (iron 0.05%, silicon 0.1%, manganese 0.03%, chromium 0.02%); 3. Reinforcement distribution (aggregation index 2.0; silicon carbide SiC: average diameter 10μm, volume fraction 15%; aluminum oxide Al2O3: average diameter 8μm, volume fraction 10%); 4. Average grain size (aluminum 10μm, magnesium 8μm, zinc 10μm); 5. Average grain size (aluminum 10μm, magnesium 8μm, zinc 10μm); 6. Average grain size (aluminum 10μm, magnesium 8μm, zinc 10μm); 7. Average grain size (aluminum 10μm, magnesium 8μm, zinc 10μm); 8. Average grain size (aluminum 10μm, magnesium 8μm, zinc 10μm); 9. Average grain size (aluminum 10μm, magnesium 8μm, zinc 10μm); 10. Average grain size (aluminum 10μm, magnesium 8μm, zinc 10μm); 11. Average grain size (aluminum 10μm, magnesium 8μm, zinc 10μm); 12. Average grain size (aluminum 10μm, magnesium 8μm, zinc 10μm); 13. Average grain size (aluminum 10μm, magnesium 8μm, zinc 10μm); 14. Average grain size (aluminum 10μm, magnesium 8μm, zinc 10μm); 15. Average grain size (aluminum 10μm, magnesium 8μm, zinc 10μm); 16. 5μm, copper 4μm); 5. Phase distribution (α-Al phase: average grain size 8μm, volume 70%; β-Mg2Si phase: average grain size 7μm, volume 20%; Al2Cu phase: average grain size 3μm, volume 5%; Al2O3 phase: average grain size 8μm, volume 5%); 6. Average distance between grains 2μm, average distance between Al2Cu phase grains and other grains 1.5μm; 7. Average surface roughness 0.4μm, slight wear, depth less than 0.15μm;
[0084] According to the above component test data, a finite element model of the microstructure of the magnesium alloy was constructed, and a finite element simulation of the microstructure of the magnesium alloy was performed. By applying virtual loads and analyzing the structural response, the macroscopic material properties were predicted: elastic modulus 50 GPa, yield strength 300 MPa, and ultimate strength 400 MPa.
[0085] In this embodiment, the method of correcting the macro material property by the state influence factor to obtain the first material property includes:
[0086] Input the state data and component test data into the state influence function to obtain the state influence factor, the expression is:
[0087]
[0088] Affect is the state influence factor, w1, w2, w3 are the state influence factor weights, c i is the type of reinforcement in magnesium alloy, v i C in magnesium alloy i Volume content of reinforcement, a i c i The standard volume content of the reinforcement, n1 is the number of reinforcement types in the magnesium alloy, c j is the type of impurities in magnesium alloys, v j C in magnesium alloy j The volume content of impurities, n2 is the number of impurity categories in the magnesium alloy, ΔH is the difference between the ambient humidity and the standard humidity H0 during the manufacture of the magnesium alloy, ΔT is the difference between the ambient temperature and the standard temperature T0 during the manufacture of the magnesium alloy, n3 is the number of processes experienced in the manufacture of the magnesium alloy, and p k ,q k is the influence index of the kth process, t k is the operation time of the kth process, t k0 is the standard operation time of the kth process, m k The operation score of the operation time of the kth process, m k0 Score the standard operation of the kth process;
[0089] According to the state influence factor, the macro material property is corrected to obtain the first material property, which is expressed as:
[0090]
[0091] Where P 1i is the first property value of the lth material property, P 0l is the macroscopic property value of the lth material property, w4 is the weight of the first material property;
[0092] In the actual evaluation, the processes for manufacturing magnesium alloys include: smelting (operation time 120min, operation score 90, standard time 100min, standard score 85), casting (operation time 90min, operation score 95, standard time 95min, standard score 90), heat treatment (operation time 60min, operation score 88, standard time 70min, standard score 82), homogenization annealing (operation time 240min, operation score 88, standard time 240min, standard score 85), machining (operation time 150min, operation score 92, standard time 140min, standard score 88);
[0093] The state data of magnesium alloy manufacturing was obtained (standard humidity 50%, ΔH=5%, standard temperature 20°C, ΔT=2°C). When the state influence factor weights w1=0.4, w2=0.3, w3=0.3 were determined, the influence indexes of each process were 1.1 / 1.05, 1.05 / 1.1, 1.2 / 1.15, 1.15 / 1.2, respectively. The state influence factor Affect=0.6183 was calculated, and the corrected first material properties were: elastic modulus 53.4GPa, yield strength 320.4MPa, and ultimate strength 427.2MPa.
[0094] In this embodiment, the method of performing deviation analysis on the performance simulation data and the performance test data to obtain the deviation coefficient includes:
[0095] A macroscopic finite element model of magnesium alloy is constructed by combining the microstructure finite element model and the first material property. Multi-physics field coupling simulation and RVE analysis are performed on the finite element model of magnesium alloy to obtain simulation data of magnesium alloy performance.
[0096] Conduct mechanical property tests, physical property tests, chemical property tests and damage property tests on magnesium alloys to obtain magnesium alloy performance test data, and perform labeling operations based on the test results;
[0097] Determine the deviation coefficients according to the aluminum alloy data categories: determine the first deviation coefficient according to the mechanical properties and physical properties data, and determine the second deviation coefficient according to the chemical properties and damage properties data;
[0098] The performance simulation data and performance test data corresponding to the mechanical properties and physical properties are input into the first deviation function to obtain the first deviation coefficient, which is expressed as:
[0099]
[0100] Where D1 is the first deviation coefficient, ΔK is the toughness deviation of the aluminum alloy, ΔΨ is the plasticity deviation of the aluminum alloy, ΔH is the hardness deviation of the aluminum alloy, σ total is the total standard deviation of all mechanical property deviations, Δρ is the density deviation of aluminum alloy, μ1 and μ2 are deviation weight coefficients, Δλ is the thermal conductivity deviation of aluminum alloy, and u Δλ is the mean deviation of thermal conductivity of aluminum alloy, σ Δλ is the standard deviation of thermal conductivity of aluminum alloy, Δζ is the electrical conductivity deviation of aluminum alloy, u Δζ is the mean value of the conductivity deviation of aluminum alloy, σ Δζ is the standard deviation of the conductivity of aluminum alloy;
[0101] The performance simulation data and performance test data corresponding to the chemical properties and damage properties are input into the second deviation function to obtain the second deviation coefficient, which is expressed as:
[0102]
[0103] Where D2 is the second deviation coefficient, μ3, μ4, μ5 are deviation weight coefficients, Δf u-q is the deviation of the difference between the ultimate strength and yield strength of aluminum alloy, Δσ F is the mechanical fatigue property deviation of aluminum alloy, Δσ TF Thermal fatigue performance deviation of aluminum alloy, v CR is the corrosion rate deviation of aluminum alloy, α is the oxidation resistance deviation of aluminum alloy;
[0104] In the actual evaluation, the macroscopic finite element model of magnesium alloy was constructed by combining the microstructure finite element model and the first material property to obtain the performance simulation value of magnesium alloy composite material: (1. Mechanical properties: toughness 10MPa·m 1 / 2 , elongation 15%, hardness 120HV; 2. Physical properties: density 1.8g / cm 3 , thermal conductivity 60W / (m·K), electrical conductivity 2.5×10 7 S / m; 3. Chemical properties: corrosion rate 0.1mm / a, oxidation resistance 800℃; 4. Thermal fatigue 1000 cycles, stress fatigue 5000 cycles); the corresponding performance measured data are: (1. Mechanical properties: toughness 9.5MPa·m 1 / 2 , elongation 14.25%, hardness 114HV; 2. Physical properties: density 1.76g / cm 3 , thermal conductivity 57W / (m·K), electrical conductivity 2.465×10 7 S / m; 3. Chemical properties: corrosion rate 0.095mm / a, oxidation resistance 800℃; 4. Thermal fatigue 1066 cycles, stress fatigue 5121 cycles);
[0105] According to the above data, the first deviation coefficient and the second deviation coefficient are calculated to be 1.38966 and 0.01601 respectively.
[0106] In this embodiment, the method for constructing the magnesium alloy composite material performance evaluation model includes:
[0107] The deviation coefficient, state influencing factor, performance simulation data and performance test data are combined into an evaluation data set, and the evaluation data set is divided into a training set and a test set;
[0108] Constructing a magnesium alloy composite material performance evaluation model, wherein the magnesium alloy composite material performance evaluation model includes a multimodal prediction model, a strategy layer, and a particle optimization algorithm;
[0109] The multimodal prediction model predicts the performance parameters of magnesium alloys by integrating evaluation data sets, including random forest-based models, support vector machine-based models, and deep neural network-based models;
[0110] The random forest-based model predicts the mechanical and physical properties of magnesium alloys by processing the complex nonlinear relationship between the evaluation data sets, using the mean square error loss function to measure the difference between the predicted value and the true value, and using grid search to adjust the random forest parameters; the support vector machine-based model is used to perform regression prediction on the damage performance and chemical properties of magnesium alloys, using the radial basis function as the kernel function, the hinge loss function to process different categories of chemical data and damage data, and the sequential minimum optimization algorithm quadratic programming model; the deep neural network-based model is used to perform multimodal learning on the dependency relationship between performance simulation data and performance test data, state influencing factors, and deviation coefficients, and predict the performance of aluminum alloy composites, using the cross entropy loss function to measure the difference between the predicted value and the true value, and using the Adam optimizer to adjust the learning rate at different training stages;
[0111] The strategy layer uses a stacking system to integrate the prediction results of the base model, uses cross-validation and regularization techniques to select the model complexity, and selects an effective magnesium alloy performance subset through attribute simplification and feature selection; the stacking system of the body includes multi-head feature fusion and stacking model fusion;
[0112] Multi-head feature fusion effectively captures local features of data through the convolution kernel of CNN, and adds a multi-head attention mechanism for data fusion. The expression is:
[0113]
[0114] MultiHead(Q,K,V)=Conect(head1,head2,…,head h )W 0
[0115] head i =Attention(QW i Q ,KW i K ,VW i V )
[0116] Where Attention(Q,K,V) is the mapping of the query matrix Q in the attention mechanism on the key-value pair, K is the attention content key matrix, V is the attention content value matrix, QK T is the dot multiplication operation of the matrix, which is used to query the matching degree of the matrix. kis the scaling factor, the softmax(·) function is used to generate the attention weight, MultiHead(Q,K,V) is the multi-head attention mechanism representation, which is used to connect multiple attentions, W 0 , W i Q , W i K , W i V is the parameter matrix of the i-th head, head i is the output of the i-th head, and h is the number of attention heads;
[0117] The steps of the stacking strategy are: perform a p-fold cross-check on the fusion results to assign a data set to each base learner, generate data from the data set through the base learner, determine the base learner framework and the meta-learner framework, and integrate the prediction results of the base model through the learner framework. The expression is;
[0118]
[0119] g=f i (D i ),G=F i (g)
[0120] where D∈R n×m is the input data set, n is the data dimension, m is the number of data entries, D q One of the p pieces of data divided into the dataset D is input into the base learner q, and each piece D j ′ Contains part of the original dataset, [D] i+k is all the data in the i+kth column, g is the base learner framework, f(·) is the base learner function, G is the meta learner framework, and F(·) is the meta learner function;
[0121] The output layer extracts the prediction result labels and finally outputs the performance evaluation results of the magnesium alloy composite material;
[0122] The particle optimization algorithm is used to adjust the hyperparameters of the multimodal prediction model according to the error between the predicted performance and the measured performance.
[0123] In this embodiment, the method for optimizing the magnesium alloy composite material performance evaluation model according to the performance error includes:
[0124] Determine the magnesium alloy performance error objective function of the searcher, the expression is:
[0125]
[0126] where F Aimis the objective function of magnesium alloy performance error, and the optimization goal is to minimize the overall error between the predicted value and the true value of magnesium alloy performance, p ij is the true value of the i-th index of the j-th type of performance of magnesium alloy, p i ′ j is the predicted value of the i-th index of the j-th type of performance of magnesium alloy, loss ij is the loss function of the i-th index of the j-th type of performance of magnesium alloy, is the weight of the jth type of performance of magnesium alloy, is the scaling factor, g(·) is the spacing function, cosθ ij For p ij Corresponding category The cosine similarity of n j For p ij The number of indicators in the corresponding category, m is the interval term hyperparameter, and M is the total number of indicators;
[0127] Initialize the particle swarm position, perform chaotic mapping on the population position and determine the chaotic interference factor, the expression is:
[0128]
[0129] where x n is the nth element after chaotic mapping in the chaotic sequence, C is the chaotic interference factor, is the control parameter, ρ is the chaotic learning rate;
[0130] The particle speed and position are updated according to the chaos factor. The expression is:
[0131]
[0132] Where V i (t+1) is the particle velocity updated at the t+1th iteration, X i (t+1) is the particle position updated at the t+1th iteration, Φ is the inertia weight, l1 is the learning factor, r1∈(0,1) is a random number, G b (t) is the optimal position of the particle group, X i (t) is the current position of particle i, P b,i (t) is the historical optimal position of particle i, N(·) is a d-dimensional normally distributed random variable, E d is the expected value of the d-dimensional perturbation, D is the dimension of the search space, α, β, δ, γ are the perturbation intensity parameters, and f b is the best fitness of the particle in history, f d is the fitness of the particle in the current dimension, Cauchy(0,τ) represents the Cauchy distribution with center 0 and scale parameter τ;
[0133] Calculate the particle fitness, and update the optimal position of the particle group and the historical optimal position of the particle, and iterate continuously until the magnesium alloy performance error objective function is minimized or the maximum number of iterations is reached. Then, input the test data and state data of the magnesium alloy composite material to be evaluated into the optimized magnesium alloy composite material performance evaluation model to obtain the magnesium alloy performance evaluation result.
[0134] In the actual evaluation, the test data and state data of the aluminum alloy composite material to be evaluated were obtained: 1. Main elements and proportions (aluminum 89.8%, magnesium 6.2%, zinc 2.4%, copper 1.1%); 2. Impurities and proportions (iron 0.04%, silicon 0.12%, manganese 0.025%, chromium 0.03%); 3. Reinforcement distribution (aggregation index 1.8; silicon carbide SiC: average diameter 9.5μm, volume fraction 16%; alumina Al2O3: average diameter 7.8μm, volume fraction 11%); 4. Average grain size (aluminum 10.2μm, magnesium 7.8μm, zinc 5.2μm, copper 3.8μm); 5. Phase distribution (α- Al phase: average grain size 7.8μm, volume 68%; β-Mg2Si phase: average grain size 7.2μm, volume 22%; Al2Cu phase: average grain size 3.2μm, volume 6%; Al2O3 phase: average grain size 7.5μm, volume 5%); 6. Average distance between grains 2.2μm, average distance between Al2Cu phase grains and other grains 1.7μm; 7. Average surface roughness 0.38μm, slight wear, depth less than 0.14μm; 8. State data when manufacturing magnesium alloy (standard humidity 50%, ΔH=4%, standard temperature 20℃, ΔT=1.5℃);
[0135] The test data and state data of the magnesium alloy composite material to be evaluated are input into the optimized magnesium alloy composite material performance evaluation model to obtain the magnesium alloy performance evaluation results: (1. Mechanical properties: toughness 11.1MPa·m 1 / 2 , elongation 15.3%, hardness 125HV; 2. Physical properties: density 1.82g / cm 3 , thermal conductivity 58W / (m·K), electrical conductivity 2.61×10 7 S / m; 3. Chemical properties: corrosion rate 0.09mm / a, oxidation resistance 820℃; 4. Thermal fatigue 969 cycles, stress fatigue 5100 cycles).
[0136] In the second aspect, a magnesium alloy composite material performance evaluation system based on big data includes:
[0137] Data acquisition module: used to obtain test data and status data of magnesium alloy composite materials, and pre-process the test data and status data;
[0138] Data processing module: used to construct the state influencing factor, to modify the macro material property to obtain the first material property, and to perform deviation analysis on the performance simulation data and the performance test data to obtain a deviation coefficient;
[0139] Finite element module: used for constructing a microstructure finite element model according to the component test data to predict macroscopic material properties, and for performing multi-physics field finite element simulation according to the microstructure finite element model and the first material properties to obtain performance simulation data;
[0140] Model module: used to construct a magnesium alloy composite material performance evaluation model, optimize the magnesium alloy composite material performance evaluation model according to the performance error, and input the test data and state data of the magnesium alloy composite material to be evaluated into the optimized magnesium alloy composite material performance evaluation model to obtain the magnesium alloy performance evaluation result;
[0141] Intelligent supervision module: used for storing, viewing and managing the test data, status data and performance evaluation results of the magnesium alloy composite material, and displaying the performance evaluation results to the user.
[0142] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims
1. A performance evaluation method for magnesium alloy composite materials based on big data, characterized in that: The following steps are involved: S1. Acquire test data and state data of the magnesium alloy composite material, and preprocess the test data and state data; the test data includes component test data and performance test data; S2. constructing a microstructure finite element model according to the component test data, and predicting macroscopic material properties according to the microstructure finite element model; S3, determining a state influence factor according to the state data, correcting the macro material property by the state influence factor to obtain a first material property, and performing a multi-physics field finite element simulation to obtain performance simulation data; S4, performing deviation analysis on the performance simulation data and the performance test data to obtain a deviation coefficient, and constructing a performance evaluation model for a magnesium alloy composite material; S5. Optimizing the magnesium alloy composite material performance evaluation model according to the performance error, and inputting the test data and state data of the magnesium alloy composite material to be evaluated into the optimized magnesium alloy composite material performance evaluation model to obtain the magnesium alloy performance evaluation result.
2. The method for evaluating the performance of magnesium alloy composite materials based on big data according to claim 1, characterized in that: The method for predicting macroscopic material properties according to the microstructure finite element model comprises: A component test experiment is performed to obtain component test data of the magnesium alloy. The steps for obtaining the component test data include: Conduct spectral analysis to determine the constituent elements and impurities of the magnesium alloy and determine the corresponding volume fractions; Ultrasonic imaging technology is used to determine the distribution of reinforcement in magnesium alloys, and electron microscope scanning technology and image processing technology are used to determine the grain size, phase distribution and surface morphology of magnesium alloys; A finite element model of the microstructure of the magnesium alloy was constructed based on the above component test data, and a finite element simulation of the microstructure of the magnesium alloy was performed. The macroscopic material properties were predicted by applying virtual loads and analyzing the structural response.
3. The method for evaluating the performance of magnesium alloy composite materials based on big data according to claim 1, characterized in that: The method of correcting the macroscopic material property by the state influence factor to obtain the first material property comprises: Input the state data and component test data into the state influence function to obtain the state influence factor, the expression is: Affect is the state influence factor, w1, w2, w3 are the state influence factor weights, c i is the type of reinforcement in magnesium alloy, v i C in magnesium alloy i Volume content of reinforcement, a i c i The standard volume content of the reinforcement, n1 is the number of reinforcement types in the magnesium alloy, c j is the type of impurities in magnesium alloys, v j C in magnesium alloy j The volume content of impurities, n2 is the number of impurity categories in the magnesium alloy, ΔH is the difference between the ambient humidity and the standard humidity H0 during the manufacture of the magnesium alloy, ΔT is the difference between the ambient temperature and the standard temperature T0 during the manufacture of the magnesium alloy, n3 is the number of processes experienced in the manufacture of the magnesium alloy, and p k ,q k is the influence index of the kth process, t k is the operation time of the kth process, t k0 is the standard operation time of the kth process, m k The operation score of the operation time of the kth process, m k0 Score the standard operation of the kth process; According to the state influence factor, the macro material property is corrected to obtain the first material property, which is expressed as: Where P 1i is the first property value of the lth material property, P 0l is the macroscopic property value of the lth material property, and w4 is the weight of the first material property.
4. The method for evaluating the performance of magnesium alloy composite materials based on big data according to claim 1, characterized in that: The method of performing deviation analysis on the performance simulation data and the performance test data to obtain a deviation coefficient comprises: A macroscopic finite element model of magnesium alloy is constructed by combining the microstructure finite element model and the first material property. Multi-physics field coupling simulation and RVE analysis are performed on the finite element model of magnesium alloy to obtain simulation data of magnesium alloy performance. Conduct mechanical property tests, physical property tests, chemical property tests and damage property tests on magnesium alloys to obtain magnesium alloy performance test data, and perform labeling operations based on the test results; Determine the deviation coefficients according to the aluminum alloy data categories: determine the first deviation coefficient according to the mechanical properties and physical properties data, and determine the second deviation coefficient according to the chemical properties and damage properties data; The performance simulation data and performance test data corresponding to the mechanical properties and physical properties are input into the first deviation function to obtain the first deviation coefficient, which is expressed as: Where D1 is the first deviation coefficient, ΔK is the toughness deviation of the aluminum alloy, ΔΨ is the plasticity deviation of the aluminum alloy, ΔH is the hardness deviation of the aluminum alloy, σ total is the total standard deviation of all mechanical property deviations, Δρ is the density deviation of aluminum alloy, μ1 and μ2 are deviation weight coefficients, Δλ is the thermal conductivity deviation of aluminum alloy, and u Δλ is the mean deviation of thermal conductivity of aluminum alloy, σ Δλ is the standard deviation of thermal conductivity of aluminum alloy, Δζ is the electrical conductivity deviation of aluminum alloy, u Δζ is the mean value of the conductivity deviation of aluminum alloy, σ Δζ is the standard deviation of the conductivity of aluminum alloy; The performance simulation data and performance test data corresponding to the chemical properties and damage properties are input into the second deviation function to obtain the second deviation coefficient, which is expressed as: Where D2 is the second deviation coefficient, μ3, μ4, μ5 are deviation weight coefficients, Δf u-q is the deviation of the difference between the ultimate strength and yield strength of aluminum alloy, Δσ F is the mechanical fatigue property deviation of aluminum alloy, Δσ TF Thermal fatigue performance deviation of aluminum alloy, v CR is the corrosion rate deviation of aluminum alloy, and α is the oxidation resistance deviation of aluminum alloy.
5. The method for evaluating the performance of magnesium alloy composite materials based on big data according to claim 1, characterized in that: The method for constructing the magnesium alloy composite material performance evaluation model comprises: The deviation coefficient, state influencing factor, performance simulation data and performance test data are combined into an evaluation data set, and the evaluation data set is divided into a training set and a test set; Constructing a magnesium alloy composite material performance evaluation model, wherein the magnesium alloy composite material performance evaluation model includes a multimodal prediction model, a strategy layer, and a particle optimization algorithm; The multimodal prediction model predicts the performance parameters of magnesium alloys by integrating evaluation data sets, including random forest-based models, support vector machine-based models, and deep neural network-based models; The random forest-based model predicts the mechanical and physical properties of magnesium alloys by processing the complex nonlinear relationship between the evaluation data sets; the support vector machine-based model is used to perform regression prediction on the damage performance and chemical properties of magnesium alloys; the deep neural network-based model is used to perform multimodal learning on the dependency relationship between performance simulation data and performance test data, state influencing factors, and deviation coefficients, and predict the performance of aluminum alloy composite materials; The strategy layer uses a stacking system to integrate the prediction results of the base model, including multi-head feature fusion and stacking model fusion; Multi-head feature fusion effectively captures local features of data through the convolution kernel of CNN, and adds a multi-head attention mechanism for data fusion. The expression is: MultiHead(Q,K,V)=Conect(head1,head2,…,head h )W 0 head i =Attention(QW i Q ,KW i K ,VW i V ) Where Attention(Q,K,V) is the mapping of the query matrix Q in the attention mechanism on the key-value pair, K is the attention content key matrix, V is the attention content value matrix, QK T is the dot multiplication operation of the matrix, which is used to query the matching degree of the matrix. k is the scaling factor, the softmax(·) function is used to generate the attention weight, MultiHead(Q,K,V) is the multi-head attention mechanism representation, which is used to connect multiple attentions, W 0 , W i Q , W i K , W i V is the parameter matrix of the i-th head, head i is the output of the i-th head, and h is the number of attention heads; The steps of the stacking strategy are: perform a p-fold cross-check on the fusion results to assign a data set to each base learner, generate data from the data set through the base learner, determine the base learner framework and the meta-learner framework, and integrate the prediction results of the base model through the learner framework; The output layer extracts the prediction result labels and finally outputs the performance evaluation results of the magnesium alloy composite material; The particle optimization algorithm is used to adjust the hyperparameters of the multimodal prediction model according to the error between the predicted performance and the measured performance.
6. The method for evaluating the performance of magnesium alloy composite materials based on big data according to claim 1, characterized in that: The method for optimizing the magnesium alloy composite material performance evaluation model according to the performance error comprises: Determine the magnesium alloy performance error objective function of the searcher, the expression is: Among them, F Aim is the objective function of magnesium alloy performance error, and the optimization goal is to minimize the overall error between the predicted value and the true value of magnesium alloy performance, p ij is the true value of the i-th index of the j-th type of performance of magnesium alloy, p i ′ j is the predicted value of the i-th index of the j-th type of performance of magnesium alloy, loss ij is the loss function of the i-th index of the j-th type of performance of magnesium alloy, is the weight of the jth type of performance of magnesium alloy, is the scaling factor, g(·) is the spacing function, cosθ ij For p ij Corresponding category The cosine similarity of n j For p ij The number of indicators in the corresponding category, m is the interval term hyperparameter, and M is the total number of indicators; Initialize the particle swarm position, perform chaotic mapping on the population position and determine the chaotic interference factor, the expression is: where x n is the nth element after chaotic mapping in the chaotic sequence, C is the chaotic interference factor, is the control parameter, ρ is the chaotic learning rate; The particle speed and position are updated according to the chaos factor. The expression is: Where V i (t+1) is the particle velocity updated at the t+1th iteration, X i (t+1) is the particle position updated at the t+1th iteration, Φ is the inertia weight, l1 is the learning factor, r1∈(0,1) is a random number, G b (t) is the optimal position of the particle group, X i (t) is the current position of particle i, P b,i (t) is the historical optimal position of particle i, N(·) is a d-dimensional normally distributed random variable, E d is the expected value of the d-dimensional perturbation, D is the dimension of the search space, α, β, δ, γ are the perturbation intensity parameters, and f b is the best fitness of the particle in history, f d is the fitness of the particle in the current dimension, Cauchy(0,τ) represents the Cauchy distribution with center 0 and scale parameter τ; The particle fitness is calculated, and the optimal position of the particle group and the historical optimal position of the particle are updated. The iteration is continued until the magnesium alloy performance error objective function is minimized or the maximum number of iterations is reached. The test data and state data of the magnesium alloy composite material to be evaluated are input into the optimized magnesium alloy composite material performance evaluation model to obtain the magnesium alloy performance evaluation results.
7. A magnesium alloy composite material performance evaluation system based on big data, used to execute the method according to any one of claims 1 to 6, characterized in that: include: Data acquisition module: used to obtain test data and status data of magnesium alloy composite materials, and pre-process the test data and status data; Data processing module: used to construct the state influencing factor, to modify the macro material property to obtain the first material property, and to perform deviation analysis on the performance simulation data and the performance test data to obtain a deviation coefficient; Finite element module: used for constructing a microstructure finite element model according to the component test data to predict macroscopic material properties, and for performing multi-physics field finite element simulation according to the microstructure finite element model and the first material properties to obtain performance simulation data; Model module: used to construct a magnesium alloy composite material performance evaluation model, optimize the magnesium alloy composite material performance evaluation model according to the performance error, and input the test data and state data of the magnesium alloy composite material to be evaluated into the optimized magnesium alloy composite material performance evaluation model to obtain the magnesium alloy performance evaluation result; Intelligent supervision module: used for storing, viewing and managing the test data, status data and performance evaluation results of the magnesium alloy composite material, and displaying the performance evaluation results to the user.
Citation Information
Cited By
Metal powder impurity analysis method and system based on machine vision
CN120543532A
Material multi-scale performance optimization system based on thermodynamic statistical simulation
CN120727170A
