Metal material processing technology window prediction system and method

Abstract

The invention discloses a metal material processing technology window prediction system and method. The method comprises the following steps: S1, constructing a fracture strain characteristic, an average unit strain plasticity characteristic and a Zener-Hollomon parameter logarithm characteristic; s2, constructing a continuous fracture probability prediction model; s3, constructing a fracture strain prediction model; s4, constructing a stress constitutive regression model; and S5, according to the stress constitutive regression model, obtaining a judgment result of fracture and instability in the metal material machining process. According to the method, under the condition of limited experimental points, smooth and physically consistent fracture and instability judgment results can be obtained, and continuous classification of machining parameter space is achieved; the prediction system can automatically identify a fracture area, an instability area and a safe area, and process window prediction and visualization of a full-parameter space are realized.

Description

Technical Field

[0001] This invention belongs to the field of material plastic processing and intelligent prediction technology, specifically relating to a metal material processing technology window prediction system and method. Background Technology

[0002] In actual industrial conditions, in order to evaluate forming performance and predict instability or fracture tendency, it is usually necessary to conduct thermal simulation experiments under multiple combinations of temperature and strain rate, measure the corresponding stress-strain curves, and record the fracture strain. Key parameters such as peak flow stress are also included. However, due to limitations in experimental equipment and material preparation conditions, these experimental data often only cover a limited and discrete range of temperature-strain rate points, resulting in sparse data distribution and noise. Furthermore, under certain extreme conditions (such as high-temperature low-speed or low-temperature high-speed deformation), the sample may not fracture or reach a stable flow stage, leading to incompleteness in the data set regarding fracture strain or stress response.

[0003] Fracture strain The variation of ductility is influenced by the material type, microstructure, deformation temperature, and strain rate, exhibiting significant nonlinearity and non-monotonicity. For example, as temperature increases, the ductility of some alloy systems initially increases and then decreases; while at high strain rates, localized temperature rise and strain concentration effects may lead to premature fracture. Therefore, this complex multi-factor coupling relationship makes it difficult for traditional linear or piecewise empirical models to accurately describe it. Distribution trend in the full parameter space.

[0004] The instability behavior of metallic materials during hot deformation (such as local necking and shear band formation) is often related to the strain rate sensitivity of flow stress, the thermal softening rate, and its derivatives. Since these data are derived from higher-order derivative calculations of stress-strain data, discrete experimental data are prone to amplifying errors during calculation, leading to increased uncertainty and volatility in instability determination. With the increasing complexity of forming conditions and the limited amount of data, maintaining the physical consistency and numerical stability of instability criteria under small sample conditions has become one of the core challenges in the field of metal plastic processing.

[0005] In the field of metal processing, existing experimental data are usually sparse and noisy. The criteria for fracture and instability depend on nonlinear and multivariate coupling relationships. Traditional methods (such as interpolation, fitting or single empirical models) are significantly lacking in accuracy, continuity and generalization ability, making it difficult to accurately, continuously and computably characterize the fracture and instability regions of the entire processing parameter space. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide a system and method for predicting the processing window of metal materials.

[0007] To solve the above-mentioned technical problems, the present invention adopts the following technical solution.

[0008] A method for predicting the processing window of metal materials includes the following steps: S1. Obtain the true stress-true strain curves of the metallic material under different temperature and strain rate conditions, extract the fracture strain, perform normalization processing, and construct fracture strain characteristics; based on the fracture strain, obtain the average unit strain plasticity, perform normalization processing, and construct the average unit strain plasticity characteristics; based on the strain rate and temperature, obtain the logarithmic characteristics of the Zener–Hollomon parameters. S2. Based on fracture strain characteristics, average unit strain plasticity characteristics, and logarithmic characteristics of the Zener–Hollomon parameter, a nonlinear combination function is used. φ The fracture probability is obtained, forming discrete temperature-strain rate-fracture probability data points; based on the temperature-strain rate-fracture probability data points, a continuous fracture probability prediction model is constructed, and the model parameters are obtained by nonlinear fitting of temperature and strain rate characteristics. S3. Construct a fracture strain prediction model based on the continuous fracture probability prediction model; S4. Based on the fracture strain prediction model, construct a stress constitutive regression model that is second-order continuously differentiable for true strain, strain rate, and temperature. S5. Based on the stress constitutive regression model, obtain the determination results of fracture and instability during the processing of metal materials.

[0009] The above-mentioned method for predicting the processing window of metal materials is further improved in step S1, where the fracture strain characteristics are as shown in equation (1). (1); in, Characteristic of fracture strain For fracture strain, To minimize the fracture strain, This represents the maximum fracture strain. The average unit strain plasticity characteristics are shown in Equation (2); (2); (3); (4); in, The average unit strain plastic properties, For plasticity integral, The average unit strain plasticity is given by ε, where ε is the true strain. For fracture strain, The minimum average unit strain plasticity, The maximum average unit strain plasticity; The logarithmic characteristics of the Zener–Hollomon parameters are shown in equation (5); log Z = ln Z (5); (6); Among them, log Z For the logarithmic characteristics of the Zener–Hollomon parameters, Z For the Zener–Hollomon parameter, For strain rate, Q As an apparent activation energy, R The gas constant is T For temperature, T The unit is K.

[0010] In a further improvement to the aforementioned method for predicting the processing window of metal materials, in step S2, the nonlinear combination function... φ This includes a weighted summation of fracture strain characteristics, average unit strain plasticity characteristics, and logarithmic characteristics of the Zener–Hollomon parameters, along with a parameterized combination of interaction terms and / or nonlinear terms; the logarithmic characteristics of the Zener–Hollomon parameters are subjected to a monotonically compressed function. After transformation, it participates in the combination, where, , These are parameters to be determined.

[0011] The above-mentioned method for predicting the processing window of metal materials is further improved in step S2, where the prediction model for the probability of continuous fracture is shown in equation (7): (7); (8); (9); (10); (11); (12); in, This represents the probability of continuous breakage. It is a temperature-dependent sigmoid function. The strain rate-dependent sigmoid function. The coefficients of the interaction term related to temperature and strain rate, The coefficients of the quadratic term related to temperature. The coefficients of the quadratic term related to the strain rate, To normalize the temperature characteristics, To normalize the strain rate characteristics, , , These are parameters related to temperature or strain rate. , Control the steepness and midpoint of the temperature direction. , Controlling the steepness and midpoint of the strain rate direction, where T is the temperature. For strain rate, This is the logarithm of the strain rate.

[0012] In a further improvement to the above-mentioned method for predicting the processing window of metal materials, in step S3, the fracture strain prediction model is as shown in equation (13); applying a strain to samples with high fracture probability. Constraints are in place to ensure conservatism. This represents the actual fracture strain; (13); in, To predict fracture strain, , The linear parameter of the fracture threshold. This represents the probability of continuous breakage.

[0013] In a further improvement to the aforementioned method for predicting the processing window of metallic materials, in step S3, the parameters k and b of the fracture strain prediction model are obtained by minimizing an objective function that includes a goodness-of-fit term and a constraint violation penalty term; the penalty term is used to suppress... Violation scenarios; in low fracture probability regions, adopt... The segmented form is used to ensure conservatism.

[0014] The above-mentioned method for predicting the processing window of metal materials is further improved in step S3, where the true strain ≥ If the true strain is less than 1, it is determined to be a fracture zone; It was determined to be a non-fracture zone.

[0015] The above-mentioned method for predicting the processing window of metal materials is further improved in step S4, where the stress constitutive regression model is shown in equations (14) to (16). (14); (15); (16); in, For stress, For strain rate sensitivity, The strain rate sensitivity derivative, Let ε be a second-order continuously differentiable activation function, and let ε be the true strain. Where is the strain rate and T is the temperature.

[0016] In a further improvement to the aforementioned method for predicting the processing window of metal materials, in step S5, if... <0, is considered an unstable region, if If the value is ≥0, it is considered a safe zone.

[0017] As a general technical concept, the present invention also provides a metal material processing window prediction system for performing the above-described metal material processing window prediction method, the metal material processing window prediction system comprising: The data preprocessing unit is used to obtain the true stress-true strain curves of metallic materials under different temperatures and strain rates, extract the fracture strain, perform normalization processing, and construct fracture strain characteristics; based on the fracture strain, obtain the average unit strain plasticity, perform normalization processing, and construct the average unit strain plasticity characteristics; and based on the strain rate and temperature, obtain the logarithmic characteristics of the Zener–Hollomon parameters. The fracture probability modeling unit, based on fracture strain characteristics, average unit strain plasticity characteristics, and the logarithmic characteristics of the Zener–Hollomon parameters, utilizes a nonlinear combination function. φ The fracture probability is obtained, forming discrete temperature-strain rate-fracture probability data points; a continuous fracture probability prediction model is constructed based on the temperature-strain rate-fracture probability data points, and the model parameters are obtained by nonlinear fitting of temperature and strain rate characteristics. The fracture strain prediction unit constructs a fracture strain prediction model based on the continuous fracture probability prediction model. A second-order continuously differentiable stress regression element is used to construct a stress constitutive regression model that is second-order continuously differentiable for true strain, strain rate, and temperature. The instability criterion calculation and region identification unit, based on the stress constitutive regression model, determines fracture and instability during the processing of metal materials; The process window generation unit outputs the determination result of the metal material processing technology.

[0018] Compared with the prior art, the advantages of the present invention are as follows: To address the problems of difficulty in identifying fracture zones, unstable zone determination, and limited computational efficiency and simulation scalability in existing metal processing window prediction methods that rely on interpolation or empirical formulas, this invention creatively proposes a metal material processing window prediction method. By constructing a continuous fracture probability prediction model, a highly conservative fracture probability function is generated under small sample conditions, ensuring that no real fractures are missed. By constructing a stress constitutive regression model, instability parameters such as strain rate sensitivity and strain rate sensitivity derivative can be directly calculated. Thus, fracture and instability determination are integrated to form a three-state process window of safe zone, unstable zone, and fracture zone, realizing the determination of fracture and instability during metal material processing. The method of this invention has the following advantages: a) It enables fracture prediction and instability identification under small sample conditions. By constructing a continuous fracture probability prediction model, it does not rely on large-scale experimental data and can form a continuously distributed fracture probability function under finite temperature and strain rate combinations. This function has smoothness and conservatism characteristics, which can effectively avoid misjudgment and omission in traditional interpolation methods under sparse sample conditions, and achieve reliable prediction of fracture zones; b) It significantly improves the continuity and differentiability of the model. A second-order continuously differentiable activation function is used to construct a stress constitutive regression model, which keeps the relationship between stress-strain-temperature-strain rate smooth in all dimensions. The automatic differentiation output of the first and second-order partial derivatives with respect to true strain, strain rate, and temperature provides a stable basis for calculating instability parameters such as strain rate sensitivity and strain rate sensitivity derivative, eliminating the problem of traditional differential noise amplification; c) A unified judgment mechanism for fracture and instability integrates the predicted fracture strain and instability criteria. With strain rate sensitivity derivative The invention establishes a three-state process window (fracture, instability, and safety) based on the joint judgment of the model. This process window has continuous, differentiable boundaries and clear physical meaning. It can be used for quantitative evaluation of experimental data and can be directly embedded into finite element simulation and process optimization processes to achieve consistency between data and simulation. d) It improves computational efficiency and scalability. The modeling framework of this invention has automatic differentiation capabilities, eliminating the need for repeated numerical difference and interpolation calls, significantly reducing computational load and error accumulation. At the same time, the model structure has good scalability, and the input features and constraints can be adjusted according to the material system or experimental design. It is applicable to various alloy systems and hot processing processes, such as hot rolling, hot forging, and hot extrusion. e) It enhances simulation adaptability and engineering practicality. The continuous fracture probability prediction model and stress constitutive regression model of this invention are both continuously differentiable and can be directly imported into numerical simulation software as material constitutive or instability criterion modules. At the same time, the three-state process window output by the model is clearly visualized and accurately classified, providing intuitive basis for process parameter optimization, defect prediction, and material selection, significantly improving the accuracy and efficiency of engineering design. The method of this invention can obtain smooth and physically consistent fracture and instability determination results under limited experimental point conditions, and realize continuous classification of processing parameter space, that is, identify fracture zone, instability zone and safe zone under multiple temperature, multiple strain rate and multiple strain conditions. It is applicable to material selection and window design of various hot / warm forming processes. Attached Figure Description

[0019] Figure 1 This is a flowchart of the metal material processing technology window prediction method in Embodiment 1 of the present invention.

[0020] Figure 2 This is a contour plot of the fracture probability with respect to temperature and strain rate in Embodiment 1 of the present invention.

[0021] Figure 3 This is a comparison diagram of the predicted fracture strain value and the experimental value in Embodiment 1 of the present invention.

[0022] Figure 4 This is a comparison chart of the predicted stress value and the experimental value in Embodiment 1 of the present invention.

[0023] Figure 5 This is a material processing diagram generated in a three-dimensional process space based on predicted fracture strain and instability criteria in Embodiment 1 of the present invention.

[0024] Figure 6 This is a diagram showing the proportion of different regions in the comprehensive processing interval generated based on predicted fracture strain and instability criteria in Embodiment 1 of the present invention. Detailed Implementation

[0025] The present invention will be further described below with reference to the accompanying drawings and specific preferred embodiments, but this does not limit the scope of protection of the present invention. All materials and instruments used in the following embodiments are commercially available.

[0026] Example 1: A method for predicting the processing window of metal materials according to the present invention, such as Figure 1 As shown, taking the hot compression test data of Mg-9Gd-4Y-Zr alloy as an example, the following steps are included: S1. Obtain the true stress-true strain curves of the metallic material under different temperature and strain rate conditions, extract the fracture strain, perform normalization processing, and construct the fracture strain characteristics; based on the fracture strain, obtain the average unit strain plastic property, perform normalization processing, and construct the average unit strain plastic property characteristics; based on the strain rate and temperature, obtain the logarithmic characteristics of the Zener–Hollomon parameters.

[0027] True stress and true strain data of Mg-9Gd-4Y-Zr alloy at different temperatures and strain rates were collected (598K~798K, 0.001 s⁻¹). -1 ~1 s -1 (A total of 20 groups) were used to clean up outliers and interpolate and smooth the strain at a uniform strain step size (Δε=0.001); fracture strain was identified based on the rapid stress decrease criterion. For the unbroken sample, the maximum strain is taken. As an upper limit, the plastic property integral PE and the average unit strain plastic property are calculated based on the fracture strain. Calculate the Zener–Hollomon parameters and logarithm. For fracture strain... Average unit strain plasticity The logarithm of the Zener–Hollomon parameter is normalized (e.g., by extreme value / standard or piecewise linear methods) to obtain the normalized characteristic fracture strain feature. Average unit strain plasticity characteristics The Zener–Hollomon parameter log Z is the output of a supervision signal used for the next step of modeling.

[0028] (1); in, Characteristic of fracture strain For fracture strain, To minimize the fracture strain, This represents the maximum fracture strain. (2); (3); (4); in, The average unit strain plasticity characteristic is given by PE, where PE is the plasticity integral. The average unit strain plasticity is given by ε, where ε is the true strain. For fracture strain, The minimum average unit strain plasticity, This represents the maximum average unit strain plasticity.

[0029] log Z = ln Z (5); (6); Among them, log Z The logarithmic characteristic of the Zener-Hollomon parameters is given by Z, where Z is the Zener-Hollomon parameter. R is the strain rate, Q is the apparent activation energy (Q = 200 kJ / mol in this embodiment), R is the gas constant, and T is the temperature, with the unit of T being K.

[0030] S2. Based on fracture strain characteristics, average unit strain plasticity characteristics, and logarithmic characteristics of the Zener–Hollomon parameter, a nonlinear combination function is used. φ The fracture probability is obtained, forming discrete temperature-strain rate-fracture probability data points; based on the temperature-strain rate-fracture probability data points, a continuous fracture probability prediction model is constructed, and the model parameters are obtained by nonlinear fitting of the temperature and strain rate characteristics.

[0031] The nonlinear combination function φ includes a weighted summation of fracture strain characteristics, average unit strain plasticity characteristics, logarithmic characteristics of the Zener–Hollomon parameters, and a parameterized combination of interaction terms and / or nonlinear terms. In this embodiment, the nonlinear combination function used to obtain the fracture probability is... ,in, Normalized characteristic fracture strain characteristics, The average unit strain plasticity characteristic is given by logZ, and logZ is the logarithmic characteristic of the Zener–Hollomon parameter. ,、 , The weights can be set by cross-validation or empirically. In this embodiment, the logarithmic features of the Zener–Hollomon parameters are subjected to a monotonic compression function. After transformation, it participates in combination.

[0032] Obtain discrete data points (T, , After that, with (T, ) as input, with To achieve this goal, a continuous fracture probability prediction model p(T, The preferred approach employs a parametric double Sigmoid structure (with smooth transition terms for both temperature and strain rate). Interaction and quadratic correction terms can be added to enhance the fitting flexibility, ultimately outputting a continuous fracture probability distribution, such as... Figure 2 The contour lines shown are schematic. Even with a small sample size, this model maintains a stable probability range, a trend consistent with the high risk associated with low temperature, high strain rate, and no significant numerical oscillations, thus meeting the requirements for subsequent conservative fracture strain threshold calculations and process window integration.

[0033] (7); (8); (9); (10); (11); (12); in, This represents the probability of continuous breakage. It is a temperature-dependent sigmoid function. The strain rate-dependent sigmoid function. The coefficients of the interaction term related to temperature and strain rate, The coefficients of the quadratic term related to temperature. The coefficients of the quadratic term related to the strain rate, To normalize the temperature characteristics, To normalize the strain rate characteristics, , , These are parameters related to temperature or strain rate. , Control the steepness and midpoint of the temperature direction. , Controlling the steepness and midpoint of the strain rate direction, where T is the temperature. Where SR_log is the strain rate, and SR_log is the logarithm of the strain rate. In this embodiment, after fitting, the following is obtained: , , , , , , The values ​​are -0.074, 0.029, 0.063, 0.018, 0.86, 579.414, and -4.999, respectively.

[0034] S3. Based on the continuous fracture probability prediction model, construct the fracture strain prediction model.

[0035] To provide a clear boundary for subsequent fracture determination, this invention proposes a conservative fracture strain prediction model. This model is based on the continuous fracture probability obtained in step S2. The fracture strain is predicted by calculating the following rules. Apply to samples with high fracture probability Constraints are in place to ensure conservatism. This represents the actual fracture strain.

[0036] (13); in, To predict fracture strain, , The linear parameter of the fracture threshold. This represents the probability of continuous breakage.

[0037] The determination of parameters k and b follows a key principle of conservatism: predicting the fracture strain under all experimental conditions. The initial fracture strain must be less than or equal to the experimentally observed actual fracture strain. Under this constraint, the objective function is optimized using a differential evolution algorithm (DE algorithm, population size 50, maximum number of iterations 100). Where μ = 100 is the penalty coefficient, the optimal values ​​of k and b are determined. In this embodiment, the optimized values ​​are k = 0.82 and b = 0.05.

[0038] Figure 3 This is a graph showing the comparison between the predicted and experimental values ​​of fracture strain in Example 1. Figure 3 There are a total of 20 samples. = 0.996, with the majority of data points located below the 1:1 diagonal, indicating that the prediction is conservative and safe.

[0039] In this invention, the parameters k and b of the fracture strain prediction model are obtained by minimizing an objective function that includes a goodness-of-fit term and a constraint violation penalty term; the penalty term is used to suppress... Violation scenarios; in low fracture probability regions, adopt... The segmented form is used to ensure conservatism.

[0040] S4. Based on the fracture strain prediction model, construct a stress constitutive regression model that is second-order continuously differentiable for true strain, strain rate, and temperature.

[0041] Using the fracture strain prediction model established in step S3, the three-dimensional process space (T, ...) generated and inherited in step S1 is analyzed. Interpolation / mesh data points are evaluated one by one (the mesh is not rebuilt, and the node system is completely reused from the preprocessing stage output). For any data point, it is determined whether the true strain ε under the corresponding conditions is greater than the predicted fracture strain. ;like If so, mark it as a "fracture zone" and remove it. If the data points are not broken, they are marked as "non-fractured regions". After filtering the nodes, 103,143 non-fractured region data points are obtained.

[0042] A second-order continuously differentiable stress constitutive regression model was constructed based on the non-fracture dataset (a total of 103,143 points), as shown in equations (14) to (16). A multi-layer feedforward neural network was adopted, with (ε, ln (T) is the input. For output; Network structure: Input layer 3 → Hidden layer 1 (64, Swish) → Hidden layer 2 (64, Swish) → Hidden layer 3 (32, Swish) → Output layer 1. The training / validation / test sets are divided in an 8:1:1 ratio (approximately: training ≈ 82514, validation ≈ 10314, test ≈ 10315, slight rounding error). The optimizer is Adam, with an initial learning rate of 0.001 and cosine annealing scheduling. Training is performed for 50,000 epochs, with regularization parameters λ1 = 0.001 and λ2 = 0.0001. Overall (all 103143 points) fit evaluation and... Figure 4 Consistent, overall R² = 0.999, overall RMSE = 1.679 MP. Gradient continuity test shows... No abnormal spikes across the entire domain, second derivative Smoothed; variance of 10 repeated automatic differentiation results <1e -6 The numerical stability was verified. The second-order differentiability of the model makes the subsequent instability parameters m, The calculation does not require a noise-amplified differential operation.

[0043] (14); (15); (16); in, For stress, For strain rate sensitivity, The strain rate sensitivity derivative, Let ε be a second-order continuously differentiable activation function, and let ε be the true strain. Where is the strain rate and T is the temperature.

[0044] S5. Based on the stress constitutive regression model, the prediction results of fracture and instability during the processing of metal materials are obtained.

[0045] Uniform / stratified sampling is performed on the temperature-strain rate-strain space to construct an evaluation grid of approximately 3000-4000 nodes (only for visualization and proportion statistics, without affecting the trained model and judgment rules). The continuous fracture probability prediction model in step S2, the fracture strain prediction model in step S3, and the stress constitutive regression model in step S4 are directly called for inference.

[0046] Calculation process: 1) Calculate each (T, according to the continuous fracture probability prediction model in step S2) The corresponding fracture probability.

[0047] 2) Obtain the predicted fracture strain threshold according to the fracture strain prediction model in step S3.

[0048] 3) Determine the true strain ε according to the stress constitutive regression model in S4; if the nodal strain If it is determined to be a fracture zone, the instability calculation is skipped; If it is, then it is marked as a "non-fracture zone".

[0049] 4) For nodes in the non-fracture zone, the stress constitutive regression model in S4 is used for determination, i.e., automatic differentiation to calculate m, ;like <0, is considered an unstable region, if If the value is ≥0, it is considered a safe zone.

[0050] 5) Summarize the three types of regions: fracture zone, instability zone, and safe zone, and generate... Figure 5 , Figure 6 . Figure 5 , Figure 6 In the comparison, a conventional interpolation method is set up.

[0051] Regional proportion: • Conventional interpolation: Safe zone 76.9%, unstable zone 19.2%, fracture zone 4.0% • In this embodiment: the safe zone is 67.9%, the unstable zone is 26.1%, and the fracture zone is 5.9%. The metal material processing window prediction method in this embodiment more fully identifies the unstable zone and fracture zone (increases by 6.9% for the unstable zone and 1.9% for the fracture zone), and the safe zone is more conservative (decreases by 9.0%), which can reduce the probability of misjudging risky working conditions.

[0052] The present invention also provides a metal material processing window prediction system for performing the above-described metal material processing window prediction method, the metal material processing window prediction system comprising: The data preprocessing unit is used to obtain the true stress-true strain curves of metallic materials under different temperatures and strain rates, extract the fracture strain, perform normalization processing, and construct fracture strain characteristics; based on the fracture strain, obtain the average unit strain plasticity, perform normalization processing, and construct the average unit strain plasticity characteristics; and based on the strain rate and temperature, obtain the logarithmic characteristics of the Zener–Hollomon parameters. The fracture probability modeling unit, based on fracture strain characteristics, average unit strain plasticity characteristics, and the logarithmic characteristics of the Zener–Hollomon parameters, utilizes a nonlinear combination function. φ The fracture probability is obtained, forming discrete temperature-strain rate-fracture probability data points; a continuous fracture probability prediction model is constructed based on the temperature-strain rate-fracture probability data points, and the model parameters are obtained by nonlinear fitting of temperature and strain rate characteristics. The fracture strain prediction unit constructs a fracture strain prediction model based on the continuous fracture probability prediction model. A second-order continuously differentiable stress regression element is used to construct a stress constitutive regression model that is second-order continuously differentiable for true strain, strain rate, and temperature. The instability criterion calculation and region identification unit, based on the stress constitutive regression model, determines fracture and instability during the processing of metal materials; The process window generation unit outputs the determination result of the metal material processing technology.

[0053] The metal material processing window prediction method of this invention is adaptable to small sample sizes and highly nonlinear conditions, is computationally efficient, and facilitates subsequent secondary development and simulation expansion. The metal material processing window prediction system of this invention can automatically identify fracture zones, instability zones, and safe zones under small to medium-sized or asymmetric sample experimental conditions, achieving full-parameter space process window prediction and visualization.

[0054] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make many possible variations and modifications to the technical solutions of the present invention using the methods and techniques disclosed above, or modify them into equivalent embodiments with equivalent changes, without departing from the spirit and technical essence of the present invention. Therefore, any simple modifications, equivalent substitutions, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solutions of the present invention shall still fall within the protection scope of the technical solutions of the present invention.

Claims

1. A method for predicting the processing window of metal materials, characterized in that, Includes the following steps: S1. Obtain the true stress-true strain curves of metallic materials under different temperatures and strain rates, extract the fracture strain, perform normalization processing, and construct fracture strain characteristics. Based on the fracture strain, the average unit strain plastic property is obtained, normalized, and the average unit strain plastic property characteristics are constructed; based on the strain rate and temperature, the logarithmic characteristics of the Zener–Hollomon parameters are obtained. S2. Based on fracture strain characteristics, average unit strain plasticity characteristics, and logarithmic characteristics of the Zener–Hollomon parameter, a nonlinear combination function is used. φ The fracture probability is obtained, forming discrete temperature-strain rate-fracture probability data points; based on the temperature-strain rate-fracture probability data points, a continuous fracture probability prediction model is constructed, and the model parameters are obtained by nonlinear fitting of temperature and strain rate characteristics. S3. Construct a fracture strain prediction model based on the continuous fracture probability prediction model; S4. Based on the fracture strain prediction model, construct a stress constitutive regression model that is second-order continuously differentiable for true strain, strain rate, and temperature. S5. Based on the stress constitutive regression model, obtain the determination results of fracture and instability during the processing of metal materials.

2. The method for predicting the processing window of metal materials according to claim 1, characterized in that, In step S1, the fracture strain characteristics are as shown in equation (1); （1）； in, Characteristic of fracture strain For fracture strain, To minimize the fracture strain, This represents the maximum fracture strain. The average unit strain plasticity characteristics are shown in Equation (2); （2）； （3）； （4）； in, The average unit strain plastic properties, For plasticity integral, The average unit strain plasticity is given by ε, where ε is the true strain. For fracture strain, The minimum average unit strain plasticity, The maximum average unit strain plasticity; The logarithmic characteristics of the Zener–Hollomon parameters are shown in equation (5); log Z = ln Z (5); （6）； Among them, log Z For the logarithmic characteristics of the Zener–Hollomon parameters, Z For the Zener–Hollomon parameter, For strain rate, Q As an apparent activation energy, R The gas constant is T For temperature, T The unit is K.

3. The method for predicting the processing window of metal materials according to claim 1, characterized in that, In step S2, the nonlinear combination function φ This includes a weighted summation of fracture strain characteristics, average unit strain plasticity characteristics, and logarithmic characteristics of the Zener–Hollomon parameters, along with a parameterized combination of interaction terms and / or nonlinear terms; the logarithmic characteristics of the Zener–Hollomon parameters are subjected to a monotonically compressed function. After transformation, it participates in the combination, where, , These are parameters to be determined.

4. The method for predicting the processing window of metal materials according to claim 1, characterized in that, In step S2, the continuous fracture probability prediction model is shown in equation (7); （7）； （8）； （9）； （10）； （11）； （12）； in, This represents the probability of continuous breakage. It is a temperature-dependent sigmoid function. The strain rate-dependent sigmoid function. The coefficients of the interaction term related to temperature and strain rate, The coefficients of the quadratic term related to temperature. The coefficients of the quadratic term related to the strain rate, To normalize the temperature characteristics, To normalize the strain rate characteristics, , , These are parameters related to temperature or strain rate. , Control the steepness and midpoint of the temperature direction. , Controlling the steepness and midpoint of the strain rate direction, where T is the temperature. For strain rate, This is the logarithm of the strain rate.

5. The method for predicting the processing window of metal materials according to claim 1, characterized in that, In step S3, the fracture strain prediction model is shown in equation (13); Apply to high fracture probability samples Constraints are in place to ensure conservatism. This represents the actual fracture strain; （13）； in, To predict fracture strain, , The linear parameter of the fracture threshold. This represents the probability of continuous breakage.

6. The method for predicting the processing window of metal materials according to claim 5, characterized in that, In step S3, the parameters k and b of the fracture strain prediction model are obtained by minimizing an objective function that includes a goodness-of-fit term and a constraint violation penalty term; the penalty term is used to suppress... The circumstances of the violation; In the low fracture probability region, adopt The segmented form is used to ensure conservatism.

7. The method for predicting the processing window of metal materials according to claim 5, characterized in that, In step S3, if the true strain ≥ If the true strain is less than 1, it is determined to be a fracture zone; It was determined to be a non-fracture zone.

8. The method for predicting the processing window of metal materials according to claim 1, characterized in that, In step S4, the stress constitutive regression model is shown in equations (14) to (16); （14）； （15）； （16）； in, For stress, For strain rate sensitivity, The strain rate sensitivity derivative, Let ε be a second-order continuously differentiable activation function, and let ε be the true strain. Where is the strain rate and T is the temperature.

9. The method for predicting the processing window of metal materials according to claim 8, characterized in that, In step S5, within the non-fracture zone, if <0, is considered an unstable region, if If the value is ≥0, it is considered a safe zone.

10. A metal material processing technology window prediction system, characterized in that, For performing the metal material processing window prediction method as described in any one of claims 1 to 9, the metal material processing window prediction system comprises: The data preprocessing unit is used to obtain the true stress-true strain curves of metallic materials under different temperatures and strain rates, extract the fracture strain, perform normalization processing, and construct fracture strain characteristics; based on the fracture strain, obtain the average unit strain plasticity, perform normalization processing, and construct the average unit strain plasticity characteristics; and based on the strain rate and temperature, obtain the logarithmic characteristics of the Zener–Hollomon parameters. The fracture probability modeling unit, based on fracture strain characteristics, average unit strain plasticity characteristics, and logarithmic characteristics of the Zener–Hollomon parameters, utilizes a nonlinear combination function. φ The fracture probability is obtained, forming discrete temperature-strain rate-fracture probability data points; a continuous fracture probability prediction model is constructed based on the temperature-strain rate-fracture probability data points, and the model parameters are obtained by nonlinear fitting of temperature and strain rate characteristics. The fracture strain prediction unit constructs a fracture strain prediction model based on the continuous fracture probability prediction model. A second-order continuously differentiable stress regression element is used to construct a stress constitutive regression model that is second-order continuously differentiable for true strain, strain rate, and temperature. The instability criterion calculation and region identification unit, based on the stress constitutive regression model, determines fracture and instability during the processing of metal materials; The process window generation unit outputs the determination result of the metal material processing technology.

Images