Collaborative inversion method for mechanical property parameters of plating layer and alloy layer of pre-plated metal plate

Abstract

A collaborative inversion method for mechanical property parameters of a plating layer and an alloy layer of a pre-plated metal plate comprises the following steps: after a nanoindentation experiment is implemented on the surface of the plating layer of the metal plate to obtain indentation three-dimensional contour data, generating a load-displacement curve and indentation morphology data under each parameter combination through three-dimensional simulation as a training set; and learning a nonlinear mapping relationship between the multi-dimensional features and the mechanical property parameters of the material by using a neural network, and then determining the mechanical property parameters of the microcell of the multi-layer structure through inversion calculation. According to the method, through multi-layer structure micro-area mechanical property inversion, the inversion precision of mechanical property parameters of the plating layer and the alloy layer can be improved, the problem of brother materials is effectively relieved, meanwhile, the calculation efficiency can be guaranteed, and a reliable solution is provided for mechanical property characterization of a multi-layer structure material.

Description

Technical Field

[0001] The present invention relates to a technology in the field of materials science, specifically a collaborative inversion method for mechanical property parameters of a coating and an alloy layer of a pre-plated metal sheet. Background Art

[0002] Some pre-plated metal sheets have micron-level (no more than 10μm) plating and alloy layers on their surface. The mechanical properties of these thin layers are crucial to ensuring the surface integrity and functionality of the sheet during subsequent plastic forming. Currently, the load-displacement curves obtained by nanoindentation experiments are commonly used to determine the micro-region mechanical properties. However, for pre-plated metal sheets, their surface is composed of extremely thin metal plating and alloy layers. Using nanoindentation technology, it is difficult to accurately locate and press the target thin layer. When performing nanoindentation experiments on the plating surface, the force applied by the indenter will cause the coordinated deformation of the underlying plating, alloy layer, and substrate material. This complex coupling effect will interfere with the measurement of the true mechanical properties of a single thin layer. Summary of the Invention

[0003] In response to the above-mentioned deficiencies in the prior art, the present invention proposes a collaborative inversion method for the mechanical property parameters of the coating and alloy layer of a pre-plated metal sheet. The method obtains load-displacement data by performing a nanoindentation experiment on the surface of the coating of the metal sheet; at the same time, a laser confocal microscope is used to obtain three-dimensional contour data of the indentation. These experimental data are used together as input for the inversion calculation. In addition, in order to construct a neural network training set, the present invention also generates load-displacement curves and indentation morphology data covering a variety of mechanical property parameter combinations through three-dimensional simulation. The neural network is used to learn the nonlinear mapping relationship between multi-dimensional features and the mechanical property parameters of the material, and finally the micro-region mechanical property parameters of the multi-layer structure are determined through inversion calculation. Through the inversion of the micro-region mechanical properties of the multi-layer structure, not only can the inversion accuracy of the mechanical property parameters of the coating and the alloy layer be improved, and the "brother material" problem can be effectively alleviated, but the computational efficiency can also be guaranteed to provide a reliable solution for the mechanical property characterization of multi-layer structural materials.

[0004] The present invention is achieved through the following technical solutions:

[0005] The present invention relates to a collaborative inversion method for the mechanical property parameters of the coating and alloy layer of a pre-plated metal sheet. After a nanoindentation experiment is carried out on the coating surface of the metal sheet to obtain three-dimensional indentation profile data, load-displacement curves and indentation morphology data under various parameter combinations are generated through three-dimensional simulation as a training set. A neural network is used to learn the nonlinear mapping relationship between multidimensional features and material mechanical property parameters, and then the micro-region mechanical property parameters of the multi-layer structure are determined through inversion calculation.

[0006] The experiment involved pre-treating the surface of the test material according to the national standard GB / T25898-2010 to ensure a clean sample surface. Two indentation tests at different depths, namely a calibration set and a validation set, were performed on the coating using a nanoindenter. After obtaining multiple sets of load-displacement curves, each indentation area was scanned using a laser confocal microscope to obtain high-resolution 3D indentation profile data.

[0007] The three-dimensional simulation refers to: after establishing an ABAQUS three-dimensional model based on the three-dimensional indentation profile data obtained during the calibration group nanoindentation experiment, setting multiple groups of mechanical property parameter combinations of the coating and alloy layers and completing the simulation of different parameter combinations through the ABAQUS solver, extracting the load-displacement curve and indentation morphology data under each parameter combination.

[0008] The training set includes: fitting the load-displacement curves in all experimental results and simulation results with loading curves and unloading curves to obtain characteristic parameters of the load-displacement curves, and calculating the key characteristic parameters of the indentation morphology from all experimental results and simulation results with indentation morphology data.

[0009] The characteristic parameters of the load-displacement curve include: loading section fitting coefficients a and b, contact stiffness S, residual depth ratio h f / h m and the maximum load P max ; Through the equation P = ah 2 +bh fitting loading section, extract coefficients a and b. For unloading curve, use the formula P=A(hh f ) m Fit the data before 5% unloading to calculate the contact stiffness S and residual depth ratio h f / h m , and extract the maximum load P max As a feature. Among them, h f is the final depth of the indentation after unloading is completed, h m It is the maximum penetration depth when the indenter applies the maximum load during the loading process. The calculation formula of contact stiffness is:

[0010] The key characteristic parameters of the indentation morphology include: area A, volume V, centroid position (X, Y) and maximum stacking height h max Specifically: the area is obtained by performing two-dimensional integration calculation on the profile data along the radial direction The total volume change of the indentation morphology is quantified by calculating the axisymmetric integration of the surface profile. The coordinate positions of the center of mass in the X and Y directions are determined by calculating the ratio of the first moment of the contour to the total area A. and and the highest point h of the accumulation around the indentation max =max(yy original_surface ).

[0011] The neural network learning refers to: taking the characteristic parameters of the load-displacement curve and the characteristic parameters of the indentation morphology in the simulation data as input, taking the corresponding mechanical property parameters of the coating and alloy layer as the target output, establishing a nonlinear mapping relationship between the nanoindentation test results and the mechanical properties of the coating and alloy layer based on the neural network, and completing the training of the neural network using the training set data.

[0012] The inverse calculation to determine the micro-region mechanical performance parameters of the multilayer structure specifically includes:

[0013] i) Parameter inversion: Based on the multi-feature fusion neural network trained with simulation data, the 10 characteristic parameters obtained after processing the experimental data of the calibration group are used as input to predict the mechanical properties of the coating and alloy layer and complete the inversion;

[0014] ii) Simulation Verification: The inverted mechanical property parameters of the coating and alloy layer micro-areas were input into the calibration group's three-dimensional multilayer structure nanoindentation simulation model established in ABAQUS. The nanoindentation simulation of the verification group was performed with an adjusted indentation depth to obtain the corresponding load-displacement curve and indentation morphology.

[0015] iii) Error analysis: Compare the simulation results of the validation group with the experimental results, calculate the errors and evaluate the reliability of the inversion results.

[0016] The present invention relates to a collaborative inversion system for mechanical property parameters of coatings and alloy layers of pre-plated metal sheets for realizing the above-mentioned method, comprising: a surface nano-indentation experiment and simulation database construction module, a multi-feature fusion neural network construction and training module, and a micro-area mechanical property parameter inversion and model verification module, wherein: the surface nano-indentation experiment and simulation database construction module establishes an ABAQUS three-dimensional model containing a coating, an alloy layer, and a substrate layer according to the surface nano-indentation experiment process, constructs different combinations of mechanical property parameters of coatings and alloy layers, performs batch simulation calculations, extracts characteristic parameters from the load-displacement curve and indentation profile curve obtained by simulation, and generates a data set required for neural network training in combination with the mechanical property parameters input by simulation; the multi-feature fusion neural network construction and training module ... , taking the mechanical performance parameters as the target output, the characteristic parameters of the load-displacement curve and the indentation profile curve as input, constructing a multi-feature fusion neural network, training, verifying and testing the neural network, and using the indentation profile features to introduce more constraints in the relationship between the nanoindentation results and the mechanical properties of the material, reducing the influence of "brother materials", and obtaining a nonlinear mapping relationship between the nanoindentation simulation results and the mechanical performance parameters of the coating and alloy layer; the micro-area mechanical performance parameter inversion and model verification module extracts characteristic parameters from the load-displacement curve and indentation profile curve experimental results based on the nanoindentation experimental results of the coating surface, and uses the trained multi-feature fusion neural network to input the experimental characteristic parameters, invert and obtain the mechanical performance parameters of the coating and alloy layer, and verify the reliability of the inversion results through simulation to evaluate the inversion accuracy. Technical Effects

[0017] The present invention addresses the problem that the mechanical properties of extremely thin coatings and alloy layers with a thickness of no more than 10 μm on the surface of pre-plated metal sheets cannot be directly measured through cross-sectional experiments. By using multi-feature extraction of load-displacement curves and indentation morphology data, the method can effectively reduce parameter coupling and the "brother material" problem, thereby optimizing the inversion accuracy of the mechanical performance parameters of the micro-regions of the coating and alloy layers. A nonlinear mapping relationship between multidimensional features and the mechanical properties of the micro-regions of the multi-layer structure is established through a neural network to achieve synchronous inversion of the mechanical performance parameters of the micro-regions of the coating and alloy layers, avoiding the error accumulation problem caused by layered inversion in traditional methods, and achieving the inversion of the mechanical performance parameters of the micro-region of the alloy layer, thereby providing a practical solution for the characterization of the mechanical properties of the micro-regions of multi-layer structural materials. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 Flowchart of the present invention;

[0019] Figure 2 Schematic diagram of the experimental load-displacement curve (calibration group and verification group);

[0020] Figure 3Schematic diagram of experimental indentation morphology curves (calibration group and verification group);

[0021] Figure 4 Schematic diagram of the 3D nanoindentation finite element model of metal plates with extremely thin coatings and alloy layers established based on ABAQUS;

[0022] Figure 5 Schematic diagram for extracting edge contours using the path method in finite element simulation results;

[0023] Figure 6 Schematic diagram of load-displacement curve feature extraction;

[0024] Figure 7 Schematic diagram of indentation profile curve feature extraction;

[0025] Figure 8 Schematic diagram of the multi-feature fusion neural network structure;

[0026] Figure 9 Schematic diagram of the comparison between the simulated load-displacement curve and indentation morphology curve based on the inversion parameters and the corresponding experimental data. DETAILED DESCRIPTION

[0027] like Figure 1 As shown in FIG. 1 , the collaborative inversion method for the mechanical property parameters of the coating and alloy layer of the pre-plated metal sheet in this embodiment is taken as an example. The pre-nickel-plated steel plate has an overall thickness of about 0.3 mm, wherein the thicknesses of the nickel layer and the alloy layer are 3.75 μm and 1.50 μm, respectively. The overall process includes:

[0028] Step 1: Construction of coating surface indentation experiment and simulation database, including:

[0029] 1.1 Specimen Preparation and Nanoindentation Experiment: According to the national standard GB / T25898-2010, the surface of the test material is pretreated to ensure that the sample surface is clean and meets the requirements of nanoindentation testing. First, wipe the coating surface with a non-abrasive solvent (such as isopropyl alcohol or acetone) to remove dust particles and impurities. Subsequently, a secondary cleaning with 95% ethanol is performed to further ensure that the surface is free of contamination and grease residue, completing the sample pretreatment.

[0030] 1.2 After the sample surface was pretreated, precise nanoindentation testing was performed using an advanced FT-I04 nanoindenter equipped with a standard Berkovich indenter. All experiments were conducted at room temperature. To comprehensively evaluate the mechanical properties of the material and build a reliable simulation database, two different indentation depth schemes were constructed, one for calibration and one for validation, and the load-depth curves and indentation topography curves were recorded for each.

[0031] Indentation depth scheme 1: The indentation depth was 1 μm. To ensure a stable and sufficient indentation process, the indenter indentation time was set to 20 seconds. This was recorded as the calibration group (subsequent 3D simulation modeling will be performed according to the experimental process of the calibration group to provide data for neural network model training). The load-displacement curve was obtained, and the indentation area was scanned using a VK-3000 laser confocal microscope at room temperature to obtain high-resolution morphological data.

[0032] Indentation Depth Scheme 2: Nanoindentation testing was performed at a greater indentation depth of 1.5 μm, designated the validation group. To accommodate the deeper indentation depth, the indenter insertion time was adjusted to 30 seconds. Similar to the calibration group experiments, the validation group also simultaneously recorded load-displacement curves and indentation topography data. These data will be used to independently verify the accuracy and generalizability of the inversion model trained using the calibration data.

[0033] like Figure 3 and Figure 4 Shown are the load-displacement curve and indentation morphology curve obtained.

[0034] 1.3 To simplify the calculation and ensure the reliability of the model when performing simulation modeling for the calibration group nanoindentation experiment, the following assumptions are made:

[0035] 1) During the indentation process, the influence of the indenter deformation is ignored and the indenter is regarded as an ideal rigid body;

[0036] 2) The metal film and substrate materials are both homogeneous and isotropic materials, and follow the von Mises yield criterion and isotropic hardening criterion under multiaxial stress state;

[0037] 3) Ignore the influence of interface deformation and assume that the metal film and substrate are perfectly bonded.

[0038] Based on the above assumptions, a suitable constitutive model is selected to describe the stress-strain relationship between the coating and the alloy layer. In this embodiment, the Hollomon stress-strain model is selected as a specific example. This model is suitable for uniform and isotropic materials. The stress-strain relationship of the Hollomon model is defined as: Where: i=c, a, s, respectively represent the metal coating, alloy layer and substrate material, E i is the elastic modulus, σ si is the yield strength, n i is the strain hardening exponent, ε pi is the equivalent plastic strain.

[0039] 1.4 Based on ABAQUS finite element analysis software, establish Figure 4The six-part symmetric 3D model of a Berkovich indenter shown here is used to simulate the load-displacement response and indentation topography during nanoindentation experiments. The model consists of a substrate, an alloy layer, and a coating. Displacement-controlled loading is employed, applying a predefined displacement load (D1) at the indenter reference point to simulate the indenter's motion during loading and unloading.

[0040] The loading and unloading processes are defined as independent analysis steps in the ABAQUS Step module to ensure precise control of the displacement path. The HistoryOutput function of the Step module outputs the contact reaction force (i.e., load) and displacement data of the indenter reference point for generating a load-displacement curve. Simultaneously, the FieldOutput function outputs the node coordinates and displacement information of the specimen surface for subsequent analysis of the indentation morphology. To improve simulation accuracy, a fine mesh is used in the indentation contact area to accurately capture local stress concentration and plastic deformation characteristics. Furthermore, the contact between the indenter and the specimen surface is achieved using a surface-to-surface contact model.

[0041] By consulting the data to solve the mechanical properties of the coating and alloy layer, the mechanical properties parameter range of the coating and alloy layer is set as follows: 100GPa≤E≤350GPa, 100MPa≤σ s ≤880MPa, 0.05≤n≤0.4; mechanical properties of the substrate are: E s =191GPa,σ ss =197MPa,n s =0.13;

[0042] Develop an automated process based on Python scripts to batch modify ABAQUS input files (inp files) to change the elastic-plastic constitutive parameters of the coating and alloy layers. First, prepare a basic inp file, which contains the geometric model of the multilayer structure, meshing, boundary conditions and loading steps. The material properties part defines the initial elastic-plastic parameters, such as elastic modulus E, yield strength σ s and the strain hardening exponent n. Next, the Python script reads the base inp file based on the pre-constructed parameter space, locates its material definition section (*Material), and replaces the corresponding layer's parameters with the values of the current combination. After each modification, the script generates a new inp file using a systematic naming convention and records the parameter combination information for subsequent traceability.

[0043] 1.5 Automated simulation in ABAQUS software through Python scripts: Use Python scripts to batch import inp files in ABAQUS and perform automated calculations of finite element models. Secondly, after the solution is completed, another set of Python scripts is used to post-process the generated odb files to complete batch data extraction. Specifically, the load-displacement curve data is directly extracted from the post-processing results, while the indentation morphology data is extracted by defining the path (Path) based on the following example. Figure 5 The edge contours of one-sixth of the model shown are extracted.

[0044] The above process effectively realizes the automated acquisition of large-scale simulation data and provides rich and diverse data samples for further construction of parameterized relational databases and neural network models.

[0045] Step 2: Multi-feature fusion neural network construction and training, specifically including:

[0046] 2.1 Load-displacement curve feature extraction: All experimental results of the calibration group and the validation group and the load-displacement curves in the simulation results corresponding to the calibration group test are processed. According to the definition of the characteristic parameters of the load-displacement curve in the invention content, the following are extracted: Figure 6 The key characteristic parameters shown are the nonlinear response coefficients a and b during loading, the residual depth ratio h f / h m , maximum load P max , contact stiffness S.

[0047] 2.2 Indentation profile feature extraction: All experimental results (including calibration group and validation group) and the corresponding calibration group test Figure 7 The indentation morphology data in the simulation results shown are analyzed. According to the definition of indentation morphology characteristic parameters in the invention content, the key characteristic parameters, namely area A, volume V, center of mass position X and Y, and maximum stacking height h, are extracted using MATLAB code. max .

[0048] 2.3 Build as Figure 8 The neural network shown in the figure includes: a total of 10 nodes, including an input layer of 5 load-depth curve characteristic parameters and 5 indentation profile characteristic parameters, a hidden layer containing multiple fully connected layers with the number of nodes being 512, 256, 128 and 64, and 1 node for predicting target values (such as E, σ s or n) output layer;.

[0049] The LeakyReLU activation function is used between each hidden layer to increase the nonlinear ability of the model;

[0050] 2.4 Network training: The Adam optimizer was used with an initial learning rate of 0.00001 to ensure smooth convergence during training. The weighted mean square error was used as the loss function. To prevent overfitting, an early stopping mechanism was adopted with a threshold of 200 epochs. If there was no significant improvement in the verification loss within 200 epochs, training was automatically stopped.

[0051] Step 3: Inversion of micro-area mechanical properties parameters and model verification, specifically including:

[0052] 3.1 Using the neural network trained based on simulation data, the 10 characteristic parameters extracted from the calibration group experimental data are used as input to output the mechanical properties of the coating and alloy layer, realizing the collaborative inversion of material parameters. The mechanical properties of the coating are obtained as the elastic modulus E c =162GPa, yield strength σ sc =400MPa, hardening index n c =0.095; the mechanical property parameter of the alloy layer is the elastic modulus E a =210GPa, yield strength σ sa =500MPa, hardening index n a =0.2.

[0053] 3.2 To verify the accuracy of the inversion results, the material parameters of the coating and alloy layer obtained by inversion were input into the original ABAQUS three-dimensional finite element model, and the indentation depth scheme 2 was used for simulation calculation to obtain the load-displacement curve and indentation profile curve of the verification group experiment. The simulation results were compared with the actual data measured at the same indentation depth. Figure 9 Detailed comparison shown.

[0054] 3.3 This example uses a multi-feature fusion neural network to collaboratively invert the material parameters of the coating and alloy layer. Based on the mechanical parameter results inverted in step 3.1, a finite element simulation of the nanoindentation experiment was performed. The simulated load-displacement curves and indentation profile curves were compared with the experimental curves, showing good agreement. This demonstrates that the parameters inverted using the multi-feature fusion method can well reproduce the experimental process.

[0055] For comparison, the inversion results of the mechanical properties of the coating and alloy layer using a single-feature neural network with only the load-displacement curve were examined. After training and validating the neural network using the same dataset with the indentation profile data removed, the experimentally obtained load-displacement curve was input, and different combinations with large numerical differences may be obtained. Two sets of parameters are taken as an example. The first set of parameters is: coating E c =260GPa, yield strength σ sc =120MPa, hardening index n c=0.28, the mechanical property parameter of the alloy layer is elastic modulus E a =118GPa, yield strength σ sa =240MPa, hardening index n a =0.09, using this set of parameters for simulation, the load-displacement curve and indentation profile curve obtained are less consistent with the experimental curve, and there are obvious differences. The second set of parameters is: coating elastic modulus E c =220GPa, yield strength σ sc =480MPa, hardening index n c =0.07, the mechanical property parameter of the alloy layer is elastic modulus E a =170GPa, yield strength σ sa =170MPa, hardening index n a = 0.21. Simulation results using this set of parameters also show that, while the agreement between the load-displacement curve and the indentation profile curves is somewhat less than that of the first set of comparison parameters, it is still lower than that obtained using the multi-feature fusion method. This comparison demonstrates that this method enables simultaneous and accurate inversion of the mechanical properties of the coating and alloy micro-regions. By integrating more information, this method effectively mitigates the coupling effects between parameters and the uncertainty caused by "sibling material" issues, significantly improving the accuracy and reliability of the inversion results and enabling simulations to better replicate experimental observations.

[0056] The above-mentioned specific implementation can be partially adjusted in different ways by those skilled in the art without departing from the principles and purpose of the present invention. The scope of protection of the present invention shall be based on the claims and shall not be limited by the above-mentioned specific implementation. All implementation schemes within its scope shall be subject to the constraints of the present invention.

Claims

1. A collaborative inversion method for mechanical property parameters of the coating and alloy layer of a pre-plated metal sheet, characterized in that: After conducting nanoindentation experiments on the coated surface of metal sheets to obtain three-dimensional indentation profile data, three-dimensional simulation is used to generate load-displacement curves and indentation morphology data under various parameter combinations as training sets. A neural network is used to learn the nonlinear mapping relationship between multidimensional features and material mechanical properties parameters, and then inversion calculations are used to determine the micro-area mechanical properties parameters of the multilayer structure.

2. The method for collaborative inversion of mechanical property parameters of the coating and alloy layer of a pre-plated metal sheet according to claim 1 is characterized in that: In the experiment, a nanoindenter was used to perform indentation tests at two different depths on the coating surface, namely a calibration group and a verification group. After obtaining multiple sets of load-displacement curves, each indentation area was scanned using a laser confocal microscope to obtain high-resolution three-dimensional indentation profile data.

3. The method for collaborative inversion of mechanical property parameters of the coating and alloy layer of a pre-plated metal sheet according to claim 1 is characterized in that: The three-dimensional simulation refers to: after establishing an ABAQUS three-dimensional model based on the three-dimensional indentation profile data obtained during the calibration group nanoindentation experiment, setting multiple groups of mechanical property parameter combinations of the coating and alloy layers and completing the simulation of different parameter combinations through the ABAQUS solver, extracting the load-displacement curve and indentation morphology data under each parameter combination.

4. The method for collaborative inversion of mechanical property parameters of the coating and alloy layer of a pre-plated metal sheet according to claim 1 is characterized in that: The training set includes: fitting the load-displacement curves in all experimental results and simulation results with loading curves and unloading curves to obtain characteristic parameters of the load-displacement curves, and calculating the key characteristic parameters of the indentation morphology from all experimental results and simulation results with indentation morphology data.

5. The method for collaborative inversion of mechanical property parameters of the coating and alloy layer of a pre-plated metal sheet according to claim 1 is characterized in that: The characteristic parameters of the load-displacement curve include: loading section fitting coefficients a and b, contact stiffness S, residual depth ratio h f / h m and the maximum load P max ; Through the equation P = ah 2 +bh fitting loading section, extract coefficients a and b, for unloading curve, use the formula P = A (hh f ) m Fit the data before 5% unloading to calculate the contact stiffness S and residual depth ratio h f / h m , and extract the maximum load P max As a feature, h f is the final depth of the indentation after unloading is completed, h m It is the maximum penetration depth when the indenter applies the maximum load during the loading process. The calculation formula of contact stiffness is: The key characteristic parameters of the indentation morphology include: area A, volume V, centroid position (X, Y) and maximum stacking height h max Specifically: the area is obtained by performing two-dimensional integration calculation on the profile data along the radial direction The total volume change of the indentation morphology is quantified by calculating the axisymmetric integration of the surface profile. The coordinate positions of the center of mass in the X and Y directions are determined by calculating the ratio of the first moment of the contour to the total area A. and and the highest point h of the accumulation around the indentation max =max(yy original_surface ).

6. The method for collaborative inversion of mechanical property parameters of the coating and alloy layer of a pre-plated metal sheet according to claim 1 is characterized in that: The neural network learning refers to: taking the characteristic parameters of the load-displacement curve and the characteristic parameters of the indentation morphology in the simulation data as input, taking the corresponding mechanical property parameters of the coating and alloy layer as the target output, establishing a nonlinear mapping relationship between the nanoindentation test results and the mechanical properties of the coating and alloy layer based on the neural network, and completing the training of the neural network using the training set data.

7. The method for collaborative inversion of mechanical property parameters of the coating and alloy layer of a pre-plated metal sheet according to claim 1 is characterized in that: The inverse calculation to determine the micro-region mechanical performance parameters of the multilayer structure specifically includes: i) Parameter inversion: Based on the multi-feature fusion neural network trained with simulation data, the 10 characteristic parameters obtained after processing the experimental data of the calibration group are used as input to predict the mechanical properties of the coating and alloy layer and complete the inversion; ii) Simulation Verification: The inverted mechanical property parameters of the coating and alloy layer micro-areas were input into the calibration group's three-dimensional multilayer structure nanoindentation simulation model established in ABAQUS. The nanoindentation simulation of the verification group was performed with an adjusted indentation depth to obtain the corresponding load-displacement curve and indentation morphology. iii) Error analysis: Compare the simulation results of the validation group with the experimental results, calculate the errors and evaluate the reliability of the inversion results.

8. A collaborative inversion system for mechanical property parameters of coating and alloy layer of pre-plated metal sheet material for implementing the method according to any one of claims 1 to 7, characterized in that: include: Surface nanoindentation experiment and simulation database construction module, multi-feature fusion neural network construction and training module, and micro-area mechanical property parameter inversion and model verification module, among which: the surface nanoindentation experiment and simulation database construction module establishes an ABAQUS three-dimensional model containing a coating, an alloy layer, and a substrate layer according to the results of the surface nanoindentation experiment, constructs different combinations of mechanical property parameters of the coating and alloy layers for batch simulation calculations, extracts characteristic parameters for the load-displacement curve and indentation profile curve obtained by simulation, and generates the data set required for neural network training by combining the mechanical property parameters input by simulation; the multi-feature fusion neural network construction module The construction and training module uses mechanical performance parameters as target outputs and characteristic parameters of load-displacement curve and indentation profile curve as inputs to construct a multi-feature fusion neural network and perform training, verification and testing to obtain the nonlinear mapping relationship between nanoindentation simulation results and mechanical performance parameters of coating and alloy layers; the micro-area mechanical performance parameter inversion and model verification module extracts characteristic parameters from the load-displacement curve and indentation profile curve experimental results based on the nanoindentation experimental results on the coating surface, and uses the trained multi-feature fusion neural network to invert the mechanical performance parameters of coating and alloy layers, and verifies the reliability of the inversion results through simulation to evaluate the inversion accuracy.

Images