Physical vapor deposition (PVD) coating design quality control method based on machine learning

Abstract

The invention relates to the technical field of PVD coating, in particular to a PVD coating design quality control method based on machine learning. The method comprises the following steps: obtaining a coating substrate object, carrying out PVD coating design, and carrying out substrate PVD coating simulation based on the design to obtain related simulation data; analyzing mechanical property parameters by using the PVD coating simulation data of the base material, evaluating the mechanical property of the coating material in combination with the mechanical data, and generating dynamic stress response data of the coating layer through a stability test; evaluating based on the chemical stability of the base material coating film, and predicting the failure condition of the coating film layer by combining the dynamic stress response data of the coating film layer; and through comprehensive analysis of dynamic stress response data and failure data of the coating layer, the coating design is further optimized. According to the invention, through the quality control of PVD coating, the quality of the PVD coating is more perfect.

Description

Technical Field

[0001] The present invention relates to the field of PVD coating technology, and in particular to a quality control method for PVD coating design based on machine learning. Background Art

[0002] Physical vapor deposition (PVD) technology is a material coating technology widely used in the field of surface engineering, especially in products in the fields of machinery, electrical appliances, electronics, mechanics, metals and cemented carbide. PVD technology forms a thin film by evaporating or sputtering solid materials onto the surface of the substrate, which has the advantages of high precision, high adhesion, good uniformity and adjustability. The required film properties are obtained by manually controlling process parameters (such as temperature, target-substrate distance, power, gas pressure, deposition rate, etc.). Although this method can guarantee product quality to a certain extent, it has many limitations. Factors such as complex film properties, fluctuations in the external environment, and stability of the equipment will lead to unstable film quality. Machine learning, as a technology that extracts rules and makes predictions from data analysis, automatically adjusts and optimizes coating process parameters based on a large amount of historical data to achieve more efficient and stable film quality control. However, the existing PVD coating design and quality control technology has the problem of inaccurate analysis of the surface characteristics of the coating substrate and inaccurate analysis of the stability of the coating layer. Summary of the invention

[0003] Based on this, it is necessary to provide a quality control method for PVD coating design based on machine learning to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a quality control method for PVD coating design based on machine learning includes the following steps:

[0005] Step S1: obtaining a coating substrate object; performing a PVD coating design according to the coating substrate object, and performing a substrate PVD coating simulation based on the substrate PVD coating design to obtain substrate PVD coating simulation data;

[0006] Step S2: Analyze the mechanical coupling behavior data of the substrate PVD coating simulation data, and analyze the multi-scale mechanical properties of the coating material based on the substrate PVD coating design and the mechanical coupling behavior data, and use the multi-scale mechanical properties of the coating material to perform a dynamic stress response test of the coating layer to generate dynamic stress response data of the coating layer;

[0007] Step S3: evaluating the chemical stability of the substrate coating based on the substrate PVD coating simulation data, and predicting the failure of the substrate coating layer based on the chemical stability of the substrate coating and the dynamic stress response data of the coating layer, to obtain the failure data of the substrate coating layer;

[0008] Step S4: performing a substrate coating quality analysis based on the coating layer dynamic stress response data and the substrate coating layer failure data, and performing substrate coating design optimization on the substrate coating quality to obtain substrate coating design optimization data.

[0009] The present invention is based on PVD (physical vapor deposition) coating technology, and the step design of the present invention can significantly improve the quality control and optimization process of substrate coating. In step S1, by obtaining the coating substrate object and performing PVD coating design, combined with the simulation of substrate PVD coating and performance influencing factors, the performance changes that occur during the substrate coating process can be accurately predicted. The substrate PVD coating simulation data provides a reliable basis for subsequent quality evaluation and design optimization, can achieve precise control of the coating process, avoid the influence of uncontrollable factors on the coating quality in traditional processes, and improve the performance and consistency of the coating layer. In step S2, by performing mechanical property analysis on the substrate PVD coating simulation data, and combining the mechanical property evaluation of the coating material, the stability of the coating layer can be effectively monitored. This process not only provides data support for the subsequent coating layer stability test, but also helps to identify the performance defects of the coating layer, thereby providing a basis for the optimization of the coating layer. The coating layer stability test further provides a reliable basis for subsequent performance estimation and failure analysis, ensuring the stability and high quality of the substrate coating layer. In step S3, by performing failure prediction based on mechanical and chemical stability evaluation and coating stability data, the potential failure areas of the coating during use are effectively identified. This analysis process helps to predict the service life of the coating in advance and provides data support for formulating appropriate maintenance strategies and improving processes. Failure prediction can not only reduce resource waste, but also implement timely adjustments and repairs before the coating fails, avoiding product performance degradation or failure, and ensuring product reliability and long-term stability. The quality analysis of the substrate coating based on the stability data and failure prediction data of the coating material can comprehensively evaluate the quality of the coating and guide subsequent design optimization. Coating design optimization can accurately adjust the process parameters, thereby achieving a higher level of quality control during the coating process and reducing the probability of defects. The optimized design data not only improves the performance and reliability of the coating, but also reduces unnecessary costs and time losses caused by quality defects during the production process. Therefore, the present invention optimizes the traditional quality control method of PVD coating design based on machine learning, solves the problem of inaccurate analysis of the surface characteristics of the coating substrate and inaccurate analysis of the stability of the coating layer in the traditional quality control method of PVD coating design based on machine learning, and improves the accuracy of the analysis of the surface characteristics of the coating substrate and the accuracy of the analysis of the stability of the coating layer.

[0010] Preferably, step S1 comprises the following steps:

[0011] Step S11: obtaining a coating substrate object;

[0012] Step S12: performing surface quality detection of the coated substrate according to the coated substrate object, thereby obtaining surface quality data of the coated substrate;

[0013] Step S13: performing a coating requirement analysis of the coating substrate object based on the coating substrate surface quality data and the coating substrate object to obtain the coating requirement of the substrate object;

[0014] Step S14: performing PVD coating design according to the coating requirements of the substrate object and the surface quality data of the coating substrate to obtain substrate PVD coating design data;

[0015] Step S15: performing substrate PVD coating simulation based on the substrate PVD coating design data to obtain substrate PVD coating simulation data.

[0016] The present invention optimizes the design and quality control in the coating process through a series of precise steps. On the basis of obtaining the coating substrate object, surface quality detection provides important input for subsequent coating demand analysis, ensuring a comprehensive understanding of the surface state of the substrate, so as to achieve targeted coating solutions. By combining surface quality data with substrate objects, the coating requirements of the substrate can be clarified, and then accurate parameter support can be provided for PVD coating design, making the design process highly adaptable and targeted. The simulation process based on design data can predict the coating effect and potential performance problems in advance, providing sufficient basis for actual production, and avoiding later rework and quality problems. The overall process improves the accuracy and efficiency of coating design, optimizes process parameters, ensures the quality and performance of the coating layer, minimizes resource waste and production costs, and achieves green production of coatings.

[0017] Preferably, step S13 comprises the following steps:

[0018] Step S131: extracting the geometric structure of the coated substrate according to the coated substrate object, thereby obtaining the geometric structure data of the coated substrate;

[0019] Step S132: analyzing the surface characteristics of the coating substrate according to the surface quality data of the coating substrate to obtain surface characteristic data of the coating substrate;

[0020] Step S133: dividing the coated substrate based on the geometric structure data of the coated substrate to obtain a flat substrate and a curved substrate;

[0021] Step S134: calculating the substrate coating adhesion ability according to the coating substrate surface characteristic data to obtain the substrate coating adhesion ability data;

[0022] Step S135: estimating the difficulty of coating the flat substrate according to the substrate coating adhesion ability data to obtain the difficulty data of coating the flat substrate;

[0023] Step S136: performing a curved surface substrate coating complexity analysis on the curved surface substrate according to the substrate coating adhesion ability data to obtain curved surface substrate coating complexity data;

[0024] Step S137: performing coating requirement analysis on the coating substrate object according to the coating complexity data of the curved substrate and the coating difficulty data of the flat substrate to obtain the coating requirement of the substrate object.

[0025] The present invention provides detailed basic data for subsequent coating design by accurately extracting the geometric structure of the coating substrate and analyzing the surface quality. The extraction of the geometric structure not only clarifies the morphological characteristics of the substrate, but also provides data support for the subsequent substrate division, so that the different properties of the planar substrate and the curved substrate can be accurately distinguished. Through surface characteristic analysis, combined with the adhesion ability evaluation of the substrate, the coating difficulty and complexity analysis of the flat and curved substrates is further optimized. This process ensures a more accurate estimate of the coating difficulty, thereby effectively reducing quality problems caused by uneven coating or insufficient adhesion. The coating demand analysis based on these data enables the coating scheme to be optimized according to the characteristics of different substrates, improves the predictability and reliability of the coating effect, and ensures the stability and efficiency of the coating process.

[0026] Preferably, step S136 includes the following steps:

[0027] Step S1361: extracting irregularities of the curved surface substrate according to the curved surface substrate to obtain irregularities of the curved surface substrate;

[0028] Step S1362: Calculating the curvature of the curved substrate according to the irregularity of the curved substrate to obtain the curvature of the curved substrate;

[0029] Step S1363: performing differential estimation of the deposition angle of the curved substrate based on the curvature and irregularity of the curved substrate to obtain differential deposition angle data;

[0030] Step S1364: performing substrate sputtering deposition distribution difference detection according to the irregularity of the curved substrate to obtain substrate sputtering deposition distribution difference data;

[0031] Step S1365: Calculating the heterogeneity of deposition rate on the substrate surface based on the substrate sputtering deposition distribution difference data and the deposition angle difference data to obtain the heterogeneity of deposition rate on the substrate surface;

[0032] Step S1366: Performing a curved surface substrate coating complexity analysis on the substrate surface deposition rate heterogeneity according to the substrate coating adhesion ability data to obtain the curved surface substrate coating complexity data.

[0033] The present invention can accurately identify the morphological features of the curved substrate by extracting the irregularity of the curved substrate and calculating the curvature, thereby providing key data for the subsequent coating design. The combination of the irregularity and curvature information of the curved substrate can predict the difference in deposition angles, thereby optimizing the coating effects of different areas during the coating process. This process ensures uniform deposition of the coating material and avoids the problem of uneven coating caused by angle differences. The detection of differences in the distribution of substrate sputtering deposition effectively reveals the unevenness of the airflow during the deposition process, and combined with the calculation of the heterogeneity of the sedimentation velocity, provides a quantitative basis for the difference between the airflow and the deposition velocity during the actual deposition process, thereby reducing the errors generated during the deposition process. The analysis of the coating complexity of the curved substrate combined with the coating adhesion ability data can accurately predict and optimize the coating scheme of the curved substrate, effectively improve the stability and quality of the coating, and ensure that the coating quality of the curved substrate is guaranteed under complex conditions.

[0034] Preferably, the mechanical coupling behavior analysis in step S2 includes:

[0035] Conducting a coating material surface rigidity test on the substrate PVD coating simulation data to obtain coating material surface rigidity data;

[0036] Calculate the friction coefficient of the coating material surface using the coating material surface rigidity data;

[0037] Evaluate the wear resistance of the coating material based on the friction coefficient of the coating material surface;

[0038] The wear state of the coating layer is estimated based on the surface friction coefficient of the coating material and the wear resistance of the coating material to obtain the wear state data of the coating layer;

[0039] Use nano-indentation technology to test the fracture toughness of the coating material by measuring the surface rigidity of the coating material;

[0040] Evaluate the external load capacity of the coating layer based on the fracture toughness of the coating material;

[0041] Based on the external load capacity of the coating layer and the wear state data of the coating layer, a mechanical coupling behavior analysis is performed to obtain the mechanical coupling behavior data.

[0042] The present invention can quantify the lubricity and anti-friction properties of the coating by calculating the surface friction coefficient, thereby evaluating the wear resistance of the coating. The wear resistance data can intuitively reflect the loss of the coating under long-term friction. Combined with the friction coefficient, the wear state of the coating layer can be accurately predicted, and potential wear modes and failure risks can be identified in advance. Nanoindentation technology can accurately measure the fracture toughness to ensure that the coating has sufficient crack resistance and avoid brittle failure under high load or impact conditions. By evaluating the external load capacity of the coating layer based on the fracture toughness, it is possible to determine the limit of the coating's ability to withstand mechanical loads and ensure its stability in complex stress environments. By performing mechanical coupling behavior analysis on the external load capacity and wear state data of the coating layer, the mechanical response characteristics of the coating under different loads and friction conditions can be revealed, providing data support for optimizing the coating process and improving the service life and performance stability of the coating.

[0043] Preferably, the multi-scale mechanical property analysis of the coating layer in step S2 includes:

[0044] The thickness of the substrate coating layer is calculated using the substrate PVD coating design to obtain the thickness of the substrate coating layer;

[0045] The substrate PVD coating design is used to analyze the microscopic particles of the substrate coating layer and obtain the microscopic particle data of the substrate coating layer;

[0046] Extracting the microscopic particle distribution characteristics of the substrate in the coating layer of the substrate with a particle size distribution ranging from 10nm to 5000nm;

[0047] The surface hardness of the coating layer to estimate the microscopic particle characteristics of the coating layer;

[0048] The thickness of the substrate coating layer is measured with a measurement interval of 0.1-100 μm to obtain the thickness difference of the substrate coating layer;

[0049] The stress concentration area in the coating layer is calculated by using the thickness difference of the substrate coating layer and the microscopic particle distribution characteristics of the substrate;

[0050] Predict the impact resistance of the coating layer based on the mechanical coupling behavior data and the stress concentration area in the coating layer;

[0051] The multi-scale mechanical properties of the coating layer are analyzed based on the surface hardness of the coating layer and the impact resistance of the coating layer to generate the multi-scale mechanical properties of the coating layer.

[0052] The present invention calculates the thickness of the coating layer through the substrate PVD coating design, and can accurately determine the thickness distribution of the coating layer, providing basic data for further analyzing the mechanical properties of the coating layer. Based on the analysis of microscopic particle data of the substrate PVD coating design, the microstructural characteristics inside the coating layer, especially the size distribution of the particles, can be revealed, which is crucial to evaluating the physical and mechanical properties of the coating layer. By extracting the particle size distribution characteristics in the range of 10nm to 5000nm, the microstructure of the coating layer can be understood in more detail, and the surface hardness of the coating layer can be inferred. The thickness difference of the substrate coating layer is measured at a spacing of 0.1-100μm, and the uniformity and local differences of the coating layer can be quantified, which is crucial to optimizing the coating process. Through the combination of thickness difference and microscopic particle distribution characteristics, the stress concentration area in the coating layer can be effectively calculated, providing a basis for analyzing the potential weaknesses of the coating layer. At the same time, the prediction of the impact resistance of these stress concentration areas and mechanical coupling behavior data helps to identify the failure area of ​​the coating layer in advance and optimize material selection and process parameters. Finally, by integrating the surface hardness and impact resistance of the coating layer for a comprehensive evaluation, we can fully understand the mechanical properties of the coating material, provide a reliable basis for the application of the material, and ensure that the coated product has sufficient durability and stability in the actual environment.

[0053] Preferably, the dynamic stress response test of the coating layer in step S2 includes:

[0054] The tensile rate of 1 mm / min was used to simulate the multi-scale mechanical properties of the coating layer and obtain the tensile data of the coating layer;

[0055] Calculate the deformation degree of the coating layer from the tensile data of the coating layer;

[0056] Calculate the probability of cracks in the coating layer when the deformation degree of the coating layer is 50nm;

[0057] The elongation at break of the coating layer is estimated according to the deformation degree of the coating layer and the crack probability of the coating layer under the stress of 10-500MPa, and the elongation at break of the coating layer is obtained;

[0058] Analyze the coating layer strength of the coating layer elongation at break;

[0059] The dynamic stress response test of the coating layer is carried out based on the strength of the coating layer and the elongation at break of the coating layer to obtain the dynamic stress response data of the coating layer.

[0060] The present invention can simulate the deformation behavior of the coating layer under stress during actual use by performing a tensile simulation on the mechanical properties of the coating material, and obtain the performance data of the coating layer under different tensile conditions. By calculating the degree of deformation of the coating layer, the deformation of the coating layer during the stretching process can be quantified, reflecting the flexibility and deformability of the material. This data helps to further infer the stress response of the coating layer under different usage environments and to pre-evaluate the material. Based on the degree of deformation, the crack probability is calculated, the failure risk of the coating layer under different conditions is predicted, and potential crack development areas are identified. Combining the analysis of the degree of deformation and the crack probability, the elongation at break of the coating layer can be estimated, which is crucial for evaluating the tensile resistance of the coating layer and the fracture performance of the material. Through further analysis of the elongation at break of the coating layer, the strength data of the coating layer can be obtained, providing an important reference for material selection and process adjustment.

[0061] Preferably, step S3 comprises the following steps:

[0062] Step S31: performing substrate coating material detection on substrate PVD coating simulation data to obtain substrate coating material data;

[0063] Step S32: evaluating the chemical stability of the substrate coating according to the substrate coating material data to obtain the chemical stability of the substrate coating;

[0064] Step S33: performing corrosion resistance test on the substrate coating according to the chemical stability of the substrate coating to obtain corrosion resistance data of the substrate coating;

[0065] Step S34: performing failure prediction of the substrate coating layer based on the corrosion resistance data of the substrate coating layer and the dynamic stress response data of the coating layer to obtain failure data of the substrate coating layer.

[0066] The present invention detects the substrate PVD coating simulation data and can comprehensively obtain various physical, chemical and mechanical performance data of the coating material, providing basic information for subsequent evaluation and optimization. By analyzing the substrate coating material data, the chemical stability of the substrate can be accurately evaluated to ensure that the coating layer will not degrade due to chemical reactions under different environmental conditions, thereby extending the service life. Corrosion resistance tests are carried out based on chemical stability data to further verify the performance of the coating material in a corrosive environment, help predict the corrosion resistance of the coating layer, and provide a quantitative basis for material optimization and selection. Combined with the coating layer stability data, the failure of the substrate coating layer can be accurately predicted, and its failure mode under different application conditions can be predicted to ensure the safety and reliability of the product in use.

[0067] Preferably, step S34 includes the following steps:

[0068] Step S341: performing external impact simulation according to the dynamic stress response data of the coating layer to obtain external impact simulation data of the coating layer;

[0069] Step S342: estimating the degradation fatigue trend of the coating layer based on the external impact simulation data of the coating layer to obtain degradation fatigue trend data of the coating layer;

[0070] Step S343: performing corrosion environment simulation according to the corrosion resistance data of the substrate coating to obtain corrosion environment simulation data of the coating layer;

[0071] Step S344: performing a corrosion extension estimation of the coating layer on the corrosion resistance data of the substrate coating based on the corrosion environment simulation data of the coating layer to obtain corrosion extension data of the coating layer;

[0072] Step S345: estimating the coating layer shedding state according to the coating layer corrosion extension data and the coating layer degradation fatigue trend data to obtain coating layer shedding state data;

[0073] Step S346: performing a failure estimation of the substrate coating layer according to the coating layer peeling state data and the coating layer corrosion extension data to obtain substrate coating layer failure data.

[0074] The present invention uses the dynamic stress response data of the coating layer for external impact simulation, accurately reflects the stress distribution and damage evolution of the coating under different impact conditions, and ensures that the coating has good impact resistance. Degradation fatigue trend estimation based on external impact simulation data can identify the fatigue accumulation effect of the coating under long-term periodic loads, and detect micro crack extension and performance degradation in advance. The corrosion resistance data of the coating is used for corrosion environment simulation, which can reproduce the degradation behavior of the coating under different corrosive media and environmental conditions, and quantify the impact of corrosion on the stability of the coating. Corrosion expansion estimation based on corrosion environment simulation data can reveal the development trend of corrosion from local pitting to comprehensive corrosion, and evaluate the corrosion resistance and service life of the coating. The shedding state is estimated by combining the corrosion expansion data of the coating layer and the degradation fatigue trend data, and the degree of peeling of the coating under the combined action of mechanical fatigue and corrosion is quantified, providing a basis for optimizing the coating structure.

[0075] Preferably, step S4 comprises the following steps:

[0076] Step S41: evaluating the quality of the substrate coating based on the dynamic stress response data of the coating layer and the failure data of the substrate coating layer to obtain the substrate coating quality data;

[0077] Step S42: performing a substrate coating quality defect analysis according to the substrate coating quality data to obtain substrate coating quality defect data;

[0078] Step S43: Optimizing the substrate coating design based on the substrate coating quality defect data based on machine learning to obtain substrate coating design optimization data.

[0079] The present invention performs quality assessment based on the stability data of the coating material and the failure data of the substrate coating layer, accurately identifies the quality problems and their impacts that occur during the coating process, and provides data support for further improving the coating technology. By performing defect analysis on the substrate coating quality data, various defects that occur during the coating process can be accurately identified, and a basis can be provided for defect root cause analysis to ensure that process parameters are adjusted in time during the coating process. Based on machine learning, design optimization of coating quality defect data can be performed, and a large amount of data can be intelligently analyzed through algorithms to optimize the coating design scheme, making the coating effect more uniform and stable, improving the overall quality of the substrate coating, and reducing the probability of failure, thereby improving the long-term performance and reliability of the product.

[0080] The present invention is that, based on PVD (physical vapor deposition) coating technology, the step design of the present invention can significantly improve the quality control and optimization process of substrate coating. In step S1, by obtaining the coating substrate object and performing PVD coating design, combined with the simulation of substrate PVD coating and performance influencing factors, the performance changes occurring in the substrate coating process can be accurately predicted. The substrate PVD coating simulation data provides a reliable basis for subsequent quality evaluation and design optimization, can achieve precise control of the coating process, avoid the influence of uncontrollable factors on the coating quality in traditional processes, and improve the performance and consistency of the coating layer. In step S2, by performing mechanical property analysis on the substrate PVD coating simulation data, combined with the mechanical property evaluation of the coating material, the stability of the coating layer can be effectively monitored. This process not only provides data support for the subsequent coating layer stability test, but also helps to identify the performance defects of the coating layer, thereby providing a basis for the optimization of the coating layer. The coating layer stability test further provides a reliable basis for subsequent performance estimation and failure analysis, ensuring the stability and high quality of the substrate coating layer. In step S3, by performing failure prediction based on mechanical and chemical stability evaluation and coating stability data, the potential failure areas of the coating during use are effectively identified. This analysis process helps to predict the service life of the coating in advance and provides data support for formulating appropriate maintenance strategies and improving processes. Failure prediction can not only reduce resource waste, but also implement timely adjustments and repairs before the coating fails, avoiding product performance degradation or failure, and ensuring product reliability and long-term stability. The quality analysis of the substrate coating based on the stability data and failure prediction data of the coating material can comprehensively evaluate the quality of the coating and guide subsequent design optimization. Coating design optimization can accurately adjust the process parameters, thereby achieving a higher level of quality control during the coating process and reducing the probability of defects. The optimized design data not only improves the performance and reliability of the coating, but also reduces unnecessary costs and time losses caused by quality defects during the production process. Therefore, the present invention optimizes the traditional quality control method of PVD coating design based on machine learning, solves the problem of inaccurate analysis of the surface characteristics of the coating substrate and inaccurate analysis of the stability of the coating layer in the traditional quality control method of PVD coating design based on machine learning, and improves the accuracy of the analysis of the surface characteristics of the coating substrate and the accuracy of the analysis of the stability of the coating layer. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 A schematic flow chart of the steps of a quality control method for PVD coating design based on machine learning;

[0082] Figure 2 for Figure 1 Detailed implementation steps of step S3 in FIG.

[0083] Figure 3 for Figure 2 Detailed implementation steps of step S34 are shown in the flowchart;

[0084] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0085] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0086] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0087] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0088] To achieve this, please refer to Figures 1 to 3 , a quality control method for PVD coating design based on machine learning, comprising the following steps:

[0089] Step S1: obtaining a coating substrate object; performing a PVD coating design according to the coating substrate object, and performing a substrate PVD coating simulation based on the substrate PVD coating design to obtain substrate PVD coating simulation data;

[0090] Step S2: Analyze the mechanical coupling behavior data of the substrate PVD coating simulation data, and analyze the multi-scale mechanical properties of the coating material based on the substrate PVD coating design and the mechanical coupling behavior data, and use the multi-scale mechanical properties of the coating material to perform a dynamic stress response test of the coating layer to generate dynamic stress response data of the coating layer;

[0091] Step S3: evaluating the chemical stability of the substrate coating based on the substrate PVD coating simulation data, and predicting the failure of the substrate coating layer based on the chemical stability of the substrate coating and the dynamic stress response data of the coating layer, to obtain the failure data of the substrate coating layer;

[0092] Step S4: performing a substrate coating quality analysis based on the coating layer dynamic stress response data and the substrate coating layer failure data, and performing substrate coating design optimization on the substrate coating quality to obtain substrate coating design optimization data.

[0093] In the embodiment of the present invention, reference Figure 1 As shown, in this example, the quality control method of PVD coating design based on machine learning includes the following steps:

[0094] Step S1: obtaining a coating substrate object; performing a PVD coating design according to the coating substrate object, and performing a substrate PVD coating simulation based on the substrate PVD coating design to obtain substrate PVD coating simulation data;

[0095] In an embodiment of the present invention, a high-precision industrial scanner is used to perform a three-dimensional scan on the substrate to obtain the geometric morphology data of the substrate. Then, a mechanical microscope and an electron microscope are combined to detect the microstructure of the substrate surface, and the surface roughness, particle distribution and surface defect information of the substrate are recorded. After completing the data acquisition of the substrate, a three-dimensional model of the substrate is established using computer-aided design (CAD) software, and material properties such as elastic modulus, Poisson's ratio, thermal expansion coefficient, etc. are input into the finite element analysis (FEA) tool to ensure that the physical properties of the substrate are accurately reflected in the simulation process. Based on the three-dimensional model of the substrate, physical vapor deposition (PVD) simulation software such as COMSOL Multiphysics or ANSYS Fluent is used to set the coating process parameters, including deposition rate, target material type, substrate temperature, gas flow rate, etc., and simulate the coating process. After multiple iterative calculations, the deposition thickness distribution, particle growth state and deposition uniformity data of the coating layer on the surface of the substrate are extracted to form the substrate PVD coating simulation data.

[0096] Step S2: Analyze the mechanical coupling behavior data of the substrate PVD coating simulation data, and analyze the multi-scale mechanical properties of the coating material based on the substrate PVD coating design and the mechanical coupling behavior data, and use the multi-scale mechanical properties of the coating material to perform a dynamic stress response test of the coating layer to generate dynamic stress response data of the coating layer;

[0097] In an embodiment of the present invention, the simulation data of the PVD coating of the substrate is analyzed to extract its mechanical coupling behavior data. The mechanical coupling behavior mainly involves the deformation, stress distribution and interaction between the coating material and the substrate under the external load. The internal stress distribution and strain state of the coating layer are calculated by finite element analysis (FEA) to obtain the mechanical response of the coating layer under various loading conditions. Next, the multi-scale mechanical properties of the coating material are analyzed based on the substrate PVD coating design data and the mechanical coupling behavior data. This process includes nanoscale microscopic particle analysis, and the surface roughness, particle distribution and other microstructural characteristics of the coating material are obtained by scanning probe microscopy (SPM). Combined with molecular dynamics simulation, the macroscopic mechanical properties of the coating layer are further inferred. The dynamic stress response test of the coating layer is carried out using the multi-scale mechanical properties of the coating material, and the coating layer is tested by tensile test or indentation test. The loading rate in the tensile test is 1 mm / min, and the deformation degree of the coating layer is measured and the probability of crack extension is calculated. The dynamic stress response data of the coating layer is generated according to the dynamic stress response data, including stress, strain, deformation degree, etc.

[0098] Step S3: evaluating the chemical stability of the substrate coating based on the substrate PVD coating simulation data, and predicting the failure of the substrate coating layer based on the chemical stability of the substrate coating and the dynamic stress response data of the coating layer, to obtain the failure data of the substrate coating layer;

[0099] In an embodiment of the present invention, when evaluating the chemical stability of the substrate coating on the substrate PVD coating simulation data, the chemical composition of the coating layer is first analyzed by X-ray photoelectron spectroscopy (XPS), and the element distribution is analyzed by energy dispersive X-ray spectroscopy (EDS) to determine whether there is component segregation in the coating layer. Subsequently, an extreme environment is simulated in a high-temperature and high-humidity environment box, and the oxidation rate of the coating layer is detected through a long-term exposure experiment, and the change in surface roughness is measured using an atomic force microscope (AFM) to evaluate the chemical stability of the coating layer. For the failure prediction of the coating layer, an accelerated aging experiment is used, and the corrosion morphology and peeling of the coating layer are observed using equipment such as a salt spray corrosion test box and a wet heat aging test box. Combined with the dynamic stress response data of the coating layer, the fracture mechanics analysis method is used to calculate the critical stress of cracking of the coating layer under different environmental conditions to obtain the failure data of the substrate coating layer.

[0100] Step S4: performing a substrate coating quality analysis based on the coating layer dynamic stress response data and the substrate coating layer failure data, and performing substrate coating design optimization on the substrate coating quality to obtain substrate coating design optimization data.

[0101] In an embodiment of the present invention, when the quality analysis of the substrate coating is performed based on the dynamic stress response data of the coating layer and the failure data of the substrate coating layer, the surface morphology, microstructure, and thickness distribution of the coating layer are comprehensively evaluated to determine whether the coating layer meets the design requirements. Subsequently, an X-ray diffractometer (XRD) is used to analyze the crystal structure of the coating layer, and the grain size and orientation are calculated to evaluate the internal stress of the coating layer. In combination with the aforementioned analysis data, a coating layer quality evaluation index system is established, and the quality data under different coating process parameters are compared to screen out the optimal coating conditions. In the process of substrate coating design optimization, a data-driven optimization algorithm, such as a genetic algorithm or a particle swarm optimization algorithm, is used to adjust the PVD coating process parameters, including target power, deposition time, gas flow rate, etc., to optimize the mechanical properties and chemical stability of the coating layer, and the PVD coating simulation is re-performed according to the optimized parameters to verify the effectiveness of the optimization scheme, and output the substrate coating design optimization data.

[0102] Preferably, step S1 comprises the following steps:

[0103] Step S11: obtaining a coating substrate object;

[0104] Step S12: performing surface quality detection of the coated substrate according to the coated substrate object, thereby obtaining surface quality data of the coated substrate;

[0105] Step S13: performing a coating requirement analysis of the coating substrate object based on the coating substrate surface quality data and the coating substrate object to obtain the coating requirement of the substrate object;

[0106] Step S14: performing PVD coating design according to the coating requirements of the substrate object and the surface quality data of the coating substrate to obtain substrate PVD coating design data;

[0107] Step S15: performing substrate PVD coating simulation based on the substrate PVD coating design data to obtain substrate PVD coating simulation data.

[0108] In an embodiment of the present invention, a coating substrate object is obtained, and subsequent operations are performed based on the material properties, shape, size and surface features of the object. The methods for obtaining the substrate object include but are not limited to manual operation and automated scanning. In the automated process, a high-resolution three-dimensional scanner or laser scanning technology is used to comprehensively scan the substrate to obtain its accurate three-dimensional model data. These data will provide support for coating design and performance evaluation in the subsequent processing process. It can also be combined with technologies such as X-ray computed tomography (CT) to obtain the internal and external structural features of the substrate. The high-precision data obtained from these scans are used to construct a digital substrate model as the basic data for subsequent coating analysis and design. The obtained substrate object must have sufficient geometric information and surface characteristics, and the surface detection tool is used to perform quality inspection on the coated substrate to ensure the uniformity and cleanliness of the substrate surface. Commonly used surface quality inspection methods include white light interferometry, scanning electron microscopy (SEM) observation, surface roughness measurement, etc. White light interferometry can obtain the tiny concave and convex conditions of the substrate surface with high precision, and scanning electron microscopy can reveal the surface microstructure and pollutants or defects. The surface roughness is measured by a surface roughness meter to obtain surface roughness parameters such as Ra (arithmetic mean roughness) value and Rz (ten-point height). These test data include surface irregularities, residual pollutants, tiny cracks and other defect information. Combined with the surface quality data of the coated substrate obtained in step S12, the coating requirements of the coated substrate object are analyzed. This demand analysis determines the key requirements such as the thickness, type, mechanical properties, and mechanical properties of the coating layer based on factors such as the surface quality, geometry, and use environment of the substrate. Through multivariate analysis methods, such as factor analysis or regression analysis, combined with data such as the roughness of the substrate surface, material properties, and expected service life, the appropriate coating design requirements are calculated. For example, a substrate with a higher surface roughness requires a thicker coating layer to fill the irregularities. For substrates with high requirements for corrosion resistance, a coating material with higher chemical stability needs to be selected. According to the coating requirement analysis results in step S13, the surface quality data of the coating substrate is obtained to perform PVD coating design. Use advanced design tools and methods, such as simulation software based on physical models (such as COMSOL Multiphysics), to set coating process parameters considering the characteristics and requirements of the substrate surface. These parameters include deposition rate, distance between the target and the substrate, coating atmosphere, selection of target materials, etc. Through numerical simulation, simulate the coating process under different process parameters, optimize the deposition conditions, so that the coating layer can evenly cover the surface of the substrate and meet the mechanical and mechanical performance requirements of the design. At the same time, according to information such as the roughness of the substrate surface, adjust the thickness and uniformity of the coating layer to ensure the quality of the coating layer.The PVD coating design data includes deposition conditions, coating layer thickness distribution, expected mechanical properties and mechanical properties, etc., to form a detailed design plan, and perform substrate PVD coating simulation according to the substrate PVD coating design data obtained in step S14. Use special simulation software (such as Filmdoctor, etc.) to further simulate the substrate and coating process. During the simulation process, the parameters such as the surface shape of the substrate, the designed coating layer thickness, and the atmosphere conditions are input into the simulation software to simulate the flow of airflow, the transmission of particles, and the uniformity of deposition during the coating process. Through simulation, the deposition rate, coating thickness, and film quality data at different positions in the coating process are obtained. These data can predict the weak areas, thick film areas, or uneven phenomena that occur in the actual coating process, and the obtained substrate PVD coating simulation data include parameters such as film thickness, surface uniformity, and deposition rate at each position.

[0109] Preferably, step S13 comprises the following steps:

[0110] Step S131: extracting the geometric structure of the coated substrate according to the coated substrate object, thereby obtaining the geometric structure data of the coated substrate;

[0111] Step S132: analyzing the surface characteristics of the coating substrate according to the surface quality data of the coating substrate to obtain surface characteristic data of the coating substrate;

[0112] Step S133: dividing the coated substrate based on the geometric structure data of the coated substrate to obtain a flat substrate and a curved substrate;

[0113] Step S134: calculating the substrate coating adhesion ability according to the coating substrate surface characteristic data to obtain the substrate coating adhesion ability data;

[0114] Step S135: estimating the difficulty of coating the flat substrate according to the substrate coating adhesion ability data to obtain the difficulty data of coating the flat substrate;

[0115] Step S136: performing a curved surface substrate coating complexity analysis on the curved surface substrate according to the substrate coating adhesion ability data to obtain curved surface substrate coating complexity data;

[0116] Step S137: performing coating requirement analysis on the coating substrate object according to the coating complexity data of the curved substrate and the coating difficulty data of the flat substrate to obtain the coating requirement of the substrate object.

[0117] In an embodiment of the present invention, the geometric structure of the coated substrate object is extracted to obtain information such as the shape, size, surface and angle of the substrate. This step is usually performed by scanning the substrate through a three-dimensional scanning device (such as a laser scanner or a white light interferometer) to obtain its high-precision three-dimensional data. Through scanning, the geometric model of the substrate obtained includes information such as surface contour, external dimensions, and edge features. For substrates with complex geometric shapes, computer-aided design (CAD) software is used to post-process the scan data to convert the three-dimensional point cloud data into geometric data that can be used for analysis and design. The extracted geometric structure data is usually stored in formats such as point clouds, meshes, or parameterized surfaces for subsequent analysis and calculation. These data will serve as the basis for coating design to ensure that the design parameters match the geometric characteristics of the substrate, thereby affecting the uniformity and quality of the coating. According to the surface quality data of the coated substrate obtained in step S12, the surface characteristics of the substrate are analyzed. During the analysis process, the roughness, surface contaminants, microscopic defects, etc. of the substrate surface should be evaluated. Surface roughness measuring instruments (such as surface roughness meters, profilometers, etc.) are used to measure the surface height difference distribution, calculate the Ra value, Rz value and other indicators, and obtain the surface micro-roughness. At the same time, a scanning electron microscope (SEM) is used to observe the surface to detect defects such as microcracks and residual pollutants on the surface. In the chemical property analysis, X-ray photoelectron spectroscopy (XPS) and other technologies are used to analyze the elemental composition and chemical state of the surface material. Combining these data, through multi-dimensional analysis, the specific characteristics of the substrate surface are obtained, including surface roughness, chemical composition, micro-defects and other data. According to the obtained geometric structure data of the coated substrate, the substrate is divided into a flat substrate and a curved substrate. By analyzing the shape characteristics of the substrate surface, it is determined which areas belong to the flat substrate and which areas are the curved substrate. For a flat substrate, the points on the surface have the same normal vector, and the surface is uniformly stressed during the coating process; while the surface of the curved substrate has a curvature, and there is a problem of uneven coating deposition. This division is usually completed through geometric analysis tools. The CAD software is used to classify the substrate model, and the area with a smaller radius of curvature is detected as a curved substrate, while the area with a higher flatness is classified as a flat substrate. In addition, a shape analysis algorithm is used to automatically divide the substrate into different areas according to the change of the surface normal vector, and the coating adhesion ability of the substrate is calculated based on the surface characteristic data of the coated substrate. The coating adhesion ability refers to the bonding force between the coating material and the substrate surface, which is usually calculated by evaluating the surface energy, surface roughness, surface treatment of the substrate. Surface energy is a key factor affecting the coating adhesion ability, which is usually obtained through contact angle measurement or water drop test method to calculate the hydrophilicity or hydrophobicity of the substrate surface. The roughness of the substrate surface will also affect the adhesion of the coating. The rougher the surface, the stronger the adhesion is usually, but excessive roughness leads to uneven coating.By combining the surface chemical properties and physical properties, the classical adhesion calculation formula (such as JKR theory) is used for quantitative analysis to obtain the adhesion ability data of the substrate coating. Based on the substrate coating adhesion ability data calculated in step S134, the coating difficulty of the planar substrate is estimated. By analyzing the coating adhesion ability of the substrate, the difficulties encountered during the coating process are predicted. The coating difficulty is mainly affected by the surface energy, roughness and substrate material. In the estimation of the planar substrate, if the substrate surface adhesion ability is strong, the coating process is relatively smooth and the difficulty is low; conversely, if the adhesion ability is weak, the coating process will have problems such as uneven coating or film shedding, and the coating difficulty is high. Accurately evaluate the coating difficulty of the planar substrate, by establishing a set of coating difficulty evaluation models, which combines factors such as surface roughness, adhesion, and material properties, and obtains a specific coating difficulty value by calculation. Based on the obtained substrate coating adhesion ability data, the coating complexity analysis is performed for the curved substrate. Since the surface of the curved substrate has a curvature, there is a certain degree of deposition non-uniformity during the coating process, especially in the uneven area of ​​the curved surface, resulting in inconsistent coating quality. By analyzing the surface characteristics of the curved substrate, combined with the coating adhesion data, the computational fluid dynamics (CFD) method is used to simulate the airflow distribution and particle deposition, and the complexity data of the curved substrate coating is obtained. The complexity evaluation takes into account factors such as the curved surface shape, the roughness of the substrate surface, and the adhesion, and a value representing the complexity of the curved substrate coating is calculated by the model. Combined with the flat substrate coating difficulty data obtained in step S135 and the curved substrate coating complexity data obtained in step S136, the coating demand analysis of the coating substrate object is performed. By comprehensively evaluating the coating difficulty and complexity of different areas (planes and curved surfaces), the overall coating demand is obtained. This demand analysis not only takes into account factors such as the adhesion, film thickness, and uniformity of the coating, but also combines the feasibility of the coating process and the adaptability of the process conditions. During the analysis process, the weighted average method or multi-factor decision analysis method is used to weight the coating characteristics of different areas to obtain the overall coating demand data.

[0118] Preferably, step S136 includes the following steps:

[0119] Step S1361: extracting irregularities of the curved surface substrate according to the curved surface substrate to obtain irregularities of the curved surface substrate;

[0120] Step S1362: Calculating the curvature of the curved substrate according to the irregularity of the curved substrate to obtain the curvature of the curved substrate;

[0121] Step S1363: performing differential estimation of the deposition angle of the curved substrate based on the curvature and irregularity of the curved substrate to obtain differential deposition angle data;

[0122] Step S1364: performing substrate sputtering deposition distribution difference detection according to the irregularity of the curved substrate to obtain substrate sputtering deposition distribution difference data;

[0123] Step S1365: Calculating the heterogeneity of deposition rate on the substrate surface based on the substrate sputtering deposition distribution difference data and the deposition angle difference data to obtain the heterogeneity of deposition rate on the substrate surface;

[0124] Step S1366: Performing a curved surface substrate coating complexity analysis on the substrate surface deposition rate heterogeneity according to the substrate coating adhesion ability data to obtain the curved surface substrate coating complexity data.

[0125] In an embodiment of the present invention, the surface morphology of the curved substrate is scanned with high precision to extract its irregularities. The surface of the substrate is scanned finely using a scanning electron microscope (SEM) or a three-dimensional laser scanner to obtain three-dimensional point cloud data of the substrate surface. These point cloud data contain microscopic irregularity information on the surface of the substrate, such as concave-convex, cracks, roughness, etc. Then, the scanned data is processed by an algorithm (such as a filtering algorithm based on fast Fourier transform or a wavelet transform method) to extract the local irregularity characteristics of the substrate surface. Assuming that in the scanned data, the root mean square value of the surface roughness is 0.8 μm and the peak-to-valley height is 2 μm, these irregularity data can be effectively extracted for subsequent curvature calculation and deposition angle analysis. The irregularity data of the curved substrate is obtained by this method, and the curvature of the substrate surface is calculated using a surface fitting method according to the substrate irregularity data extracted in step S1361. The specific operation is to use a quadratic surface fitting or a B-spline curve fitting method to perform surface modeling based on the obtained irregularity data. Taking 3D scanning data as input, the surface curvature value is obtained by calculating the normal vector and surface direction of each point on the surface. Assuming that the curvature calculation range of the substrate surface is ±1.0 (curvature radius from 5mm to 10mm), the curvature value of each point is calculated based on this range. By summarizing the curvature data of all points, an overall curvature distribution map is formed, thereby obtaining the local and global curvature information of the curved substrate. Combining the curvature of the curved substrate obtained in step S1362 and the irregularity data in step S1361, the complexity of the deposition process is estimated by calculating the difference in deposition angles at different positions. In this step, the deposition angles at different positions are first calculated based on the incident angle (such as 45°, 60°, etc.) in the coating process combined with the local curvature of the substrate surface. For curved substrates, the incident angle will change due to the change in surface curvature, which will affect the thickness and uniformity of the deposited layer. Assume that on a substrate with a concave area, the curvature of the substrate surface in this area is negative, resulting in a smaller deposition angle, while in the convex area, the curvature is positive and the deposition angle is larger. By estimating the deposition angles of different areas, the deposition angle differentiation data of the coating layer is obtained, and the difference in the sputtering deposition distribution of the substrate is detected based on the surface irregularity of the curved substrate. First, the physical vapor deposition (PVD) sputtering model is used to simulate the substrate surface to calculate the deposition rate and thickness at different positions. In this simulation process, it is assumed that under different substrate surface irregularities, the angle between the incident angle of the sputtered particles and the surface normal vector will be different, thus affecting the deposition distribution. For example, in a concave area, the deposited particles will be concentrated in a smaller area, while in a convex area, the particles will spread to a larger range.By using the simulation technology based on the Monte Carlo method, the distribution map of the deposited particles on the substrate surface is obtained, the thickness difference and uneven deposition generated during the deposition process are detected, and the substrate sputtering deposition distribution difference data is obtained through statistical analysis of the sputtering deposition distribution. Combined with the substrate sputtering deposition distribution difference data in step S1364 and the deposition angle differentiation data in step S1363, the heterogeneity of the deposition rate on the substrate surface is calculated. By analyzing the deposition speed and angle differences at different positions, combined with the average kinetic energy of the sputtering particles, the deposition rate of different areas of the substrate is calculated using a theoretical formula (such as the Langevin model). For example, in the concave area, the deposition rate of the particles is relatively slow, while in the convex area, the deposition rate is relatively fast due to the increase in the particle incident angle. Combined with the above data, the deposition rate heterogeneity of each point on the substrate surface is calculated using the differential method, and the corresponding velocity distribution map is output. Finally, by combining the deposition rate with the surface irregularity, the heterogeneity data of the deposition rate on the substrate surface is obtained. Based on the heterogeneity data of the deposition rate on the substrate surface obtained in step S1365, combined with the coating adhesion ability data of the substrate, the coating complexity analysis of the curved substrate is performed. First, the adhesion ability of the substrate is tested experimentally (for example, through adhesion test, shear strength test, etc.), and the adhesion data of different positions on the substrate surface are obtained. Then, these adhesion data are combined with the deposition rate heterogeneity data to analyze how the coating adhesion of different areas on the substrate surface affects the formation of the coating layer. It is assumed that during the deposition process, due to the low deposition rate in some areas, the adhesion of the coating layer is weak, which in turn affects the quality of the coating layer. Through further analysis of these data, the substrate coating complexity data is obtained.

[0126] Preferably, the mechanical coupling behavior analysis in step S2 includes:

[0127] Conducting a coating material surface rigidity test on the substrate PVD coating simulation data to obtain coating material surface rigidity data;

[0128] Calculate the surface friction coefficient of the coating material using the surface rigidity data of the coating material;

[0129] Evaluate the wear resistance of the coating material based on the friction coefficient of the coating material surface;

[0130] The wear state of the coating layer is estimated based on the surface friction coefficient of the coating material and the wear resistance of the coating material to obtain the wear state data of the coating layer;

[0131] Use nano-indentation technology to test the fracture toughness of the coating material by measuring the surface rigidity of the coating material;

[0132] Evaluate the external load capacity of the coating layer based on the fracture toughness of the coating material;

[0133] The mechanical coupling behavior analysis is performed based on the external load capacity of the coating layer and the wear state data of the coating layer to obtain the mechanical coupling behavior data.

[0134] In an embodiment of the present invention, a scanning electron microscope (SEM) is used to observe the surface morphology of the coating layer involved in the substrate PVD coating simulation data to confirm its microstructural characteristics, including the surface roughness, particle distribution and porosity of the coating layer. Then, a nanoindentation test device (such as CSEM NanoIndenter or KLA Tencor) is used to test the surface rigidity of the coating material. The nanoindentation test applies an indentation of a known depth on a micrometer or nanometer scale, and determines the hardness and elastic modulus of the coating material by the relationship between the indentation depth and the applied force. In the experiment, the loading rate used was 0.1mN / s, the load peak was set to 1mN, the maximum indentation depth was 50nm, and the application time was 10 seconds. By analyzing the indentation test data, the rigidity data of the coating material, including the surface hardness and elastic modulus, are obtained. According to the surface rigidity data obtained in the previous step, the friction coefficient of the surface of the coating material is calculated using a tribological experiment, and a linear friction test is performed on a tribological testing machine (such as Anton PaarTribometer). In this experiment, a steel ball or a ceramic ball is used as a friction material to slide on the surface of the coating material. The load set in the friction test is 5N, the sliding speed is 1mm / s, and the sliding distance is 10mm. The friction coefficient of the surface of the coating material is calculated by measuring the ratio of friction force to normal pressure. During the experiment, environmental factors such as temperature and humidity must also be controlled to ensure the accuracy of the friction coefficient test. The test process is repeated, and after obtaining multiple test results, the average value is taken as the friction coefficient. After obtaining the surface friction coefficient of the coating material, the wear resistance of the coating material is evaluated by comparing the wear conditions under different friction coefficients, and a wear test is carried out. The coating layer is evaluated using a wear tester (such as a Pin-on-Disk tester). The abrasive used in the experiment is standard quartz sand, the applied pressure is 10N, the wear time is 60 minutes, and the speed is set to 1000rpm. The wear rate of the material is estimated by measuring the difference in the mass of the coating layer before and after wear. The wear resistance evaluation result of the coating material is obtained through the relationship between the friction coefficient and the wear rate. During the specific analysis, it is compared with the wear resistance of standard materials or different coating layers to further determine its wear resistance in actual use. According to the friction coefficient data and wear resistance data obtained in the previous steps, the wear state of the coating layer is estimated, and the wear model of the coating layer is established by comparing the friction coefficient and wear rate of different materials. In this process, statistical regression analysis methods (such as the least squares method) are used to establish a wear prediction model based on the relationship between the friction coefficient and the wear rate obtained from the experiment. For each coating material under different conditions, simulations are performed to predict its wear state under specific working conditions. During the simulation process, parameters such as temperature, friction, and sliding rate are used as input variables, and the wear rate of the coating layer under different working conditions is calculated in combination with the tribological model.Through feedback correction of multiple experimental data, the wear state data of the coating layer under specific conditions are obtained. After analyzing the surface rigidity test data of the coating material, the fracture toughness of the coating material is further tested using nanoindentation technology. In the experiment, a nanoindenter (such as Hysitron TriboIndenter) is used for loading, applying high pressure load and measuring the material fracture phenomenon caused by it. The experimental load is gradually increased, the indentation depth is controlled within 50nm, and the crack extension of the coating material under high pressure is gradually observed. By recording the relationship between load and indentation depth, combined with the fracture toughness theory, the fracture toughness data of the coating material is calculated. The test results will include the critical fracture load of the material, fracture propagation mode, crack tip stress distribution, etc. According to the obtained fracture toughness data of the coating material, the external load capacity of the coating layer is evaluated. Combined with the fracture toughness data measured in the experiment, the mechanical behavior of the coating layer under external load is calculated using a mechanical model. By applying classical fracture mechanics theory (such as linear elastic fracture mechanics LEFM), the critical stress intensity factor of the coating layer under different external loads is calculated. According to the calculation results, the external load capacity of the coating layer is obtained, including its maximum load capacity and critical conditions for rupture under different load conditions. After obtaining the external load capacity and wear state data of the coating layer, the mechanical coupling behavior analysis of the coating layer is carried out. By using multi-physics coupling analysis software (such as COMSOL Multiphysics or ABAQUS), the mechanical response of the coating layer under external load and wear is simulated and analyzed. During the simulation process, parameters such as external load, wear rate, friction coefficient, etc. are used as input variables to simulate their influence on the mechanical properties of the coating layer. The mechanical coupling behavior data is obtained.

[0135] Preferably, the multi-scale mechanical property analysis of the coating layer in step S2 includes:

[0136] The thickness of the substrate coating layer is calculated using the substrate PVD coating design to obtain the thickness of the substrate coating layer;

[0137] The substrate PVD coating design is used to analyze the microscopic particles of the substrate coating layer and obtain the microscopic particle data of the substrate coating layer;

[0138] Extracting the microscopic particle distribution characteristics of the substrate in the coating layer of the substrate with a particle size distribution ranging from 10nm to 5000nm;

[0139] The surface hardness of the coating layer to estimate the microscopic particle characteristics of the coating layer;

[0140] The thickness of the substrate coating layer is measured with a measurement interval of 0.1-100 μm to obtain the thickness difference of the substrate coating layer;

[0141] The stress concentration area in the coating layer is calculated by using the thickness difference of the substrate coating layer and the microscopic particle distribution characteristics of the substrate;

[0142] Predict the impact resistance of the coating layer based on the mechanical coupling behavior data and the stress concentration area in the coating layer;

[0143] The multi-scale mechanical properties of the coating layer are analyzed based on the surface hardness of the coating layer and the impact resistance of the coating layer to generate the multi-scale mechanical properties of the coating layer.

[0144] In an embodiment of the present invention, it is necessary to determine the PVD coating design parameters of the substrate, including sputtering power, atmosphere pressure, gas type, distance between the substrate and the target, etc. Assuming that appropriate coating process conditions are selected, the physical vapor deposition (PVD) model is used to calculate the thickness of the coating layer. In the specific calculation process, the coating rate is first estimated based on the gas flow rate, the distance between the substrate and the target, and the sputtering power using experimental data or calculation formulas (such as Fick's law or Knudsen flow formula). Assuming that the coating rate is 2nm / s, after a 5-minute deposition process, the thickness of the substrate coating layer is calculated to be 600nm. In addition, considering the differences in deposition rates at different locations, it is necessary to use high-precision measurement tools (such as a film thickness gauge) to perform actual thickness measurements at multiple points to further optimize the calculation model. The coating layer is microstructurally analyzed using a scanning electron microscope (SEM) or an atomic force microscope (AFM). By observing the surface morphology of the coating layer, the size, shape and distribution information of the microscopic particles are extracted. When using AFM to scan the surface of the substrate coating layer, set the scanning resolution to 0.5nm and the scanning area to 10μm×10μm to obtain high-precision microscopic particle data. Assume that through measurement, it is found that the microscopic particles of the coating layer are mainly distributed in the range of 20nm to 3000nm, and the particles are round or elliptical in shape. The size, distribution and shape of these particles affect the physical properties of the coating layer, such as hardness, friction coefficient, etc. By further analyzing the distribution characteristics of the particles, the overall quality of the coating layer is evaluated, and the size distribution of the particles in the substrate coating layer is extracted and analyzed in detail. Image processing technology is used to extract information such as particle diameter, shape, and arrangement from SEM or AFM images. Assume that in a certain experiment, the size range of the particles in the substrate coating layer is 10nm to 5000nm. Image analysis software (such as ImageJ) is used to binarize the image, and the size and morphological characteristics of the particles are obtained through morphological analysis algorithms. The particle size range is further divided into different groups (e.g., 10-100nm, 100-1000nm, 1000-5000nm) to analyze its distribution in the coating layer. For example, in a certain area of ​​the coating layer, the main size of the particles is concentrated between 100nm and 1000nm, and the particles are relatively evenly distributed, while in another area, the particle size is uneven. Based on the particle size distribution data extracted in step S3, the surface hardness of the coating layer is inferred by theoretical models or experimental data. Assuming that the Hershel hardness test method is used, the hardness test of the coating layer is performed at multiple locations, and the distribution of hardness values ​​is closely related to the particle size, shape and density in the coating layer. Use Vickers hardness or nanoindentation testing technology to obtain the hardness value of the coating layer. Assuming that the test is performed in different areas of the sample, the distribution range of the hardness value is 6-12GPa.By analyzing the correlation between the particle distribution characteristics and the hardness data, the hardness of the coating layer is predicted by using a regression model or a machine learning method (such as a support vector machine). Assuming that the improvement in surface hardness mainly comes from the dense distribution of particle sizes in the range of 100nm to 1000nm according to the distribution characteristics of the particles, a high-precision laser scanning thickness gauge is used to measure the thickness of the substrate coating layer, with the measurement interval ranging from 0.1μm to 100μm, to ensure that the thickness difference of the coating layer at the micron level can be obtained. Assuming that multiple measurement points are selected in different areas of the coating layer with a measurement interval of 1μm, the thickness difference data is finally obtained. For example, in one area, the thickness of the coating layer is 550nm, while in another area it is 630nm, and the measurement difference is 80nm. These thickness difference data reveal the problem of uneven deposition during the coating process. By combining the thickness difference data in step S5 and the particle distribution characteristics in step S3, the stress concentration area in the coating layer is calculated. First, the thickness difference of the coating layer and the particle size and distribution are input into the simulation system using a finite element analysis (FEA) model. Assume that in some areas, the thickness of the coating layer is thin and the particles are large, resulting in uneven stress distribution in the coating layer, forming a stress concentration area. Through simulation analysis, the position and size of the stress concentration area are obtained. Assume that in a certain area, the maximum stress concentration value is 150MPa. Based on the stress concentration area data calculated in step S6, the impact resistance of the coating layer is evaluated by mechanical coupling analysis. A dynamic mechanical simulation method (such as a finite element method) is used to couple the external impact force and the internal stress distribution of the coating layer. Assuming that the external impact force is 1N, the impact resistance of the coating layer is simulated to be able to withstand an impact energy of 1.5J, and the maximum stress of the coating layer is 200MPa. According to the surface hardness data obtained in step S4 and the impact resistance data in step S7, the multi-scale mechanical properties of the coating layer are analyzed. A multi-scale modeling method (such as a macro-micro coupling model) is used to comprehensively consider the hardness, impact resistance and distribution characteristics of microscopic particles of the coating layer to generate multi-scale mechanical property data of the coating layer. For example, under certain process conditions, the hardness of the coating layer is 9 GPa and the impact resistance is 1.5 J. After multi-scale analysis, the comprehensive mechanical properties of the coating layer are obtained.

[0145] Preferably, the dynamic stress response test of the coating layer in step S2 includes:

[0146] The tensile rate of 1 mm / min was used to simulate the multi-scale mechanical properties of the coating layer and obtain the tensile data of the coating layer;

[0147] Calculate the deformation degree of the coating layer from the tensile data of the coating layer;

[0148] Calculate the probability of cracks in the coating layer when the deformation degree of the coating layer is 50nm;

[0149] The elongation at break of the coating layer is estimated according to the deformation degree of the coating layer and the crack probability of the coating layer at a stress of 500 MPa, and the elongation at break of the coating layer is obtained;

[0150] Analyze the coating layer strength of the coating layer elongation at break;

[0151] The dynamic stress response test of the coating layer is carried out based on the strength of the coating layer and the elongation at break of the coating layer to obtain the dynamic stress response data of the coating layer.

[0152] In the embodiment of the present invention, the tensile simulation of the coating layer is performed on the multi-scale mechanical properties of the coating layer at a tensile rate of 1 mm / min to obtain the tensile data of the coating layer. The coating layer tensile experiment is carried out on a precision mechanical loading platform using a micro-nanomechanical testing system. The two ends of the coating sample are fixed to ensure that the stress applied by the clamp to the coating layer is evenly distributed to avoid the introduction of additional stress concentration due to improper clamping. The loading device adopts a constant tensile rate of 1 mm / min, and uses a high-precision displacement sensor to record the change in tensile length during the whole process, and measures the applied tensile load through a force sensor. The digital image correlation (DIC) method is used simultaneously to track the deformation field distribution on the surface of the coating layer at the micrometer scale, obtain the strain distribution of the coating layer during the tensile process, and form a tensile data set of the coating layer. The deformation degree of the coating layer of the tensile data of the coating layer is calculated. The finite element analysis method is used to input the tensile data into the material deformation calculation model to analyze the strain distribution of the coating layer at different positions during the tensile process. The average strain value is calculated using the full-field strain data measured by DIC, and the true stress-strain curve of the coating layer is calculated in combination with the tensile load data. At the nanoscale, an atomic force microscope (AFM) was used to measure the surface morphology and analyze the deformation of the coating layer under stress, especially to observe the displacement changes of nano-scale particles. Combined with the mechanical interference measurement method, the thickness change of the coating layer was analyzed to calculate its overall deformation degree. The crack probability of the coating layer with a deformation degree of 50nm was calculated. The microscopic crack morphology of the coating layer after stretching was characterized by scanning electron microscopy (SEM) and transmission electron microscopy (TEM), and the image processing algorithm was used to identify the crack starting point and crack propagation path. For the deformation area of ​​50nm scale, the crack density statistical method was used to calculate the number of cracks per unit area, and the probability of crack occurrence at 50nm scale was calculated in combination with the mechanical constitutive relationship of the coating material. Based on the principle of fracture mechanics, the stress intensity factor method was used to evaluate the critical conditions for crack propagation, and the probability of crack initiation was calculated by probability statistics. The elongation at break of the coating layer is estimated according to the deformation degree of the coating layer under stress of 10-500MPa and the probability of cracks in the coating layer, and the elongation at break of the coating layer is obtained. The strain data of the coating layer under stress of 10-500MPa is substituted into the fracture mechanics model to calculate the critical strain value. The strain-crack probability curve is fitted with the experimental data to determine the probability of crack occurrence under different stress levels, and the elongation at break of the coating layer is calculated in combination with the fracture toughness parameter. The experimental results are used for data comparison to verify the accuracy of the estimated elongation at break, and the calculation model is corrected according to the experimental results to make it more consistent with the actual test data. The strength of the coating layer is analyzed for the elongation at break of the coating layer. The hardness and Young's modulus of the coating layer are measured by the nanoindentation test method, and the strength change of the coating layer under different strain conditions is evaluated in combination with the elongation at break data.The ultimate tensile strength of the coating layer was calculated by using the micro-tensile test data, and the fracture mechanism of the coating layer was determined by combining the fracture morphology analysis results. The residual stress of the coating layer was measured by X-ray diffraction (XRD) method, and the comprehensive mechanical properties of the coating layer were evaluated by combining the tensile test data. The dynamic stress response test of the coating layer was carried out based on the strength of the coating layer and the elongation at break of the coating layer, and the dynamic stress response data of the coating layer was obtained. The stress change of the coating layer under dynamic loading was monitored in real time by laser ultrasonic technology, and the deformation process of the coating layer under impact load was recorded by a high-speed camera system. The high strain rate impact test of the coating layer was carried out using the Split Hopkinson Pressure Bar (SHPB), and the dynamic stress response curve of the coating layer was recorded at nanosecond time resolution. Combined with the finite element simulation method, the dynamic stress distribution of the coating layer at different loading rates was calculated, and the model was corrected by experimental data to generate the dynamic stress response data of the coating layer.

[0153] Preferably, step S3 comprises the following steps:

[0154] Step S31: performing substrate coating material detection on substrate PVD coating simulation data to obtain substrate coating material data;

[0155] Step S32: evaluating the stability of the substrate coating structure according to the substrate coating material data to obtain the stability of the substrate coating structure;

[0156] Step S33: performing corrosion resistance test on the substrate coating according to the chemical stability of the substrate coating to obtain corrosion resistance data of the substrate coating;

[0157] Step S34: performing failure prediction of the substrate coating layer based on the corrosion resistance data of the substrate coating layer and the dynamic stress response data of the coating layer to obtain failure data of the substrate coating layer.

[0158] As an example of the present invention, refer to Figure 2 As shown, in this example, step S3 includes:

[0159] Step S31: performing substrate coating material detection on substrate PVD coating simulation data to obtain substrate coating material data;

[0160] In an embodiment of the present invention, the composition information of the substrate coating layer is extracted from the PVD coating simulation process, and materials are tested using equipment such as X-ray diffraction (XRD), scanning electron microscope (SEM) and energy spectrum analysis (EDS). XRD can be used to analyze the crystal structure of the coating layer to ensure that the phase composition of the coating material is consistent with the target design. SEM is used to observe the surface morphology of the coating layer and evaluate its microstructure and uniformity. Combined with EDS, the distribution of chemical elements in the coating layer is quantitatively analyzed to obtain detailed data of the coating material, including the chemical composition, element content, grain size and distribution of the coating layer. By integrating these data, comprehensive information about the substrate coating material is obtained.

[0161] Step S32: evaluating the chemical stability of the substrate coating according to the substrate coating material data to obtain the chemical stability of the substrate coating;

[0162] In the embodiment of the present invention, based on the substrate coating material data obtained from the coating material test, the chemical corrosion test is conducted by immersing the coating sample in different acid and alkali solutions to simulate its corrosion behavior in the actual use environment. According to these experimental results, combined with the composition and structural characteristics of the coating material, the chemical stability of the coating layer under specific conditions is calculated to obtain the chemical stability data of the substrate coating layer.

[0163] Step S33: performing corrosion resistance test on the substrate coating according to the chemical stability of the substrate coating to obtain corrosion resistance data of the substrate coating;

[0164] In the embodiments of the present invention, the coating layer is tested by simulating the corrosion conditions in the actual use environment, using methods such as salt spray test, damp heat test or immersion test. In the salt spray test, the coating sample is placed in a salt spray box and exposed to the salt spray environment under certain temperature and humidity conditions. The corrosion resistance is evaluated by testing the corrosion signs on the surface of the coating layer. The damp heat test evaluates the corrosion under extreme humidity conditions by exposing the coating sample to a high temperature and high humidity environment. The immersion test is to immerse the coating layer sample in different corrosive media to observe the corrosion rate and degradation of the coating layer. Through these tests, combined with the chemical stability data of the coating material, the corrosion resistance data of the substrate coating is obtained.

[0165] Step S34: performing failure prediction of the substrate coating layer based on the corrosion resistance data of the substrate coating layer and the dynamic stress response data of the coating layer to obtain failure data of the substrate coating layer.

[0166] In an embodiment of the present invention, a failure analysis method is used to predict the failure of the substrate coating layer. A failure mode model of the coating layer is established by combining the corrosion resistance test and the dynamic stress response data of the coating layer. The model takes into account the various stresses, environmental conditions and corrosion effects faced by the coating layer during actual use, and evaluates the failure risk of the coating layer under different working conditions. Stress analysis is performed using numerical simulation methods (such as finite element analysis), and combined with experimental data of corrosion failure, the failure time and failure mode (such as peeling, cracking, corrosion penetration, etc.) of the coating layer are predicted.

[0167] Preferably, step S34 includes the following steps:

[0168] Step S341: performing external impact simulation according to the dynamic stress response data of the coating layer to obtain external impact simulation data of the coating layer;

[0169] Step S342: estimating the degradation fatigue trend of the coating layer based on the external impact simulation data of the coating layer to obtain degradation fatigue trend data of the coating layer;

[0170] Step S343: performing corrosion environment simulation according to the corrosion resistance data of the substrate coating to obtain corrosion environment simulation data of the coating layer;

[0171] Step S344: performing a corrosion extension estimation of the coating layer on the corrosion resistance data of the substrate coating based on the corrosion environment simulation data of the coating layer to obtain corrosion extension data of the coating layer;

[0172] Step S345: estimating the coating layer shedding state according to the coating layer corrosion extension data and the coating layer degradation fatigue trend data to obtain coating layer shedding state data;

[0173] Step S346: performing a failure estimation of the substrate coating layer according to the coating layer peeling state data and the coating layer corrosion extension data to obtain substrate coating layer failure data.

[0174] As an example of the present invention, refer to Figure 3 As shown, in this example, step S34 includes:

[0175] Step S341: performing external impact simulation according to the dynamic stress response data of the coating layer to obtain external impact simulation data of the coating layer;

[0176] In an embodiment of the present invention, an external impact test is performed on the dynamic stress response data of the coating layer by a mechanical simulation experiment or a numerical simulation method. Commonly used simulation tools include finite element analysis (FEA) software, which is used to simulate the application and propagation of external impact loads. The impact load is usually applied by a hard object to the surface of the coating layer with impact forces of different energies to simulate the physical impact of the external environment on the coating layer. An impact testing machine was used in the experiment to perform actual tests, applying impact forces of different frequencies and intensities, and monitoring the deformation, crack propagation, surface damage and other behaviors of the coating layer through sensors. By analyzing the deformation and damage during the impact simulation process, the simulation data of the coating layer under external impact is obtained, such as the starting point of crack generation, the propagation speed, and the tolerance limit.

[0177] Step S342: estimating the degradation fatigue trend of the coating layer based on the external impact simulation data of the coating layer to obtain degradation fatigue trend data of the coating layer;

[0178] In an embodiment of the present invention, the fatigue trend of the coating layer degradation is estimated based on the external impact simulation data of the coating layer obtained in step S341. In this process, the fatigue life is predicted by collecting the stress-strain cycle data in the simulation and combining the SN curve (stress-life curve). Specifically, different impact frequency and amplitude data are used in combination with experimental data (for example, multiple impact tests are performed on the sample using a fatigue testing machine) to establish a fatigue damage model of the coating layer under actual working conditions. Assuming that the fatigue life of the coating layer at a frequency of 50Hz is 5000 times and the fatigue life at a frequency of 100Hz is 3000 times, the simulation data shows that as the impact load increases, the crack propagation rate of the coating layer will also increase. By analyzing the fatigue test data, the degradation fatigue trend data of the coating layer is obtained, and then the durability and damage accumulation of the coating layer in long-term use are evaluated.

[0179] Step S343: performing corrosion environment simulation according to the corrosion resistance data of the substrate coating to obtain corrosion environment simulation data of the coating layer;

[0180] In an embodiment of the present invention, the corrosion resistance data of the substrate coating is obtained, and the reaction of the coating layer in various corrosion environments is simulated through corrosion simulation experiments. The coating sample is exposed using a corrosion test box, and the test includes environmental simulations such as salt spray, acid and alkali, and damp heat. By controlling factors such as temperature, humidity, and pH, the corrosion behavior of the coating layer under different actual working conditions is simulated. By regularly detecting changes on the surface of the coating layer (such as corrosion holes, discoloration, crack extension, etc.), the impact of different corrosion environments on the coating layer is evaluated. During the experiment, the weight loss method, surface morphology analysis method, X-ray diffraction (XRD) and other techniques are combined to obtain the microstructural changes and corrosion rate of the coating layer before and after corrosion, and then obtain the corrosion environment simulation data of the coating layer.

[0181] Step S344: performing a corrosion extension estimation of the coating layer on the corrosion resistance data of the substrate coating based on the corrosion environment simulation data of the coating layer to obtain corrosion extension data of the coating layer;

[0182] In an embodiment of the present invention, the corrosion extension model is used to predict the corrosion extension behavior of the coating layer based on the obtained corrosion environment simulation data of the coating layer in combination with the chemical stability data of the coating layer. A model-based prediction method is used to calculate the progress of corrosion of the coating layer under different corrosion conditions through the corrosion extension rate equation. The specific method includes using a linear or nonlinear corrosion model to simulate the corrosion extension process of the coating layer within different time ranges. By considering the influence of temperature, humidity, pH, and environmental factors on the corrosion rate, the corrosion extension data of the coating layer under the actual use environment is calculated. In addition, dynamic monitoring technology, such as electrochemical impedance spectroscopy (EIS), is used to evaluate the corrosion behavior of the coating layer in real time, and the corrosion extension data of the coating layer under different corrosion conditions are obtained.

[0183] Step S345: estimating the coating layer shedding state according to the coating layer corrosion extension data and the coating layer degradation fatigue trend data to obtain coating layer shedding state data;

[0184] In an embodiment of the present invention, the shedding state of the coating layer is estimated by combining the corrosion expansion data obtained in step S344 and the fatigue degradation trend data in step S342. In this process, the multi-physics field coupling analysis method is used to combine the effects of corrosion expansion and fatigue degradation for simulation. Assuming that under a certain fatigue load, corrosion expansion will accelerate the shedding of the coating layer, and the corresponding shedding state data is generated. By combining the corrosion rate, fatigue cycle number and mechanical properties of the coating layer under different working conditions, the shedding state of the coating layer under specific conditions is predicted. For example, if crack expansion occurs in the stress concentration area of ​​the coating layer under a corrosive environment, it can be inferred that the coating layer begins to fall off after 3000 impacts. Through this estimation, the shedding state data of the coating layer is generated, and the shedding probability distribution is provided.

[0185] Step S346: performing a failure estimation of the substrate coating layer according to the coating layer peeling state data and the coating layer corrosion extension data to obtain substrate coating layer failure data.

[0186] In an embodiment of the present invention, the failure prediction model is used to estimate the failure of the substrate coating layer in combination with the obtained coating layer shedding state data and coating layer corrosion extension data. Based on the material mechanics and corrosion extension theory, the finite element method (FEA) is used to perform fatigue analysis and failure analysis of the coating layer. By constructing a failure model, the failure process of the coating layer is simulated in combination with the combined effects of different environmental conditions, external impacts, wear and corrosion. The effects of corrosion extension and wear state on the coating layer are analyzed, and the critical conditions and failure time of failure are calculated. Through dynamic monitoring technology, the corrosion extension and wear conditions of the coating layer are obtained in real time to provide data support for failure prediction. Based on the failure model and experimental data, the failure data of the substrate coating layer is obtained, including failure time, failure mode and failure cause.

[0187] Preferably, step S4 comprises the following steps:

[0188] Step S41: evaluating the quality of the substrate coating based on the dynamic stress response data of the coating layer and the failure data of the substrate coating layer to obtain the substrate coating quality data;

[0189] Step S42: performing a substrate coating quality defect analysis according to the substrate coating quality data to obtain substrate coating quality defect data;

[0190] Step S43: Optimizing the substrate coating design based on the substrate coating quality defect data based on machine learning to obtain substrate coating design optimization data.

[0191] In an embodiment of the present invention, the dynamic stress response data of the coating layer and the failure data of the substrate coating layer are used for comprehensive analysis to evaluate the quality of the substrate coating. By obtaining the stability data of the coating material under different environmental conditions (such as corrosion, wear, thermal stability, etc.), and the failure data of the substrate coating layer during use (such as cracks, peeling, shedding, etc.), a substrate coating quality evaluation model is constructed. In specific operations, the corrosion resistance test results and wear test data under different environments are used, combined with the failure analysis theory, to analyze the life, durability and overall performance of the coating layer. By constructing a multi-factor evaluation model, the influence of different process parameters, material properties and environmental factors on the coating quality is calculated, and the substrate coating quality data is obtained, including the thickness uniformity, strength, hardness, corrosion resistance and other key performance indicators of the coating layer. According to the substrate coating quality data obtained in step S41, the substrate coating quality defect analysis is further performed. By statistics and analysis of the substrate coating quality data, the type of quality defects that exist is determined. Specifically, defect identification methods such as mechanical microscopy, scanning electron microscopy (SEM), X-ray fluorescence analysis (XRF), etc. are used to conduct detailed inspections on the surface and interior of the coating layer to find defects such as surface cracks, pores, peeling, and uneven coating. In the defect analysis process, the relationship between different defect types and the strength, toughness, and impact resistance of the coating layer is analyzed in combination with mechanical performance data. Digital image processing technology is used to quantitatively analyze the defects, identify the specific location, size, and distribution of the defects, and obtain the substrate coating quality defect data through data mining and pattern recognition methods, including the probability and type of defects and their impact on product performance. Based on the substrate coating quality defect data obtained in step S42, a machine learning method is applied to optimize the coating design. The collected coating quality defect data is used as a training set to construct a machine learning model. By analyzing the relationship between different design parameters (such as coating temperature, gas flow rate, deposition rate, etc.) and defect data, the training model identifies which process parameters are closely related to coating quality defects. Using these key features, the machine learning model will be able to predict the quality defects of the coating layer under different process parameter settings. After the model training is completed, the process parameters are adjusted and optimized based on the optimization algorithm (such as genetic algorithm, particle swarm optimization algorithm, etc.) to reduce or eliminate the quality defects of the coating layer, and then optimize the coating design. The obtained substrate coating design optimization data includes the optimized process parameters, improved coating quality and the corresponding defect reduction rate.

[0192] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A quality control method for PVD coating design based on machine learning, characterized in that: The following steps are involved: Step S1: obtaining a coating substrate object; performing a PVD coating design according to the coating substrate object, and performing a substrate PVD coating simulation based on the substrate PVD coating design to obtain substrate PVD coating simulation data; Step S2: Analyze the mechanical coupling behavior data of the substrate PVD coating simulation data, and analyze the multi-scale mechanical properties of the coating material based on the substrate PVD coating design and the mechanical coupling behavior data, and use the multi-scale mechanical properties of the coating material to perform a dynamic stress response test of the coating layer to generate dynamic stress response data of the coating layer; Step S3: evaluating the chemical stability of the substrate coating based on the substrate PVD coating simulation data, and predicting the failure of the substrate coating layer based on the chemical stability of the substrate coating and the dynamic stress response data of the coating layer, to obtain the failure data of the substrate coating layer; Step S4: performing a substrate coating quality analysis based on the coating layer dynamic stress response data and the substrate coating layer failure data, and performing substrate coating design optimization on the substrate coating quality to obtain substrate coating design optimization data.

2. The quality control method of PVD coating design based on machine learning according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: obtaining a coating substrate object; Step S12: performing surface quality detection of the coated substrate according to the coated substrate object, thereby obtaining surface quality data of the coated substrate; Step S13: performing a coating requirement analysis of the coating substrate object based on the coating substrate surface quality data and the coating substrate object to obtain the coating requirement of the substrate object; Step S14: performing PVD coating design according to the coating requirements of the substrate object and the surface quality data of the coating substrate to obtain substrate PVD coating design data; Step S15: performing substrate PVD coating simulation based on the substrate PVD coating design data to obtain substrate PVD coating simulation data.

3. The quality control method of PVD coating design based on machine learning according to claim 2, characterized in that: Step S13 includes the following steps: Step S131: extracting the geometric structure of the coated substrate according to the coated substrate object, thereby obtaining the geometric structure data of the coated substrate; Step S132: analyzing the surface characteristics of the coating substrate according to the surface quality data of the coating substrate to obtain surface characteristic data of the coating substrate; Step S133: dividing the coated substrate based on the geometric structure data of the coated substrate to obtain a flat substrate and a curved substrate; Step S134: calculating the substrate coating adhesion ability according to the coating substrate surface characteristic data to obtain the substrate coating adhesion ability data; Step S135: estimating the difficulty of coating the flat substrate according to the substrate coating adhesion ability data to obtain the difficulty data of coating the flat substrate; Step S136: performing a curved surface substrate coating complexity analysis on the curved surface substrate according to the substrate coating adhesion ability data to obtain curved surface substrate coating complexity data; Step S137: performing coating requirement analysis on the coating substrate object according to the coating complexity data of the curved substrate and the coating difficulty data of the flat substrate to obtain the coating requirement of the substrate object.

4. The quality control method of PVD coating design based on machine learning according to claim 3 is characterized in that: Step S136 includes the following steps: Step S1361: extracting irregularities of the curved surface substrate according to the curved surface substrate to obtain irregularities of the curved surface substrate; Step S1362: Calculating the curvature of the curved substrate according to the irregularity of the curved substrate to obtain the curvature of the curved substrate; Step S1363: performing differential estimation of the deposition angle of the curved substrate based on the curvature and irregularity of the curved substrate to obtain differential deposition angle data; Step S1364: performing substrate sputtering deposition distribution difference detection according to the irregularity of the curved substrate to obtain substrate sputtering deposition distribution difference data; Step S1365: Calculating the heterogeneity of deposition rate on the substrate surface based on the substrate sputtering deposition distribution difference data and the deposition angle difference data to obtain the heterogeneity of deposition rate on the substrate surface; Step S1366: Performing a curved surface substrate coating complexity analysis on the substrate surface deposition rate heterogeneity according to the substrate coating adhesion ability data to obtain the curved surface substrate coating complexity data.

5. The quality control method of PVD coating design based on machine learning according to claim 1, characterized in that: The mechanical coupling behavior analysis described in step S2 includes: Conducting a coating material surface rigidity test on the substrate PVD coating simulation data to obtain coating material surface rigidity data; Calculate the surface friction coefficient of the coating material using the surface rigidity data of the coating material; Evaluate the wear resistance of the coating material based on the friction coefficient of the coating material surface; The wear state of the coating layer is estimated based on the surface friction coefficient of the coating material and the wear resistance of the coating material to obtain the wear state data of the coating layer; Use nano-indentation technology to test the fracture toughness of the coating material by measuring the surface rigidity of the coating material; Evaluate the external load capacity of the coating layer based on the fracture toughness of the coating material; The mechanical coupling behavior analysis is performed based on the external load capacity of the coating layer and the wear state data of the coating layer to obtain the mechanical coupling behavior data.

6. The quality control method of PVD coating design based on machine learning according to claim 1, characterized in that: The multi-scale mechanical property analysis of the coating layer described in step S2 includes: The thickness of the substrate coating layer is calculated using the substrate PVD coating design to obtain the thickness of the substrate coating layer; The substrate PVD coating design is used to analyze the microscopic particles of the substrate coating layer and obtain the microscopic particle data of the substrate coating layer; Extracting the microscopic particle distribution characteristics of the substrate in the coating layer of the substrate with a particle size distribution ranging from 10nm to 5000nm; The surface hardness of the coating layer to estimate the microscopic particle characteristics of the coating layer; The thickness of the substrate coating layer is measured with a measurement interval of 0.1-100 μm to obtain the thickness difference of the substrate coating layer; The stress concentration area in the coating layer is calculated by using the thickness difference of the substrate coating layer and the microscopic particle distribution characteristics of the substrate; Predict the impact resistance of the coating layer based on the mechanical coupling behavior data and the stress concentration area in the coating layer; The multi-scale mechanical properties of the coating layer are analyzed based on the surface hardness of the coating layer and the impact resistance of the coating layer to generate the multi-scale mechanical properties of the coating layer.

7. The quality control method of PVD coating design based on machine learning according to claim 1, characterized in that: The dynamic stress response test of the coating layer described in step S2 includes: The tensile rate of 1 mm / min was used to simulate the multi-scale mechanical properties of the coating layer and obtain the tensile data of the coating layer; Calculate the deformation degree of the coating layer from the tensile data of the coating layer; Calculate the probability of cracks in the coating layer when the deformation degree of the coating layer is 50nm; The elongation at break of the coating layer is estimated according to the deformation degree of the coating layer and the crack probability of the coating layer at a stress of 500 MPa, and the elongation at break of the coating layer is obtained; Analyze the coating layer strength of the coating layer elongation at break; The dynamic stress response test of the coating layer is carried out based on the strength of the coating layer and the elongation at break of the coating layer to obtain the dynamic stress response data of the coating layer.

8. The quality control method of PVD coating design based on machine learning according to claim 6, characterized in that: Step S3 includes the following steps: Step S31: performing substrate coating material detection on substrate PVD coating simulation data to obtain substrate coating material data; Step S32: evaluating the chemical stability of the substrate coating according to the substrate coating material data to obtain the chemical stability of the substrate coating; Step S33: performing corrosion resistance test on the substrate coating according to the chemical stability of the substrate coating to obtain corrosion resistance data of the substrate coating; Step S34: performing failure prediction of the substrate coating layer based on the corrosion resistance data of the substrate coating layer and the dynamic stress response data of the coating layer to obtain failure data of the substrate coating layer.

9. The quality control method of PVD coating design based on machine learning according to claim 8, characterized in that: Step S34 includes the following steps: Step S341: performing external impact simulation according to the dynamic stress response data of the coating layer to obtain external impact simulation data of the coating layer; Step S342: estimating the degradation fatigue trend of the coating layer based on the external impact simulation data of the coating layer to obtain degradation fatigue trend data of the coating layer; Step S343: performing corrosion environment simulation according to the corrosion resistance data of the substrate coating to obtain corrosion environment simulation data of the coating layer; Step S344: performing a corrosion extension estimation of the coating layer on the corrosion resistance data of the substrate coating based on the corrosion environment simulation data of the coating layer to obtain corrosion extension data of the coating layer; Step S345: estimating the coating layer shedding state according to the coating layer corrosion extension data and the coating layer degradation fatigue trend data to obtain coating layer shedding state data; Step S346: performing a failure estimation of the substrate coating layer according to the coating layer peeling state data and the coating layer corrosion extension data to obtain substrate coating layer failure data.

10. The quality control method of PVD coating design based on machine learning according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: evaluating the quality of the substrate coating based on the dynamic stress response data of the coating layer and the failure data of the substrate coating layer to obtain the substrate coating quality data; Step S42: performing a substrate coating quality defect analysis according to the substrate coating quality data to obtain substrate coating quality defect data; Step S43: Optimizing the substrate coating design based on the substrate coating quality defect data based on machine learning to obtain substrate coating design optimization data.

Images