A method for predicting mechanical properties of martensitic heat-resistant steel by analyzing EBSD detection data

Abstract

The application discloses a method for predicting the mechanical properties of martensitic heat-resistant steel by using EBSD detection data analysis, adopts area weighted average value to obtain corresponding EBSD data, and overcomes the geometric distortion problem of a traditional equivalent circle diameter method when processing non-equiaxed and slender lath martensite; the concept of martensite lath fitting length-diameter ratio is innovatively proposed, and through a large amount of experimental data analysis, the correlation between the concept and the mechanical properties of the martensitic heat-resistant steel is found; through a large amount of data, a prediction formula between the martensite lath fitting length-diameter ratio and the mechanical properties can be established, so that the mechanical properties of the final product can be accurately predicted by measuring the EBSD data of the sample in the subsequent production process.

Description

Technical Field

[0001] This invention belongs to the field of steel production quality control technology, specifically relating to a method for analyzing and predicting the mechanical properties of martensitic heat-resistant steel using EBSD test data. Background Technology

[0002] Martensitic heat-resistant steel, a key engineering material that achieves high-temperature service performance through microstructure control via heat treatment, has become one of the core materials in the fields of energy and power, aerospace, and high-end equipment manufacturing since its emergence in the early 20th century. Its core value lies in its ability to maintain excellent strength, oxidation resistance, and creep resistance at high temperatures of 600-650℃, significantly driving breakthroughs in ultra-supercritical power generation technology and the lightweight development of industrial equipment. Its historical evolution can be traced back to the late 19th century, when German metallurgist Adolf Martens first discovered martensite in hard minerals in 1895, later named by French scholar Osment in honor of his contribution. In 2025, China's independently developed G115 steel became the world's only martensitic heat-resistant steel that can be engineered for metal wall temperatures of 630-650℃, achieving a leap from technological catching up to international leadership.

[0003] The technical principle of martensitic heat-resistant steel is based on the non-diffusional phase transformation from austenite to martensite. When austenite is rapidly cooled to below the martensitic transformation temperature (Ms), carbon atoms form a supersaturated solid solution in α-Fe, and the crystal structure changes from face-centered cubic to body-centered tetragonal, forming a high-hardness microstructure. The microstructure is altered through composition design and heat treatment processes to balance strength and toughness. In terms of high-temperature performance, this material maintains high creep strength at 600-650℃ (e.g., the creep limit of G115 steel ≥120MPa after 100,000 hours), and its oxidation weight gain is ≤1.5g / m³. 2 Furthermore, by controlling the tempering temperature, an optimized match between hardness (HRC30-40) and impact toughness (KV20℃≥60J) is achieved to avoid brittle fracture. Therefore, the structural morphology and grain size of martensite largely determine the final product performance.

[0004] Due to the complex composition and high processing difficulty of G115 (or G115-like) steel, the original austenite grains are generally quite coarse (usually -2 or -1.5 grade), making it difficult to correlate post-metallographic corrosion observations with mechanical properties. Although some corrosion methods can reveal the martensitic lath structure, the mixed and interwoven martensitic laths make it difficult for metallographic methods to distinguish the orientation of the martensitic lath grains, and the size of the martensitic laths is difficult to statistically analyze. This leads to difficulties in statistically analyzing the morphology of the martensitic laths and the martensitic grain size, which cannot fully reflect the product quality. Therefore, electron backscatter diffraction (EBSD) technology is usually introduced in the structural analysis of martensitic steel.

[0005] Electron backscatter diffraction (EBSD) is a crystallographic analysis technique based on scanning electron microscopy (SEM). It determines the microscopic crystal structure of materials by analyzing the backscattered electron diffraction pattern (Kikuchi pattern) generated by the interaction of an electron beam with the sample. Originating from the first observation of the Kikuchi line in 1928, this technique was first applied to SEM in 1973 and automated data acquisition was achieved in the 1990s. EBSD boasts high spatial resolution (0.1 μm), high angular resolution (0.5°), and rapid acquisition capability (360,000 points / hour). Currently, it is widely used in grain orientation analysis, texture research, grain boundary property determination, and strain distribution measurement, providing crucial microscopic information for materials science. It is a core tool for understanding the relationship between the microstructure and macroscopic properties of materials, and widely serves materials research and development, quality control, and failure analysis.

[0006] Therefore, EBSD, through high-resolution crystal orientation imaging, can accurately characterize the crystal structure of martensite (such as body-centered cubic or body-centered orthorhombic phases) and its twin boundary characteristics, effectively distinguishing between low-angle and high-angle grain boundaries in martensite. EBSD provides quantitative orientation difference distribution data, offering crucial evidence for elucidating the martensitic phase transformation mechanism, microstructure evolution, and the correlation with mechanical properties. Compared to traditional X-ray diffraction, it offers simpler sample preparation, higher spatial resolution (up to 100 nm), and supports in-situ micro-area analysis, making it particularly suitable for rapid, non-destructive characterization of fine structures in martensitic steels. Currently, there are only six national standards (GB / T) and four ISO international standards for EBSD technology (with some overlap with national standards). Besides general technical guidance, phase identification, and grain size measurement, only standards concerning austenite structure are available.

[0007] Therefore, developing EBSD testing technology related to martensitic laths is crucial for the research, development, quality control, and failure analysis of martensitic heat-resistant steels. Summary of the Invention

[0008] The purpose of this invention is to provide a method for predicting the mechanical properties of martensitic heat-resistant steel by analyzing EBSD test data. During the production process, the mechanical properties of the final product can be accurately predicted by characterizing the microstructure of the sample. This method combines high efficiency, accuracy, and engineering applicability.

[0009] To achieve the above objectives, the technical solution of the present invention is as follows: A method for predicting the mechanical properties of martensitic heat-resistant steel using EBSD test data analysis includes the following steps: S1: Mechanical grinding and vibration polishing were performed on the surface of the martensitic heat-resistant steel sample. The vibration polishing time was 4-6 hours. S2: Perform EBSD detection on the sample. The collection area should include at least one original austenite grain boundary intersection region and one region located inside the original austenite grain. Based on the EBSD data, identify the grain boundaries and reconstruct the martensite lath grains. Perform ellipse fitting on each reconstructed grain to obtain the area-weighted average of the maximum diameter of the fitted ellipse, the area-weighted average of the maximum Freret diameter, and the area-weighted average of the aspect ratio of the fitted ellipse R*. If the ratio of the maximum diameter area-weighted average of the fitted ellipse of the reconstructed grain to the corresponding maximum Ferrette diameter area-weighted average is ≥70%, then the maximum diameter area-weighted average, maximum Ferrette diameter area-weighted average, and aspect ratio area-weighted average of the fitted ellipse in the collected EBSD data are considered valid. The major-to-minor axis ratio of the fitted martensitic lath is calculated using the area-weighted average (R*) of the fitted ellipse, as shown in the following formula: Martensitic lath fitting aspect ratio = R* 2 ; S3: Using the polynomial regression statistical method, a univariate m-degree polynomial regression equation was established between the aspect ratio of martensitic laths and the mechanical properties of martensitic heat-resistant steel. The formula is as follows: , Where: y represents the room temperature tensile strength, room temperature plastic elongation strength, room temperature impact absorbed energy, or 630℃ high temperature plastic elongation strength of martensitic heat-resistant steel; when y represents the room temperature tensile strength, room temperature plastic elongation strength, or 630℃ high temperature plastic elongation strength of martensitic heat-resistant steel, the unit is MPa; when y represents the room temperature impact absorbed energy, the unit is J. x is the aspect ratio of the martensitic lath fitting, which is dimensionless; b is a constant; S4: In subsequent testing, the fitted aspect ratio of the martensitic lath of the sample to be tested is obtained by EBSD detection. The fitted aspect ratio of the martensitic lath is then substituted into the pre-established quantitative correspondence between the aspect ratio of the martensitic lath and the mechanical properties of martensitic heat-resistant steel to calculate the predicted values ​​of one or more target mechanical properties.

[0010] Preferably, in step S1, the polishing liquid used for vibration polishing is a silica nano suspension.

[0011] Preferably, in step S2, the orientation difference threshold used when identifying grain boundaries is ≥15°.

[0012] Preferred, , in, d i For the first iThe maximum Ferrette diameter of each grain, the maximum diameter of the fitted ellipse, and the aspect ratio of the fitted ellipse are given in μm; the aspect ratio of the fitted ellipse is dimensionless. A i For the first i The area corresponding to each grain, in μm 2 ; n To reconstruct the total number of grains.

[0013] Preferably, in steps S4 and S5, the mechanical properties of the martensitic heat-resistant steel include at least one of room temperature tensile strength, room temperature plastic elongation strength, room temperature impact absorbed energy, and 630°C high temperature plastic elongation strength.

[0014] Furthermore, when the martensitic heat-resistant steel is G115 steel, the quantitative correspondence between the aspect ratio of the martensitic laths and the mechanical properties of G115 steel is as follows: Tensile strength at room temperature: y1 = 0.2x 2 +0.93x+662, unit MPa, x is the aspect ratio of the martensitic lath fitting, dimensionless; The room temperature ductile elongation strength y2 = 0.06x 2 +5.17x+484, unit MPa, x is the aspect ratio of the martensitic lath fitting, dimensionless; The energy absorbed by the shock at room temperature is y3 = 0.38x. 2 -3.59x+357, unit J, x is the aspect ratio of the martensitic lath fitting, dimensionless; The high-temperature plastic elongation strength at 630℃ is y4 = 0.17x. 2 +1.69x+91, unit MPa, x is the aspect ratio of the martensitic lath fitting, dimensionless.

[0015] The grain size of lath martensite is not a single scalar quantity, but a composite parameter determined by its three-level structure of laths, blocks, and bundles. Its macroscopic mechanical response (including the evolution of strength and toughness) is mainly regulated by the synergistic control of lath length, cross-sectional width, orientation difference distribution, and three-dimensional spatial arrangement. Although the current national standard GB / T 36165–2018 is based on electron backscatter diffraction (EBSD) technology for microscopic analysis, its statistical framework still adheres to a simplified model of equivalent circle diameter, which makes it impossible to effectively characterize the extension directionality and true three-dimensional geometric features of the laths. Numerous studies (e.g., Morris et al., ActaMaterialia, 2003; Kitahara et al., Materials Science and Engineering A, 2006) have shown that the aspect ratio of martensitic laths typically ranges from 5:1 to 20:1. Their strengthening mechanism primarily stems from dislocation pile-up at high-angle grain boundaries, and grain boundary density exhibits a significant positive correlation with lath length (correlation coefficient r > 0.85). This provides crucial evidence for understanding the intrinsic relationship between lath structure and mechanical properties. Therefore, characterizing lath solely by area or equivalent diameter severely underestimates the scale of the actual load-bearing unit. Furthermore, typical analyses focus only on grain size, failing to accurately reflect the morphology of martensitic laths.

[0016] Due to the high anisotropy of lath martensite, it can more realistically and accurately reflect the essential state of its grain size and morphology, presenting it in a more intuitive and physically meaningful way, which can significantly improve the correlation between grain characterization and material mechanical properties (such as yield strength, impact toughness, high-temperature performance, etc.).

[0017] Grain evaluation in national or international standards is mostly based on equivalent circle diameter, intercept method, or area averaging method. These methods are effective when dealing with equiaxed grains, but when faced with highly non-equiaxed, slender, and orientation-sensitive lath martensite, they often lose key structural information due to over-simplification. Furthermore, because lath martensite is composed of numerous parallel, fine laths separated by subgrain boundaries (substructural interfaces), this unique morphology significantly enhances the material's toughness—subgrain boundaries effectively hinder crack propagation paths, requiring cracks to bypass multiple laths to propagate, thus suppressing brittle fracture. Simultaneously, grain size (specifically lath width and martensite block size) plays a decisive role in strength: finer laths, by increasing subgrain boundary density, strengthen the fine-grain strengthening effect (following the Hall-Petch relationship), significantly improving yield strength and hardness. This invention achieves high-fidelity quantification of the martensite microstructure by introducing a geometric parameter system that more closely matches the actual morphology and combining it with crystallographic boundary identification criteria.

[0018] Furthermore, conventional metallographic observation is limited to two-dimensional cross-sections, while the interlacing, stacking, and penetrating arrangement of laths in three-dimensional space often results in what appears to be "small grains" on the surface actually being truncated fragments of larger laths. Regardless of whether the laths in the EBSD image are randomly arranged ( Figure 1 , Figure 2 ), parallel arrangement ( Figure 3 ) or staggered stacking ( Figure 4 These localized small regions can significantly interfere with statistical results based on arithmetic means, leading to systematic underestimation. Although three-dimensional grain size reconstruction (such as through focused ion beam SEM (FIB-SEM) tomography or X-ray microtomography) can partially solve this problem, a single detection can take tens of hours, the equipment cost exceeds ten million yuan, and the data processing is complex, making it difficult to apply to production line-level quality control. Furthermore, the method of this invention has no special requirements for the coordinate axes (usually represented by x, y, z) of the inverse pole figure (IPF diagram). Figure 1 x-axis Figure 2 , Figure 3 , Figure 4 All axes are z-axis, which have no effect on the results.

[0019] The first aspect of this invention improves the method for calculating grain size.

[0020] If we follow the arithmetic mean and standard deviation method specified in the national standard GB / T 36165–2018, even after correction by combining the maximum Freret diameter or the maximum diameter of the fitted ellipse, the resulting coefficient of variation (CV), although lower than when using the equivalent circle diameter, is still at a high level—measured data show that the CV is still close to 170%. This indicates that the dispersion of grain size data is still significant, thus weakening the correlation between martensitic grain size and the macroscopic mechanical properties of the material (see Table 1, the result is based on the statistical average of 100 samples), reflecting that small grains still have a large interference with the overall statistical results.

[0021]

[0022] To address the aforementioned issues, this invention proposes using the area-weighted mean as the core indicator for grain size evaluation, and its specific calculation formula is as follows: , in, d i For the first i The maximum Feret diameter of each grain, the maximum diameter of the fitted ellipse, and the aspect ratio of the fitted ellipse are described. d i No. iThe maximum Ferrette diameter of each grain and the maximum diameter of the fitted ellipse are in μm. d i For the first i The aspect ratio of the fitted ellipse of each grain is dimensionless; A i For the first i The area corresponding to each grain, in μm 2 ; n To reconstruct the total number of grains.

[0023] This invention significantly reduces statistical fluctuations caused by a large number of tiny grains (mainly from cross-sectional slices of three-dimensional lath structures) by assigning higher weights to larger grains, thus more accurately characterizing the overall size distribution of martensitic laths. As shown in Table 1, after implementing area weighting, the coefficient of variation is significantly reduced to below 40%, and both data centrality and statistical reliability are significantly optimized. This scheme not only conforms to the weighted statistical specifications for non-equiaxed grains in the ASTM E112 appendix, but also highly echoes the "effective grain size" theory in cutting-edge international research (such as Gao et al., Scripta Materialia, 2020).

[0024] Table 1 compares the effects of different fitting algorithms and the selection of fitting data on the final martensitic lath grain size (100 sets of data). The maximum diameter of the fitted ellipse (area-weighted) has the lowest coefficient of variation, which means that the data dispersion is low, the reliability is stronger, and the data regression is more accurate.

[0025] The second aspect of this invention optimizes the EBSD detection scheme.

[0026] Typically, EBSD testing requires random selection of multiple locations, and the accuracy of the test results is strongly correlated with the number of selected locations. This invention exposes characteristic locations during sample preparation, thereby reducing the amount of EBSD testing, increasing data utilization, and enabling better performance prediction. To reduce the impact of pitting corrosion on EBSD results, this invention does not recommend using electrolytic polishing for polishing martensitic heat-resistant steel. The overall sample preparation method still follows the specifications in GB / T 13298-2015. However, in the final step of vibratory polishing of martensitic heat-resistant steel using OP-S (silica nano-suspension polishing slurry), the time is set at 4-6 hours, longer than the commonly recommended vibratory polishing time for hard materials (2 hours). OP-S (silica nano-suspension polishing slurry) is typically alkaline with a pH of around 10, and prolonged polishing can have a certain corrosive effect on the proto-austenite grain boundaries in the martensitic structure. However, this effect is relatively weak and unlikely to affect the interlamellar grain boundaries of martensite, nor will it affect the Kikuchi pattern and calibration rate in EBSD analysis. It can also eliminate surface stress, making the grain structure more realistic.

[0027] Extensive experimental research has shown that controlling the polishing time to 4-6 hours maximizes the advantages and minimizes the disadvantages, facilitating the selection of subsequent EBSD detection areas. Vibratory polishing exposes the original austenite grain boundaries, making them easier to locate under electron microscopy and thus facilitating the selection of suitable detection locations during EBSD. Note that while Nital (nitric acid alcohol etchant) can also expose the original austenite grain boundaries, it affects the accuracy of EBSD data. Eliminating this effect requires very precise control of the etching time; however, the etching effect is directly related to the grain orientation and elemental distribution on the polished surface of each sample, making precise time control difficult. Although it can be considered when equipment limitations exist, it is not recommended here. Although vibratory polishing is time-consuming, it allows for the simultaneous processing of up to 20 samples (approximately 12-18 minutes per sample) and requires no manual supervision, making it a highly recommended polishing method.

[0028] Simultaneously, the area-weighted average of the maximum Freret diameter must be used as a reference. The ratio of the maximum diameter to the maximum Freret diameter of the fitted ellipse must be ≥70%. A ratio less than 70% indicates that the fitted ellipse cannot reflect the true martensite lath morphology. A ratio that is too small indicates that the fitted ellipse is too short and too wide compared to the original martensite lath, which may be due to the following three reasons: 1. If the step size is too large, the edges of the martensite laths will not be clear enough, the orientation difference will not accurately reflect the morphology, and the eccentricity of the fitted ellipse will be incorrectly reduced.

[0029] 2. The EBSD detection area was too small, and there were too many boundary grains, which prevented the martensite laths from being completely encapsulated, resulting in a smaller eccentricity of the fitted ellipse.

[0030] 3. In EBSD analysis, if the minimum grain boundary angle is selected too low (usually 15°), some subgrain boundaries will be identified as regular grain boundaries. This will cause the subgrain boundaries to be excluded when fitting the ellipse (and identified as individual grains), resulting in discontinuity of the laths and a smaller error in the eccentricity of the fitted ellipse.

[0031] During EBSD testing, parameter selection follows the GB / T 19501-2013 standard. The step size should be 1 / 20 of the average width of the martensite laths to maximize the reproduction of the martensite lath morphology. EBSD region selection includes two areas: localized austenite grain boundary junctions and regions within the same austenite grain. The region range is selected according to GB / T 36165-2018. Junction regions, being at the beginning or end of the lattice shear process, are prone to lattice distortion, which has a greater impact on the final performance. Regions within the same austenite grain can be used to characterize the uniformity of grain morphology after martensite transformation, leading to more accurate predictions of mechanical properties. When selecting the range, the calculation of the number of grains within the range must exclude boundary grains. Boundary grains do not contain all martensite laths, and when fitting an ellipse, the eccentricity will not accurately reflect the martensite lath morphology. Grain size calculations also need to exclude boundary grains to ensure the accuracy of the results.

[0032] The third aspect of this invention is an improvement in the field of mechanical property testing of metallic materials.

[0033] Traditional methods have long relied on sampling from multiple locations and standard mechanical testing of samples, but their limitations are becoming increasingly apparent.

[0034] First, the sample preparation process is significantly affected by the processing quality. For example, tiny deviations in the cutting, grinding, or heat treatment processes can directly change the surface microstructure, leading to systematic shifts in mechanical test data such as tensile and impact tests.

[0035] Secondly, human factors such as sample positioning, fixture installation, and differences in the experience of testing personnel further amplify the uncertainty of the results, making it difficult to meet industry standards for repeatable testing. This is especially true for high-value-added materials like martensitic heat-resistant steel, where their mechanical properties (such as strength and toughness) are strongly correlated with their microstructure (such as the size, distribution, and orientation of martensitic laths). However, traditional testing methods only focus on macroscopic performance indicators and cannot incorporate microscopic characteristics into the real-time evaluation system. This leads to a disconnect between test results and actual product characteristics; for example, the impact of lath size variations on yield strength is ignored, resulting in distorted quality assessments.

[0036] More importantly, mechanical performance testing is usually conducted as a final step after product processing is completed, making it impossible to predict quality during the production process. Once performance is found to be substandard, rework or scrapping is often necessary, which not only increases costs but also delays delivery cycles, creating a passive "post-event remediation" situation and severely hindering the implementation of efficient manufacturing models.

[0037] This invention proposes the concept of fitting the aspect ratio of martensitic laths. Due to the random angle of martensitic laths, measuring the aspect ratio of martensitic laths is very difficult. If the minor axis length of the fitted ellipse is simply used, the result will deviate greatly from the actual result and is insufficient for finer martensitic laths (such as those with an aspect ratio of around 1:20). However, considering different situations, the calculation is difficult in actual operation. Therefore, this invention uses the aspect ratio of the fitted martensitic laths, i.e., the nominal diameter of the fitted ellipse / the maximum diameter of the fitted ellipse, as a parameter to characterize the size and morphology of martensitic lath bundles in martensitic heat-resistant steels (G115, P92, etc.).

[0038] Martensitic lath fitting aspect ratio = D / L, where D is the maximum diameter of the fitted ellipse and L is the nominal diameter of the fitted ellipse; Fitted ellipse diameter L=2b 2 / a, where b is the minor radius of the fitted ellipse, b=a / R*; a is the major radius of the fitted ellipse, a=1 / 2D, and D is the maximum diameter of the fitted ellipse; therefore, the bore diameter of the fitted ellipse L=D / R*. 2 ; Finally, the martensitic lath fitting aspect ratio is calculated as R * 2 .

[0039] This invention, for the first time, established a polynomial regression statistical relationship (goodness of fit R² > 0.85) between the aspect ratio of martensitic laths and mechanical indices such as yield strength, tensile strength, and impact toughness at both room and high temperatures through cross-validation using large-scale experimental data. This breakthrough correlation shifts the detection process from "indirect inference" to "direct mapping," significantly simplifying the quality assessment process. At the implementation level, the core of sample preparation utilizes fully automated vibration polishing technology. This technology achieves uniform surface treatment through a precision vibration system, minimizing fluctuations caused by human operation. EBSD testing itself is characterized by high precision and non-destructive nature, avoiding the dependence on equipment in traditional mechanical testing and reducing measurement errors caused by outdated equipment, calibration deviations, or maintenance downtime.

[0040] This invention does not significantly increase the requirements for EBSD detection technology. The operational requirements are only redefined in sample preparation and region selection, and this redefined approach does not significantly increase preparation time. The preparation method can provide information for region selection, and in fact, it reduces the risk of insufficient grain quantity due to selecting unknown regions. No higher requirements, special requirements, special data analysis software, or special learning process are needed for existing operators; the size and morphology analysis of martensitic lath grains can be obtained simply through standard data analysis methods and data linkage.

[0041] Compared with the prior art, the beneficial effects of the present invention are as follows: First, this invention uses the concept of area-weighted average to obtain the maximum diameter, maximum Freret diameter, and the ratio of the major and minor axes of the fitted ellipse. This overcomes the geometric distortion problem of the traditional equivalent circle diameter method when dealing with non-equiaxed, slender lath martensite, making the grain size statistical results more closely match the true microstructure of the material. As a result, the coefficient of variation (CV) of the grain size statistical results is significantly reduced from more than 300% in the traditional method to less than 40%, greatly reducing data dispersion and improving the reliability and repeatability of the results.

[0042] Second, the EBSD detection method is optimized by extending the exposure of the original austenite grain boundaries, ensuring that the sampling area includes at least one original austenite grain boundary intersection region and one region located inside the original austenite grain, and by introducing a requirement that the maximum diameter of the fitted ellipse / maximum Freret diameter must be ≥70% to ensure the accuracy of the detection results.

[0043] Third, this invention creatively proposes the concept of martensitic lath fitting aspect ratio and studies and discovers that the martensitic lath fitting aspect ratio obtained by using area-weighted average value has a certain correlation with several key mechanical properties of the material, such as room temperature tensile strength, yield strength, high temperature yield strength and impact toughness. Through a large amount of data, a predictive formula between the martensitic lath fitting aspect ratio and various mechanical properties can be established. Thus, in the subsequent production process, the mechanical properties of the final product can be accurately predicted by measuring the EBSD data of the sample.

[0044] The core advantage of the method described in this invention lies in moving quality control forward to the production process. In an actual production line, samples can be extracted as needed, and slat parameters can be quickly obtained through EBSD scanning. These parameters are then substituted into a preset linear model to predict in real time whether the mechanical properties meet the standards. If the data is within the acceptable range, the product can proceed to the next process; if it deviates from the threshold, the system immediately triggers process adjustments (such as optimizing heat treatment parameters), avoiding the remedial difficulties after finished product inspection. This not only advances the quality warning time to the middle of the processing stage but also significantly reduces the risk of production interruption—in the traditional model, equipment maintenance and metrological calibration often lead to downtime of 1-2 days, while this method minimizes such impacts by reducing reliance on large mechanical equipment. Its high efficiency stems from the industrial adaptability of the model: the linear relationship is simple and easy to use, requiring no complex algorithms, and operators can master it after short-term training, meeting the "plug-and-play" digital transformation needs of the manufacturing industry.

[0045] This invention, combined with EBSD data analysis using artificial intelligence, will further optimize model accuracy, providing a replicable solution for large-scale quality control of lath martensitic steel, with value far exceeding that of single material applications. It promotes a paradigm shift in materials science from empirical correlation of "performance-microstructure" to quantitative prediction of "structure-performance," setting a methodological benchmark for online detection of other high-performance alloys (such as austenitic stainless steel and titanium alloys). It achieves an intelligent quality closed loop across the entire production chain. In the context of increasingly stringent resource constraints, this technology, by reducing trial-and-error costs and improving material utilization, not only creates direct economic benefits for enterprises but also helps the manufacturing industry upgrade towards green and efficient directions, possessing profound industrial promotion value. Attached Figure Description

[0046] Figure 1 EBSD scan of martensitic heat-resistant steel; Figure 2 EBSD scan of martensitic heat-resistant steel; Figure 3 EBSD scan of martensitic heat-resistant steel; Figure 4 This is an EBSD scan of martensitic heat-resistant steel. Detailed Implementation

[0047] The present invention will be further described below with reference to the embodiments and accompanying drawings. The following embodiments will help those skilled in the art to further understand the present invention, but do not limit the present invention in any way.

[0048] The EBSD data acquisition described in this invention relies on a multi-source device platform, covering probe systems from mainstream manufacturers such as Bruker, Oxford Instruments, and EDAX.

[0049] To eliminate the impact of device specificity on data and ensure objectivity, the data in this invention underwent systematic cross-validation in a cross-platform environment, covering EBSD devices widely used in industry. Data processing strictly followed several national standards mentioned above, employing third-party data processing software from different vendors for post-processing and statistical analysis. Each set of EBSD data was independently validated using heterogeneous software combinations, meaning the data acquisition equipment and processing software were unrelated in terms of vendor origin, effectively avoiding systematic biases or algorithmic preferences caused by specific hardware and software coupling. All EBSD data met the technical requirements of GB / T36165-2018, and the post-processing workflow strictly followed the standard, including the minimum confidence index (CI) threshold setting (typically 0.3), large-angle grain boundary definition (orientation difference ≥15°), grain merging rules (based on grain boundary type and size), and statistical sampling area specifications (recommended ≥0.1 mm²). This method ensured the repeatability and standardization of the data generation process, avoiding evaluation distortion caused by a single device or software, and providing high-confidence input data for subsequent microstructure-mechanical property correlation analysis.

[0050] I. Reliability Analysis of the Fitting Degree of the Sample Ellipse All martensitic steels of this invention were vibratory polished using OP-S for 4-6 hours (randomly) to clearly reveal the original austenite grain boundaries under electron microscopy, ensuring that there was no under-polishing or over-polishing within this range. During electron backscatter diffraction (EBSD) analysis, the step size was measured at 1 / 20, 1 / 30, and 1 / 50 of the martensite lath width (estimated by the average value of multiple manual measurements after coarse EBSD scanning) to ensure the accuracy of the parameters provided by this invention. Table 2 summarizes the reliability of the fitted ellipse fit for random samples at different step sizes, i.e., the ratio of the maximum diameter to the maximum Feretta diameter of the fitted ellipse is ≥70%, with a small standard deviation and a very low coefficient of variation.

[0051]

[0052] II. Establishing the relationship between mechanical properties and the aspect ratio of martensitic laths This invention takes G115 martensitic heat-resistant steel as an example. Table 3 shows the maximum diameter of the fitted ellipse and mechanical properties of randomly selected martensitic heat-resistant steel. From the data analysis in Table 3, the overall conclusion is that there is no significant correlation between the maximum diameter of the fitted ellipse and the mechanical properties of the martensitic heat-resistant steel, because the maximum diameter of the fitted ellipse is only related to the grain size in the EBSD detection area and is not related to the morphology of the martensite laths.

[0053]

[0054] Therefore, this invention introduces the fitting ellipse diameter value L=2b. 2 / a, where b is the minor radius of the fitted ellipse, b=a / R*; a is the major radius of the fitted ellipse, a=1 / 2D, and D is the maximum diameter of the fitted ellipse; therefore, the bore diameter of the fitted ellipse L=D / R*. 2 ; Martensitic lath fitting aspect ratio = D / L, where D is the maximum diameter of the fitted ellipse and L is the nominal diameter of the fitted ellipse; Finally, the martensitic lath fitting aspect ratio is calculated as R * 2 .

[0055] This invention, through regression analysis of random batches of G115 martensitic heat-resistant steel data, yielded the following relationships between the fitted martensitic lath aspect ratio (x) and four mechanical properties: room temperature tensile strength (y1), room temperature specified plastic elongation strength (Rp0.2) (y2), high temperature specified plastic elongation strength (Rp0.2) (y3), and impact absorbed energy (KV2) (y4) at 20℃: Tensile strength at room temperature: y1 = 0.2x 2 +0.93x+662, unit MPa, x is the aspect ratio of the martensitic lath fitting, dimensionless; The room temperature ductile elongation strength y2 = 0.06x 2 +5.17x+484, unit MPa, x is the aspect ratio of the martensitic lath fitting, dimensionless; The energy absorbed by the shock at room temperature is y3 = 0.38x. 2 -3.59x+357, unit J, x is the aspect ratio of the martensitic lath fitting, dimensionless; The high-temperature plastic elongation strength at 630℃ is y4 = 0.17x. 2 +1.69x+91, unit MPa, x is the aspect ratio of the martensitic lath fitting, dimensionless.

[0056] Test Example 1 In this embodiment, G115 martensitic heat-resistant steel (20 samples) produced in the same batch was used. The linear fitting relationship between the aspect ratio of the martensitic laths and four conventionally tested mechanical properties was utilized to statistically calculate the error between the calculated and measured values. The error value was obtained by dividing the absolute value of the difference between the calculated and measured values ​​by the measured value. The test results are shown in Table 4.

[0057]

[0058] Test Example 2 In this embodiment, G115 martensitic heat-resistant steel (20 samples) produced in different batches was used. The linear fitting relationship between the aspect ratio of martensitic laths and four conventionally tested mechanical properties was utilized. The calculated error between the calculated and measured values ​​was statistically analyzed. The error value was obtained by dividing the absolute value of the difference between the calculated and measured values ​​by the measured value. The test results are shown in Table 5.

[0059]

[0060] Test Example 3 In this embodiment, 30 samples of G115 martensitic heat-resistant steel were randomly selected. The linear fitting relationship between the aspect ratio of martensitic laths and four conventionally tested mechanical properties was utilized. The calculated error between the calculated and measured values ​​was then statistically analyzed. The error was calculated as the absolute value of the difference between the calculated and measured values ​​divided by the measured value. The test results are shown in Table 6.

[0061]

[0062] As can be clearly seen from Tables 4 to 6, the linear relationship between the aspect ratio of the martensitic laths and the mechanical properties in Examples 1 to 3 is good, and the error rate of the fitting curve prediction results is small (all data are below 10%, and most are below 5%). Adjusting the linear relationship coefficient can more accurately predict the mechanical properties, and the overall evaluation relationship is clearer than that of other methods.

[0063] In summary, the results obtained by fitting the aspect ratio of martensitic laths (area-weighted average method) of this invention have a strong correlation with the mechanical properties of G115 steel, and the prediction results are relatively accurate, which can be used for quality management.

[0064] Meanwhile, the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention, such as changing the linear relationship coefficients, without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications and substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for analyzing and predicting the mechanical properties of martensitic heat-resistant steel using EBSD test data, characterized in that, Includes the following steps: S1: Vibratory polishing of the surface of the martensitic heat-resistant steel sample for 4-6 hours. S2: Perform EBSD detection on the sample. The collection area should include at least one original austenite grain boundary intersection region and one region located inside the original austenite grain. Based on the EBSD data, identify the grain boundaries and reconstruct the martensite lath grains. Perform ellipse fitting on each reconstructed grain to obtain the area-weighted average of the maximum diameter of the fitted ellipse, the area-weighted average of the maximum Freret diameter, and the area-weighted average of the aspect ratio of the fitted ellipse R*. If the ratio of the maximum diameter area-weighted average of the fitted ellipse of the reconstructed grain to the corresponding maximum Ferrette diameter area-weighted average is ≥70%, then the maximum diameter area-weighted average, maximum Ferrette diameter area-weighted average, and aspect ratio area-weighted average of the fitted ellipse in the collected EBSD data are considered valid. The major-to-minor axis ratio of the fitted martensitic lath is calculated using the area-weighted average (R*) of the fitted ellipse, as shown in the following formula: Martensitic lath fitting aspect ratio = R* 2 ; S3: Using the polynomial regression statistical method, a univariate m-degree polynomial regression equation was established between the aspect ratio of martensitic laths and the mechanical properties of martensitic heat-resistant steel. The formula is as follows: ， in: y represents the room temperature tensile strength, room temperature plastic elongation strength, room temperature impact absorbed energy, or 630℃ high temperature plastic elongation strength of martensitic heat-resistant steel; when y represents the room temperature tensile strength, room temperature plastic elongation strength, or 630℃ high temperature plastic elongation strength of martensitic heat-resistant steel, the unit is MPa; when y represents the room temperature impact absorbed energy, the unit is J. x is the aspect ratio of the martensitic lath fitting, which is dimensionless; b is a constant; S4: In subsequent testing, the fitted aspect ratio of the martensitic lath of the sample to be tested is obtained by EBSD detection. The fitted aspect ratio of the martensitic lath is then substituted into the pre-established quantitative correspondence between the aspect ratio of the martensitic lath and the mechanical properties of martensitic heat-resistant steel to calculate the predicted values ​​of the mechanical properties of one or more samples to be tested.

2. The method for analyzing and predicting the mechanical properties of martensitic heat-resistant steel using EBSD test data as described in claim 1, characterized in that, In step S1, the polishing fluid used for vibration polishing is a silica nano suspension.

3. The method for analyzing and predicting the mechanical properties of martensitic heat-resistant steel using EBSD test data as described in claim 1, characterized in that, In step S2, EBSD testing is performed according to national standards, preferably GB / T 36165-2018.

4. The method for analyzing and predicting the mechanical properties of martensitic heat-resistant steel using EBSD test data as described in claim 1, characterized in that, In step S2, the orientation difference threshold used when identifying grain boundaries is ≥15°.

5. The method for analyzing and predicting the mechanical properties of martensitic heat-resistant steel using EBSD test data as described in claim 1, characterized in that, ， in, d i For the first i The maximum Ferrette diameter of each grain, the maximum diameter of the fitted ellipse, and the aspect ratio of the fitted ellipse are given in μm; the aspect ratio of the fitted ellipse is dimensionless. A i For the first i The area corresponding to each grain, in μm 2 ; n To reconstruct the total number of grains.

6. The method for analyzing and predicting the mechanical properties of martensitic heat-resistant steel using EBSD test data as described in claim 1, characterized in that, In steps S3, S4 and S5, the mechanical properties of the martensitic heat-resistant steel include at least one of room temperature tensile strength, room temperature plastic elongation strength, room temperature impact absorbed energy, and 630℃ high temperature plastic elongation strength.

7. The method for analyzing and predicting the mechanical properties of martensitic heat-resistant steel using EBSD test data as described in claim 1, characterized in that, In step S3, when the martensitic heat-resistant steel is G115 steel, the quantitative correspondence between the aspect ratio of the martensitic laths and the mechanical properties of G115 steel is as follows: Tensile strength at room temperature: y1 = 0.2x 2 + 0.93x + 662, unit MPa, x is the aspect ratio of the martensitic lath fitting, dimensionless; The room temperature ductile elongation strength y2 = 0.06x 2 + 5.17x + 484, unit MPa, x is the aspect ratio of the martensitic lath fitting, dimensionless; The energy absorbed by the shock at room temperature is y3 = 0.38x. 2 - 3.59x + 357, unit J, x is the aspect ratio of the martensitic lath fitting, dimensionless; The high-temperature plastic elongation strength at 630℃ is y4 = 0.17x. 2 + 1.69x + 91, unit MPa, x is the aspect ratio of the martensitic lath fitting, dimensionless.

Images