A high-throughput method for determining and predicting martensitic transformation start temperature

Abstract

The application provides a high-throughput determination and prediction method for martensite transformation starting temperature, and belongs to the technical field of steel material detection, and comprises the following steps: treating a sample; performing austenitizing treatment on the treated sample; recording the phase change process of the sample from austenite to martensite in a cooling process by using a high-temperature confocal laser scanning microscope; frame by frame analyzing the critical temperature at which each grain is first transformed into martensite, and obtaining the martensite transformation starting temperature of different grain sizes; after the sample is treated, the depth of grain boundary erosion is not less than 3 µm, and the width of grain boundary erosion is not greater than 1 µm. The HT-CLSM is used to record the martensite transformation starting temperature of different grain sizes, and in order to realize high-throughput measurement of different grain sizes and avoid mutual influence between different grains, corresponding requirements are proposed for the treatment of the sample.

Description

Technical Field

[0001] The invention belongs to the technical field of steel material detection, and in particular relates to a high-throughput determination and prediction method of martensitic transformation starting temperature. Background Art

[0002] Martensitic steels, due to their high strength, hardness, and excellent overall mechanical properties, are widely used as high-performance structural materials in applications such as engineering machinery, molds, aerospace, and the automotive industry. During heat treatment, the Ms temperature is a key parameter that determines the final microstructure and properties. The Ms temperature is influenced by multiple factors, among which the prior austenite grain size (PAGS) has a significant influence. Generally, larger PAGS values ​​are associated with lower Ms values, which in turn influences microstructure distribution and mechanical properties. However, this relationship is governed by multiple factors, including composition, heat treatment parameters, and microstructure, making it difficult to accurately describe using simple empirical formulas.

[0003] Currently, methods for determining Ms primarily include metallography, differential thermal analysis, magnetometry, dilatometry, and high-temperature in-situ observation techniques. HT-CLSM experiments, owing to their advantages of high-temperature in-situ analysis, high resolution, and precise temperature control, have been increasingly applied to studying the austenite-martensite phase transformation process. However, traditional studies often employ single-grain-scale samples and conduct experiments one by one. This results in low efficiency, few data points, and weak statistical analysis, making it difficult to reveal the complex nonlinear relationship between PAGS and Ms. Furthermore, conventional regression methods have significant limitations in data processing, making it difficult to accurately fit complex data patterns.

[0004] In recent years, high-throughput experimental design and machine learning methods have shown great potential in materials design and microstructure property prediction. High-throughput experiments can acquire a large amount of valid variable data in a single experiment, while machine learning, particularly global optimization methods such as genetic algorithms (GAs), has been widely used to model complex nonlinear relationships and perform high-dimensional parameter regression.

[0005] However, there are no existing literature or technical solutions that systematically integrate high-throughput grain structure design, high-temperature laser in-situ observation, and machine learning modeling techniques for quantitatively determining the relationship between PAGS and Ms in martensitic steels. Therefore, there is an urgent need to develop a method with high experimental efficiency, high measurement accuracy, and strong modeling capabilities to reveal and predict the onset behavior of martensitic phase transformation under different grain structure conditions, providing reliable technical support for heat treatment process optimization and microstructure and property control. Summary of the Invention

[0006] To address these issues, the present invention provides a high-throughput method for measuring and predicting the martensitic transformation start temperature. This method uses HT-CLSM to record the martensitic transformation start temperature of grains of varying sizes. To achieve high-throughput measurement of grain sizes and avoid interactions between different grains, corresponding requirements are placed on sample preparation.

[0007] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0008] The present invention provides a high-throughput determination method for the martensite transformation starting temperature, comprising: processing a sample; austenitizing the processed sample; recording the phase transformation process of the sample from austenite to martensite using a high-temperature confocal laser scanning microscope during cooling; analyzing the critical temperature of each grain at which it first transforms into martensite frame by frame to obtain the martensite transformation starting temperature of different grain sizes; after the sample is processed, the depth of grain boundary erosion is not less than 3µm, and the width of grain boundary erosion is not more than 1µm.

[0009] Furthermore, processing the sample includes grinding, primary polishing, etching and secondary polishing.

[0010] Furthermore, the etching uses two etching agents, the first etching agent is a mixture of picric acid and alcohol, wherein the ratio of picric acid to alcohol is 4g-5g / 100ml; the second etching agent is a mixture of nitric acid and ethanol, wherein the volume proportion of nitric acid is 8%-10%.

[0011] Furthermore, the etching process is as follows: placing the sample in a first etching agent, keeping it in an ultrasonic oscillation environment for 10s-15s, and immediately placing it in anhydrous ethanol for cleaning and drying after completion; then placing the sample after the initial etching in a second etching agent, ultrasonically oscillating for 1s-2s, taking it out, removing excess etching agent from the surface of the sample, setting the observation surface upward and keeping it for at least 20s, repeating the etching in the second etching agent 3-4 times, and then cleaning and drying it after completion.

[0012] Furthermore, the secondary polishing adopts mechanical polishing with a rotation speed of 300rpm-600rpm, a pressure of 15N-20N, a diamond suspension with a particle size of 1μm-3μm, a polishing cloth of short-pile cloth, and polishing for at least 30min.

[0013] Furthermore, the austenitizing treatment process is: heating from room temperature to 180°C-220°C at 0.5°C / s-1.0°C / s; heating to 980-1050°C at 8°C / s-12°C / s; heating to 1200°C-1280°C at 1°C / s-3°C / s and keeping warm for 150s-200s to form a mixed grain structure.

[0014] Furthermore, during the cooling process, the cooling rate is 38°C / s-42°C / s.

[0015] An embodiment of the present invention also provides a method for predicting the martensitic transformation starting temperature of a sample based on grain size, comprising: obtaining the martensitic transformation starting temperature corresponding to different grain sizes based on the above-mentioned measurement method; constructing a polynomial nonlinear regression model optimized by a genetic algorithm using the grain size and the corresponding martensitic transformation starting temperature as parameters; calculating the martensitic transformation starting temperature of the sample based on the polynomial nonlinear regression model; the sample and the sample for establishing the polynomial nonlinear regression model belong to the same material.

[0016] Furthermore, the average particle size and the deviation range of the sample are calculated, and the corresponding martensitic transformation starting temperature is obtained by calculating the average particle size; and the martensitic transformation starting temperature is corrected based on the deviation range.

[0017] Furthermore, the final martensitic transformation starting temperature of the sample is calculated as follows:

[0018] ;

[0019] ;

[0020] Where, is the corrected martensitic transformation starting temperature, is the martensitic transformation start temperature calculated by the polynomial nonlinear regression model, is the deviation amplitude, is the particle size corresponding to 30% in the distribution curve, It is the particle size corresponding to 70% in the distribution curve.

[0021] The beneficial effects brought about by the technical solution provided by the embodiments of the present invention include: the technical solution proposed in the present invention uses HT-CLSM (High-temperature confocal laser scanning microscope) to observe the phase transformation process of different grain sizes during the cooling process, and obtains the critical temperature of different grains for the first transformation to martensite through frame-by-frame analysis, thereby realizing high-throughput measurement; however, direct observation still has corresponding technical problems. First, due to the close contact between the grains, there is mutual influence between different grains, and the martensitic phase transformation is accompanied by thermal effect and volume effect, that is, heat is released or absorbed during the phase transformation, and the volume changes. The existence of stress and thermal effects makes the high-throughput detection results of the same sample inaccurate. In order to solve the above problems, the present application limits the depth and width of grain boundary erosion, and the depth and width of the above grain boundaries facilitate the subsequent measurement of the grain size using an algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0023] Figure 1 The tissue morphology of the sample provided in Example 1 of the present invention at different stages, a is the surface morphology of the sample after treatment, b is when the temperature is reduced to 349.1°C, c is when the temperature is reduced to 321.6°C, and d is when the temperature is reduced to 261.1°C;

[0024] Figure 2 This is the polynomial nonlinear regression model provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0025] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0026] An embodiment of the present invention provides a high-throughput method for determining the martensitic transformation starting temperature, comprising:

[0027] S1. Process the sample;

[0028] S2. performing austenitizing treatment on the treated sample;

[0029] S3, using a high temperature confocal laser scanning microscope to record the phase transformation process of the sample from austenite to martensite during the cooling process;

[0030] S4, analyzing the critical temperature of each grain for the first transformation into martensite frame by frame to obtain the martensite transformation starting temperature of different grain sizes;

[0031] After the sample is processed, the depth of the grain boundary erosion is not less than 3 μm, and the width of the grain boundary erosion is not greater than 1 μm.

[0032] The technical solution proposed in the present invention uses HT-CLSM (High-temperature confocal laser scanning microscope) to observe the phase transformation process of different grain sizes during the cooling process, and obtains the critical temperature of different grains for the first transformation to martensite through frame-by-frame analysis, thereby realizing high-throughput measurement; however, direct observation still has corresponding technical problems. First, due to the close contact between the grains, there is mutual influence between different grains, and the martensitic phase transformation is accompanied by thermal effect and volume effect, that is, heat is released or absorbed during the phase transformation, and the volume changes. The existence of stress and thermal effect makes the high-throughput detection results of the same sample inaccurate. In order to solve the above problems, this application limits the depth and width of grain boundary erosion, and the depth and width of the above grain boundaries facilitate the subsequent measurement of the grain size using algorithms.

[0033] The sample processing includes grinding, primary polishing, etching, and secondary polishing. Specifically, the sample surface is polished sequentially using 400#-2000# sandpaper to remove the oxide layer and uneven areas, such as using 400#, 800#, 1000#, 1200#, and 2000# sandpaper to polish the sample surface sequentially. Subsequently, a primary polishing is performed, in which the sample surface is mechanically polished using a polishing paste to achieve a mirror-like effect. The sample surface is then cleaned, preferably using anhydrous alcohol ultrasonic cleaning for 10 minutes to remove surface particles and residual impurities.

[0034] The etching process utilizes two etchants: a first etchant consisting of a mixture of picric acid and alcohol at a ratio of 4g-5g per 100ml; and a second etchant consisting of a mixture of nitric acid and ethanol at a volume ratio of 8%-10%. The first etchant helps erode grain boundaries while eroding grains relatively lightly, making it suitable for initial erosion of grain boundaries, thereby preventing the production of samples with wide grain boundaries. The second etchant, combined with subsequent etching processes, increases grain boundary erosion, ensuring a depth of no less than 3µm and a width of no more than 1µm.

[0035] The etching process is as follows: placing the sample in a first etching agent, keeping it in an ultrasonic oscillation environment for 10s-15s, and immediately placing it in anhydrous ethanol for cleaning and drying after completion; then placing the sample after the initial etching in a second etching agent, ultrasonically oscillating for 1s-2s, taking it out, removing excess etching agent from the surface of the sample, setting the observation side upward and keeping it for at least 20s, repeating the etching in the second etching agent 3-4 times, and cleaning and drying after completion. During the etching process, a first etchant is used to erode the grain boundaries of the sample, followed by a second etchant. When the second etchant is used for treatment, ultrasonic vibration is used to allow the second etchant to fully penetrate the grain boundaries of the etched sample. Excess etchant on the surface is then removed. Preferably, an air bag is used to blow away excess etchant on the sample surface to avoid further erosion of the grain surface. The following process is repeated 3-4 times: placing the sample in the second etchant, ultrasonically vibrating for 1s-2s, removing the sample and blowing away excess etchant on the surface, and holding the sample with the observation surface facing up for at least 20s. The grain boundaries are etched multiple times, ultimately achieving a grain boundary erosion depth of no less than 3µm and a grain boundary erosion width of no more than 1µm. The above technical effects are primarily achieved through the combined effect of the selection of etchants and the corresponding etching process.

[0036] The secondary polishing is performed mechanically at a speed of 300-600 rpm, a pressure of 15N-20N, a diamond suspension with a particle size of 1μm-3μm, and a short-pile polishing cloth. Polishing lasts for at least 30 minutes. The secondary polishing is primarily used to remove products produced by the reaction between the etchant and the sample during the erosion process. These process parameters are designed to avoid excessive polishing intensity, which can affect the depth and width of grain boundary erosion. After the secondary polishing, the grain surface achieves a mirror finish, and the grain boundaries are clearly visible, ensuring clear images during HT-CLSM observation.

[0037] The austenitization process involves heating from room temperature to 180-220°C at a rate of 0.5-1.0°C / s; then heating to 980-1050°C at a rate of 8-12°C / s; and finally heating to 1200-1280°C at a rate of 1-3°C / s and holding for 150-200 seconds to form a mixed-grain structure. This treatment helps induce the formation of austenite grains of varying sizes within the sample, thereby creating a mixed-grain structure. Holding at 1200-1280°C for 150-200 seconds ensures sufficient growth of large grains while retaining some small grains, ultimately forming a representative mixed-grain austenite initial structure, providing a foundation for high-throughput measurement.

[0038] During the cooling process, the cooling rate is 38°C / s-42°C / s. Using an air cooling process with a cooling rate of 38°C / s-42°C / s allows austenite to rapidly transform into martensite. This rate is higher than the critical cooling rate of most martensitic steels, ensuring that the Ms point is fully exposed during the transformation process, facilitating accurate recording.

[0039] During the cooling process, the HT-CLSM system records the evolution of the sample's surface microstructure in real-time video mode. Using laser confocal scanning, the structural changes of each grain are captured frame by frame, recording the critical temperature at which the martensite phase first appears in each grain, which is the Ms value of that grain.

[0040] The embodiment of the present invention further discloses a method for predicting the martensitic transformation starting temperature of a sample based on grain size, comprising:

[0041] S5. Obtaining the martensitic transformation starting temperature corresponding to different grain sizes based on the above measurement method;

[0042] S6. Constructing a polynomial nonlinear regression model optimized by a genetic algorithm using the grain size and the corresponding martensitic transformation starting temperature as parameters;

[0043] S7, calculating the martensitic transformation starting temperature of the sample based on the polynomial nonlinear regression model;

[0044] The sample and the sample used to establish the polynomial nonlinear regression model belong to the same material.

[0045] It should be noted that the "sample" and "sample" in this invention refer to the same material produced using different processes. This can be understood as providing a polynomial nonlinear regression model for the material through the sample. This polynomial nonlinear regression model can then be used to guide the production process of products made from the same material but using different processes. Specifically, if different austenitization processes result in different grain size distributions, the polynomial nonlinear regression model proposed in this invention can be used to design and guide processes for materials with different grain sizes.

[0046] Initial austenite structure identification and PAGS (original austenite grain size) statistics: Image frames before cooling are selected as a reference for the austenite structure. Grain boundaries are identified using deep learning image recognition methods. A geometric measurement algorithm based on region segmentation is used to calculate the equivalent circular diameter of each grain as the PAGS value. The number and size of each grain are counted, and a number-to-PAGS mapping table is created.

[0047] Martensitic Transformation Tracking and Ms Extraction: We analyze cooling image sequences frame by frame and use deep learning-based image classification and semantic segmentation models to identify the characteristic regions where martensite first appears in each grain. Combining the cooling rate with the image timestamp, we accurately calculate the transformation onset temperature for each grain and obtain the Ms data point corresponding to each PAGS.

[0048] Training dataset construction: All grain numbers, corresponding PAGS, and Ms values ​​were assembled into a dataset, divided into a training set (80%) and a test set (20%). PAGS variables were normalized to facilitate subsequent machine learning modeling.

[0049] Construction of polynomial nonlinear regression model based on genetic algorithm:

[0050] (1) Model expression:

[0051] Select the second to fifth order polynomial form to construct the Ms prediction function:

[0052] ;

[0053] The embodiment of the present invention adopts the fourth order.

[0054] (2) Genetic algorithm parameter setting:

[0055] The population size is set to 70, the crossover probability is set to 0.8, the mutation probability is set to 0.02, the maximum number of iterations is set to 300, the encoding method is set to real number encoding, and the fitness function is MSE (mean square error)

[0056] (3) Evolutionary process control:

[0057] The initial population generates model expressions using random parameters, and the prediction performance of each generation of models is evaluated through the fitness function. Selection, crossover, and mutation operations are performed iteratively to finally obtain the global optimal Ms prediction model.

[0058] (4) Model validation and generalization:

[0059] Use the test set to evaluate prediction performance and calculate metrics such as R² and RMSE to verify the model's generalization ability. If the error meets the set standard (e.g., RMSE < 5°C), the model can be used to predict Ms under unknown PAGS conditions.

[0060] Specifically, the average particle size and the deviation range of the sample are calculated, and the corresponding martensitic transformation starting temperature is obtained by calculating the average particle size; and the martensitic transformation starting temperature is corrected based on the deviation range.

[0061] Specifically, the final martensitic transformation starting temperature of the sample is calculated as follows:

[0062] ;

[0063] ;

[0064] Where, is the corrected martensitic transformation starting temperature, is the martensitic transformation starting temperature calculated by the polynomial nonlinear regression model, is the deviation amplitude, is the particle size corresponding to 30% in the distribution curve, It is the particle size corresponding to 70% in the distribution curve.

[0065] Specifically, the particle size distribution curve is obtained, that is, the proportion of different particle sizes in the total number of grains is calculated according to the particle size, with the horizontal axis being the proportion and the vertical axis being the particle size, and the particle sizes corresponding to 30% and 70% are calculated respectively.

[0066] For the convenience of description, the same martensitic steel is used in the embodiments of the present invention.

[0067] In order to better illustrate the embodiments of the present invention, the present invention is further described in detail below through specific examples.

[0068] Example 1

[0069] The present invention provides a high-throughput method for determining and predicting the martensitic transformation start temperature, comprising the following steps:

[0070] S1. Process the sample.

[0071] The sample was ground, polished once, etched, and polished twice; the etchants used in the etching process were: the first etchant was a mixture of picric acid and alcohol, wherein the ratio of picric acid to alcohol was 4g / 100ml; the second etchant was a mixture of nitric acid and ethanol, wherein the volume proportion of nitric acid was 8%.

[0072] The etching process is as follows: placing the sample in a first etching agent, keeping it in an ultrasonic oscillation environment for 10 seconds, and immediately placing it in anhydrous ethanol for cleaning and drying; then placing the sample after the initial etching in a second etching agent, ultrasonically oscillating it for 1 second, taking it out, blowing off the excess etching agent on the surface of the sample, setting the observation side upward and keeping it for 20 seconds, repeating the etching in the second etching agent for 3 times, and cleaning and drying it after the end.

[0073] The morphology of the sample after processing is as follows Figure 1 As shown in a in FIG, the depth of the grain boundary erosion is 3.0 μm, and the width of the grain boundary erosion is 0.6 μm.

[0074] S2. Performing austenitizing treatment on the treated sample.

[0075] The temperature was increased from room temperature to 200°C at 0.8°C / s, then to 1000°C at 10°C / s, and finally to 1250°C at 2°C / s.

[0076] S3. During the cooling process, a high temperature confocal laser scanning microscope is used to record the phase transformation process of the sample from austenite to martensite.

[0077] Initial austenite structure identification and PAGS (original austenite grain size) statistics: Image frames before cooling are selected as a reference for the austenite structure. Grain boundaries are identified using deep learning image recognition methods. A geometric measurement algorithm based on region segmentation is used to calculate the equivalent circular diameter of each grain as the PAGS value. The number and size of each grain are counted, and a number-to-PAGS mapping table is created.

[0078] S4, analyzing the critical temperature of each grain for the first transformation into martensite frame by frame to obtain the martensite transformation starting temperature of different grain sizes;

[0079] Martensitic Transformation Tracking and Ms Extraction: We analyze cooling image sequences frame by frame and use deep learning-based image classification and semantic segmentation models to identify the characteristic regions where martensite first appears in each grain. Combining the cooling rate with the image timestamp, we accurately calculate the transformation onset temperature for each grain and obtain the Ms data point corresponding to each PAGS.

[0080] like Figure 1 As shown in Figure b, when the temperature drops to 349.1°C, not all austenite phases undergo martensite transformation. Only the austenite phase with larger PAGS undergoes martensite transformation. Figure 1 As shown in Figure c, when the temperature is further reduced to 321.6℃, martensite begins to form in the smaller austenite phase of PAGS; Figure 1 As shown in (d), when the temperature drops to 261.1°C, all the austenite in the observation area completes the martensite transformation.

[0081] S5. Obtain the martensitic transformation starting temperature corresponding to different grain sizes based on the above measurement method.

[0082] S6. Constructing a polynomial nonlinear regression model optimized by a genetic algorithm using the grain size and the corresponding martensitic transformation starting temperature as parameters.

[0083] like Figure 2As shown in Figure 3, when the thickness of PAGS increases from 30 μm to 250 μm, the Ms point increases from 300 °C to 320 °C. It is worth noting that when the thickness of PAGS exceeds 150 μm, the slope of the fitting curve is close to 0, which indicates that the effect of the increase of PAGS on the Ms point is almost negligible.

[0084] S7. Calculating the martensitic transformation starting temperature of the sample based on the polynomial nonlinear regression model.

[0085] The corresponding martensitic transformation starting temperature is calculated by the average particle size; and the martensitic transformation starting temperature is corrected based on the deviation amplitude.

[0086] The final martensitic transformation start temperature of the sample is calculated as:

[0087] ;

[0088] ;

[0089] The average particle size of the sample in the embodiment of the present invention is 72 μm, and the martensitic transformation starting temperature calculated by the polynomial nonlinear regression model is 331°C. The particle size is 59 μm. The particle size is 85 μm.

[0090] According to calculation, the final martensitic transformation starting temperature of the sample is 308°C.

[0091] For the same sample and process, the martensitic transformation starting temperature tested by the thermal dilatometer method is 315°C. Compared with the final martensitic transformation starting temperature obtained by the technical solution of the present invention, the error is 2.2%, which can be used to guide actual production.

[0092] Example 2

[0093] The present invention provides a high-throughput method for determining and predicting the martensitic transformation start temperature, comprising the following steps:

[0094] S1. Process the sample.

[0095] The sample was ground, polished once, etched, and polished twice; the etchants used in the etching process were: the first etchant was a mixture of picric acid and alcohol, wherein the ratio of picric acid to alcohol was 4g / 100ml; the second etchant was a mixture of nitric acid and ethanol, wherein the volume proportion of nitric acid was 9%.

[0096] The etching process is as follows: placing the sample in a first etching agent, keeping it in an ultrasonic oscillation environment for 12 seconds, and immediately placing it in anhydrous ethanol for cleaning and drying; then placing the sample after the initial etching in a second etching agent, ultrasonically oscillating it for 2 seconds, taking it out, blowing off the excess etching agent on the surface of the sample, setting the observation side upward and keeping it for 20 seconds, repeating the etching in the second etching agent for 3 times, and cleaning and drying it after completion.

[0097] After the sample was processed, the depth of the grain boundary erosion was 3.5 μm, and the width of the grain boundary erosion was 0.8 μm.

[0098] S2. Performing austenitizing treatment on the treated sample.

[0099] The temperature was raised from room temperature to 180°C at 0.5°C / s, then to 980°C at 8°C / s, and finally to 1200°C at 1°C / s.

[0100] S3. During the cooling process, a high temperature confocal laser scanning microscope is used to record the phase transformation process of the sample from austenite to martensite.

[0101] S4. Analyze the critical temperature of each grain when it first transforms into martensite frame by frame to obtain the martensite transformation starting temperature of different grain sizes.

[0102] S5. Obtain the martensitic transformation starting temperature corresponding to different grain sizes based on the above measurement method.

[0103] S6. Constructing a polynomial nonlinear regression model optimized by a genetic algorithm using the grain size and the corresponding martensitic transformation starting temperature as parameters.

[0104] S7. Calculating the martensitic transformation starting temperature of the sample based on the polynomial nonlinear regression model.

[0105] The corresponding martensitic transformation starting temperature is calculated by the average particle size; and the martensitic transformation starting temperature is corrected based on the deviation amplitude.

[0106] The final martensitic transformation start temperature of the sample is calculated as:

[0107] ;

[0108] ;

[0109] The average particle size of the sample in the embodiment of the present invention is 65 μm, and the martensitic transformation starting temperature calculated by the polynomial nonlinear regression model is 330°C. The particle size is 54 μm. The particle size is 76 μm.

[0110] According to calculation, the final martensitic transformation starting temperature of the sample is 308°C.

[0111] For the same sample and process, the martensitic transformation starting temperature tested by the thermal dilatometer method is 314°C. Compared with the final martensitic transformation starting temperature obtained by the technical solution of the present invention, the error is 2.0%, which can be used to guide actual production.

[0112] Example 3

[0113] The present invention provides a high-throughput method for determining and predicting the martensitic transformation start temperature, comprising the following steps:

[0114] S1. Process the sample.

[0115] The sample was ground, polished once, etched, and polished twice; the etchants used in the etching process were: the first etchant was a mixture of picric acid and alcohol, wherein the ratio of picric acid to alcohol was 5g / 100ml; the second etchant was a mixture of nitric acid and ethanol, wherein the volume proportion of nitric acid was 10%.

[0116] The etching process is as follows: placing the sample in a first etching agent, keeping it in an ultrasonic oscillation environment for 15 seconds, and immediately placing it in anhydrous ethanol for cleaning and drying; then placing the sample after the initial etching in a second etching agent, ultrasonically oscillating it for 2 seconds, taking it out, blowing off the excess etching agent on the surface of the sample, setting the observation surface upward and keeping it for 20 seconds, repeating the etching in the second etching agent for 4 times, and then cleaning and drying it after the end.

[0117] After the sample was processed, the depth of the grain boundary erosion was 3.8 μm, and the width of the grain boundary erosion was 1.0 μm.

[0118] S2. Performing austenitizing treatment on the treated sample.

[0119] The temperature was increased from room temperature to 220°C at 1.0°C / s, then to 1050°C at 12°C / s, and finally to 1280°C at 1°C / s.

[0120] S3. During the cooling process, a high temperature confocal laser scanning microscope is used to record the phase transformation process of the sample from austenite to martensite.

[0121] S4. Analyze the critical temperature of each grain when it first transforms into martensite frame by frame to obtain the martensite transformation starting temperature of different grain sizes.

[0122] S5. Obtain the martensitic transformation starting temperature corresponding to different grain sizes based on the above measurement method.

[0123] S6. Constructing a polynomial nonlinear regression model optimized by a genetic algorithm using the grain size and the corresponding martensitic transformation starting temperature as parameters.

[0124] S7. Calculating the martensitic transformation starting temperature of the sample based on the polynomial nonlinear regression model.

[0125] The corresponding martensitic transformation starting temperature is calculated by the average particle size; and the martensitic transformation starting temperature is corrected based on the deviation amplitude.

[0126] The final martensitic transformation start temperature of the sample is calculated as:

[0127] ;

[0128] The average particle size of the sample in the embodiment of the present invention is 78 μm, and the martensitic transformation starting temperature calculated by the polynomial nonlinear regression model is 338 ° C. The particle size is 64 μm. The particle size is 85 μm.

[0129] According to calculation, the final martensitic transformation starting temperature of the sample is 319°C.

[0130] For the same sample and process, the martensitic transformation starting temperature tested by the thermal dilatometer method is 330°C. Compared with the final martensitic transformation starting temperature obtained by the technical solution of the present invention, the error is 3.3%, which can be used to guide actual production.

[0131] Experimental example

[0132] This experimental example uses the same samples as Examples 1-3, and the austenitizing process is as follows:

[0133] Experimental Example 1: The austenitizing process is: heating to 1200°C at 8°C / s and holding for 180s. The obtained sample particle size is 72±6μm.

[0134] Experimental Example 2: The austenitizing process is: heating to 1200°C at a rate of 10°C / s and holding for 180s. The obtained sample particle sizes are 65±5μm.

[0135] Experimental Example 3: The austenitizing process is: heating to 1280°C at 8°C / s and holding for 180s. The obtained sample particle sizes are 78±8μm.

[0136] It can be seen that the particle size distribution of Experimental Examples 1-3 is relatively uniform.

[0137] The martensitic transformation starting temperatures of the samples tested by the thermal dilatometer method were 343℃, 345℃ and 351℃, respectively. The martensitic transformation starting temperatures calculated by the polynomial nonlinear regression model were 331℃, 330℃ and 338℃, respectively. The error ratios between the two were 3.50%, 4.35% and 3.70%, respectively. The errors were all within 5%, proving that the model's predictions are more accurate.

[0138] Comparative Example 1

[0139] This comparative example differs from Example 1 in that, in step S1, the etching process involves placing the sample in a first etchant, maintaining it in an ultrasonic oscillation environment for 70 seconds, and then immediately rinsing and drying it in anhydrous ethanol. After the sample treatment, the depth of grain boundary etching was 1.5 µm, and the width of grain boundary etching was 3 µm.

[0140] During the cooling process, a high temperature confocal laser scanning microscope was used to record the phase transformation process of the sample from austenite to martensite.

[0141] The critical temperature of each grain transforming into martensite for the first time is analyzed frame by frame to obtain the martensite transformation starting temperature of different grain sizes.

[0142] A polynomial nonlinear regression model optimized by a genetic algorithm is constructed with the grain size and the corresponding martensitic transformation starting temperature as parameters.

[0143] The martensitic transformation start temperature calculated by the polynomial nonlinear regression model is 310℃.

[0144] The final martensitic transformation start temperature of the sample is calculated as:

[0145] ;

[0146] ;

[0147] The average particle size of the sample in the embodiment of the present invention is 76 μm. The particle size is 58 μm. The particle size is 84 μm.

[0148] According to calculation, the final martensitic transformation starting temperature of the sample is 288°C.

[0149] For the same sample and process, the martensitic transformation starting temperature tested by the thermal dilatometer method is 315°C. Compared with the final martensitic transformation starting temperature obtained by the technical solution of the present invention, the error is 8.6%, which is a large error.

[0150] Comparing Examples 1-3 with Experimental Example 1, the martensitic transformation start temperatures of samples with approximately a single particle size, measured using the gold standard dilatometer method, were similar to those calculated using the polynomial nonlinear regression model, with an error within 5%. This demonstrates the high accuracy of the polynomial nonlinear regression model proposed in the present invention based on high-throughput testing. As shown in Example 1 and Comparative Example 1, due to the inadequate grain boundary erosion depth and width in Comparative Example 1 and the mutual influence between adjacent grains, the constructed polynomial nonlinear regression model exhibited problems, resulting in a low calculated martensitic transformation start temperature, with a deviation of up to 8.6% from the martensitic transformation start temperature measured using the dilatometer method. This significant error makes it of limited practical value in guiding production.

[0151] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A high-throughput method for determining the martensitic transformation starting temperature, characterized in that: include: Process the samples; The treated samples were austenitized; During the cooling process, a high temperature confocal laser scanning microscope was used to record the phase transformation process of the sample from austenite to martensite; Analyze the critical temperature of each grain's first transformation into martensite frame by frame to obtain the martensite transformation starting temperature of different grain sizes; After the sample is treated, the depth of the grain boundary erosion is not less than 3 μm, and the width of the grain boundary erosion is not more than 1 μm; Processing the sample includes grinding, primary polishing, etching, and secondary polishing; The etching adopts two etching agents, the first etching agent is a mixture of picric acid and alcohol, wherein the ratio of picric acid to alcohol is 4g-5g / 100ml; The second etchant is a mixture of nitric acid and ethanol, wherein the volume proportion of nitric acid is 8%-10%; The etching process is as follows: placing the sample in a first etching agent, maintaining it in an ultrasonic vibration environment for 10s-15s, and immediately placing it in anhydrous ethanol for cleaning and drying; Then, place the sample after the initial etching in the second etching agent, ultrasonically vibrate for 1s-2s, then take it out to remove the excess etching agent on the surface of the sample. Set the sample with the observation surface facing up and keep it for at least 20s. Repeat the etching in the second etching agent 3-4 times. After the end, clean and dry it.

2. The measuring method according to claim 1, wherein The secondary polishing adopts mechanical polishing with a rotation speed of 300rpm-600rpm, a pressure of 15N-20N, a diamond suspension with a particle size of 1μm-3μm, a short-pile polishing cloth, and polishing for at least 30min.

3. The measuring method according to claim 1, wherein The austenitizing process is as follows: heating from room temperature to 180-220°C at 0.5-1.0°C / s; heating to 980-1050°C at 8-12°C / s; heating to 1200-1280°C at 1-3°C / s and keeping the temperature for 150-200 seconds to form a mixed grain structure.

4. The measuring method according to claim 1, wherein During the cooling process, the cooling rate is 38°C / s-42°C / s.

5. A method for predicting the martensitic transformation start temperature of a sample based on grain size, characterized in that: Obtaining the martensitic transformation starting temperature corresponding to different grain sizes based on the measurement method described in any one of claims 1 to 4; A polynomial nonlinear regression model optimized by a genetic algorithm is constructed using the grain size and the corresponding martensitic transformation starting temperature as parameters; Calculating the martensitic transformation starting temperature of the sample based on the polynomial nonlinear regression model; The sample and the sample used to establish the polynomial nonlinear regression model belong to the same material.

6. The method according to claim 5, characterized in that Calculating the average particle size and the deviation of the sample, and obtaining the corresponding martensitic transformation starting temperature based on the average particle size; The martensitic transformation start temperature is corrected based on the magnitude of the deviation.

7. The method according to claim 6, characterized in that The final martensitic transformation start temperature of the sample is calculated as: Ms=α 1 / 5 Ms‘; α=R 30 / R 70 ; Where Ms is the corrected martensitic transformation starting temperature, Ms' is the martensitic transformation starting temperature calculated by the polynomial nonlinear regression model, α is the deviation amplitude, and R 30 is the particle size corresponding to 30% of the distribution curve, R 70 It is the particle size corresponding to 70% in the distribution curve.

Images