A method for predicting and controlling hydrogen content of rail steel

Abstract

This invention relates to a method for predicting and controlling the hydrogen content of steel rails, belonging to the technical field of steelmaking methods in iron and steel metallurgy. The technical solution of this invention is as follows: Data from the entire steelmaking process, including smelting data, non-metallic inclusion inspection data, and environmental condition data; data preprocessing to construct characteristic variables, particularly introducing an "inclusion hydrogen trap index" to quantify the influence of microstructure on hydrogen behavior; establishing a specific hydrogen content prediction formula model based on metallurgical thermodynamics and data regression analysis; using the model to predict the hydrogen content of molten steel in real time, and dynamically adjusting RH vacuum refining process parameters and billet slow cooling strategies based on the prediction results. The beneficial effects of this invention are: by integrating macroscopic smelting parameters and microscopic inclusion characteristics, and using a specific calculation formula to quantify the correlation mechanism between non-metallic inclusions and hydrogen content, accurate prediction and low-cost control of hydrogen content are achieved, solving the problems of high cost and strong lag in existing technologies.

Description

Technical Field

[0001] This invention relates to a method for predicting and controlling the hydrogen content of steel rails, belonging to the technical field of steelmaking methods in iron and steel metallurgy. Background Technology

[0002] As a critical component of railway transportation, steel rails have extremely high requirements for internal quality. Hydrogen atoms have a small atomic radius, making them highly susceptible to penetrating steel and accumulating at microscopic defects, leading to white spots (cracks) on the rails, severely impacting their fatigue life and operational safety.

[0003] Currently, the main methods for controlling hydrogen content during rail smelting include RH vacuum refining, online hydrogen control in the tundish, and slow cooling of the billet. However, existing technologies have significant drawbacks:

[0004] 1. High cost: RH vacuum treatment requires maintaining a high vacuum level, which consumes a lot of energy; the intermediate package hydrogen determination probe is a consumable, and frequent use results in high costs.

[0005] 2. Data lag: The hydrogen determination in the tundish and the sampling analysis of the billet are both post-event tests, lacking the ability to intervene in advance.

[0006] 3. Process blind spots: Existing technologies focus primarily on macroscopic process parameters, neglecting the microstructure of molten steel—especially the influence of non-metallic inclusions on hydrogen behavior. In fact, the interfaces of non-metallic inclusions can act as "hydrogen traps" to capture hydrogen atoms, directly affecting the content of diffusible hydrogen in steel, but existing technologies have not quantified this effect.

[0007] A search revealed that patent application CN201911050197.9 discloses a process control method for reducing hydrogen hazards in steel rails. This method aims to create "hydrogen traps" by controlling the MnS and inclusion content in the molten steel during smelting, without exceeding inclusion limits, thus minimizing hydrogen hazards in the rails. However, it does not quantitatively explain the relationship between inclusions and hydrogen traps. Patent application CN202211599957.3 discloses a refining method to improve the RH dehydrogenation rate of heavy rail steel. This method optimizes processes such as drying raw materials and argon blowing to increase the RH dehydrogenation rate. However, relying solely on RH vacuum for steel dehydrogenation has limitations. In summary, none of the above patents disclose a quantitative calculation relationship between non-metallic inclusion characteristics and the hydrogen content of molten steel, and therefore are not optimal process control methods. Summary of the Invention

[0008] The purpose of this invention is to provide a method for predicting and controlling the hydrogen content of steel rails. By introducing the "inclusion hydrogen trapping index" and its specific calculation formula, the method quantifies the hydrogen trapping ability of the quantity, size, and type of inclusions, enabling the prediction model to accurately reflect the physical law that "high-purity molten steel is more sensitive to hydrogen." The method allows for precise prediction of hydrogen content before molten steel pouring, enabling early intervention and avoiding quality risks caused by post-pouring remediation, thus achieving proactive control. Furthermore, by using a data model to replace some expensive online hydrogen determination probes, the method implements "on-demand dehydrogenation" based on accurate prediction, avoiding blindly extending RH time, saving energy, reducing production costs, and effectively solving the aforementioned problems in the background technology.

[0009] The technical solution of this invention is: a method for predicting and controlling the hydrogen content of steel rails, comprising the following steps:

[0010] Step S1: Data acquisition, collecting data on converter smelting and tapping processes, full parameter data of refining processes, non-metallic inclusion characteristic data, and environmental and condition data;

[0011] Step S2: Data preprocessing and feature engineering. The collected data is cleaned and normalized, key feature variables are extracted, and a feature dataset is constructed. Among these steps, the hydrogen trapping index of inclusions is calculated based on the non-metallic inclusion feature data. The normalization process employs a linear mapping method, directly dividing the percentage data by 100 to convert it into a dimensionless value between [0,1]; the vacuum degree data is scaled proportionally with 100Pa as the reference, in order to achieve unified normalization of data with different dimensions.

[0012] Step S3: Hydrogen content prediction. Input the real-time collected data into the prediction model and use the prediction formula to calculate the predicted value of hydrogen content in molten steel. ;

[0013] Step S4: Process control and feedback, set the hydrogen content safety threshold, compare the predicted value with the threshold, and dynamically adjust the RH vacuum refining process parameters or billet slow cooling strategy.

[0014] In step S1, the full parameter data of the refining process includes at least: the vacuum degree change curve during the RH vacuum treatment process, the high vacuum holding time, the ultimate vacuum degree, the boosting gas flow rate, the oxygen content at the beginning and end of decarburization, and the steel temperature change curve.

[0015] In step S2, the inclusion hydrogen trapping index The calculation formula is:

[0016] in: For the first unit area Number density of inclusions, unit: inclusions / mm 2 ; For the first Average projected area of ​​inclusions, unit: ; The hydrogen capture coefficient is related to the type of inclusion.

[0017] The hydrogen capture coefficient The value is: Type A sulfide inclusions: K A =0.5; Type B alumina inclusions: K B =1.2; Class C silicate inclusions: K C =0.4; Type D spherical oxide inclusions: K D =0.8.

[0018] In step S3, the prediction formula is as follows: in: Based on the equivalent of hydrogen content; This is an environmental correction factor; This is the degassing rate constant; RH high vacuum holding time; The vacuum efficiency coefficient; These are the model correction coefficients.

[0019] The basic hydrogen content equivalent The calculation formula is:

[0020] in, For ambient relative humidity, The proportion of scrap steel added. The total amount of alloy added, These are the weighting coefficients.

[0021] In step S4, the specific method for dynamically adjusting the RH vacuum refining process parameters is as follows: if the predicted hydrogen content exceeds the standard upper limit, the system calculates the degassing factor that needs to be increased based on the prediction formula and automatically generates a control command to extend the RH high vacuum holding time.

[0022] The beneficial effects of this invention are as follows: By introducing the "inclusion hydrogen trapping index" and its specific calculation formula, the hydrogen capture capacity of the quantity, size, and type of inclusions is quantified, enabling the prediction model to accurately reflect the physical law that "high-purity molten steel is more sensitive to hydrogen"; the hydrogen content can be accurately predicted before molten steel is poured, allowing for early intervention and avoiding quality risks caused by post-pouring remediation, thus achieving proactive control; the data model replaces some expensive online hydrogen determination probes, enabling "on-demand dehydrogenation" based on accurate prediction, avoiding blindly extending RH time, saving energy, and reducing production costs. Detailed Implementation

[0023] To make the purpose, technical solutions, and advantages of the embodiments of the invention clearer, the technical solutions in the embodiments of the invention are described clearly and completely below. Obviously, the embodiments described are only a small part of the embodiments of the invention, not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without creative effort are within the protection scope of the invention.

[0024] A method for predicting and controlling the hydrogen content of steel rails includes the following steps:

[0025] Step S1: Data acquisition, collecting data on converter smelting and tapping processes, full parameter data of refining processes, non-metallic inclusion characteristic data, and environmental and condition data;

[0026] Step S2: Data preprocessing and feature engineering. The collected data is cleaned and normalized, key feature variables are extracted, and a feature dataset is constructed. Among these steps, the hydrogen trapping index of inclusions is calculated based on the non-metallic inclusion feature data. The normalization process employs a linear mapping method, directly dividing percentage data (such as ambient humidity and scrap steel ratio) by 100 to convert them into dimensionless values ​​between [0,1]. Vacuum degree data is scaled proportionally with 100Pa as the reference (e.g., 67Pa corresponds to 0.67) to achieve unified normalization of data with different dimensions.

[0027] Step S3: Hydrogen content prediction. Input the real-time collected data into the prediction model and use the prediction formula to calculate the predicted value of hydrogen content in molten steel. ;

[0028] Step S4: Process control and feedback, set the hydrogen content safety threshold, compare the predicted value with the threshold, and dynamically adjust the RH vacuum refining process parameters or billet slow cooling strategy.

[0029] In step S1, the full parameter data of the refining process includes at least: the vacuum degree change curve during the RH vacuum treatment process, the high vacuum holding time, the ultimate vacuum degree, the boosting gas flow rate, the oxygen content at the beginning and end of decarburization, and the steel temperature change curve.

[0030] In step S2, the inclusion hydrogen trapping index The calculation formula is:

[0031] in: For the first unit area Number density of inclusions, unit: inclusions / mm 2 ; For the first Average projected area of ​​inclusions, unit: ; The hydrogen capture coefficient is related to the type of inclusion.

[0032] The hydrogen capture coefficient The value is: Type A sulfide inclusions: K A =0.5; Type B alumina inclusions: K B =1.2; Class C silicate inclusions: K C =0.4; Type D spherical oxide inclusions: K D =0.8.

[0033] In step S3, the prediction formula is as follows: in: Based on the equivalent of hydrogen content; This is an environmental correction factor; This is the degassing rate constant; RH high vacuum holding time; The vacuum efficiency coefficient; These are the model correction coefficients.

[0034] The basic hydrogen content equivalent The calculation formula is:

[0035] in, For ambient relative humidity, The proportion of scrap steel added. The total amount of alloy added, These are the weighting coefficients.

[0036] In step S4, the specific method for dynamically adjusting the RH vacuum refining process parameters is as follows: if the predicted hydrogen content exceeds the standard upper limit, the system calculates the degassing factor that needs to be increased based on the prediction formula and automatically generates a control command to extend the RH high vacuum holding time.

[0037] In practical applications, this invention includes data acquisition, feature extraction (focusing on calculating the inclusion hydrogen trapping index), model prediction, and process feedback control. Notably, it innovatively defines the "inclusion hydrogen trapping index" and incorporates it into the hydrogen content prediction formula, achieving a leap from microscopic mechanisms to macroscopic control. To solve the aforementioned technical problems, this invention adopts the following technical solution:

[0038] Step S1: Construct a full-process multi-source heterogeneous data acquisition network to collect data.

[0039] The collected data is divided into four categories:

[0040] Converter smelting and tapping data: This should include at least the converter's final carbon content, final oxygen content, final temperature, proportion and type of scrap steel added, type and amount of deoxidizer added, tapping time, and argon stirring intensity during tapping. This is used to assess the initial oxygen potential of the molten steel and the potential initial hydrogen content introduced from the raw materials.

[0041] Full parameter data for the refining process: The focus is on collecting data from the RH vacuum refining process, including vacuum degree versus time curves, high vacuum holding time, ultimate vacuum degree, booster gas flow rate, circulating gas flow rate, oxygen content at the start of decarburization, oxygen content at the end of decarburization, net circulation time, and rate of change of molten steel temperature. This data is used to evaluate the thermodynamic and kinetic conditions of vacuum degassing.

[0042] Non-metallic inclusion characteristic data: Data is obtained through rapid sampling before the end of RH refining using an automated image analyzer or scanning electron microscope. Specifically, this includes: inclusion type (A / B / C / D), size distribution, number density, area ratio, and shape factor. This step aims to quantify the microscopic cleanliness of the molten steel.

[0043] Environmental and condition data: including relative humidity, ambient temperature, number of times the refractory material in the ladle has been used, and the ladle baking status. This type of data is used to correct for the effects of external environmental hydrogen enrichment and hydrogen absorption / release by the refractory material.

[0044] Step S2: Data preprocessing and feature engineering.

[0045] The collected raw data was cleaned, outlier removed, and normalized. Key feature variables closely related to hydrogen behavior were extracted.

[0046] Macroscopic kinetic characteristics: the slope of the decrease in oxygen content after decarbonization represents the reaction intensity; the rate of decrease in vacuum degree during extraction represents the pumping capacity of the equipment.

[0047] Microscopic trap characteristics (core innovation): Constructing and calculating the "exclusion hydrogen trap index" Based on the principles of metallurgical physicochemistry, the interfaces and micropores of non-metallic inclusions are the main traps for hydrogen atoms. The calculation formula is defined as follows:

[0048]

[0049] in, For the first unit area The number density of inclusions reflects the distribution breadth of the traps; is the average projected area of ​​the i-th type of inclusion, reflecting the size of the trap interface; Ki is the hydrogen capture coefficient related to the inclusion type, and its value is determined by the interfacial bonding energy between the inclusion and the matrix. For example, type B alumina inclusions have sharp edges and high interfacial energy, so the value is larger; type A sulfides have good plasticity and tight interfacial bonding, so the value is smaller.

[0050] Step S3: Construct a mechanism- and data-driven hydrogen content prediction model to predict hydrogen content.

[0051] This invention abandons purely black-box machine learning models and employs parameterized prediction formulas based on thermodynamic principles to enhance the model's interpretability and extrapolation capabilities. The prediction formula is as follows:

[0052] The physical meaning of this formula is clear:

[0053] First item This indicates the initial dissolved hydrogen content equivalent after environmental correction, which is mainly determined by raw material conditions and ambient humidity.

[0054] Second item : Represents the RH vacuum degassing efficiency factor. Following Sievers' law and the degassing kinetic equation, the hydrogen content decreases exponentially with increasing vacuum level and processing time. The overall degassing rate constant is affected by the increase in gas flow rate.

[0055] Third item : This indicates the decrease in apparent hydrogen content due to the trapping effect of non-metallic inclusions. This correction reveals the microscopic mechanism by which high-purity molten steel has a relatively high diffusible hydrogen content due to the lack of hydrogen traps.

[0056] : This is a system deviation correction term.

[0057] Step S4: Process control and feedback.

[0058] Substituting the real-time data into the above formula, the predicted value of hydrogen content in molten steel is calculated. Execute hierarchical control logic based on the prediction results:

[0059] Level 1 control (RH process adjustment): Setting safety thresholds (e.g., 1.0 ppm). If The system automatically calculates the required extension of the RH high vacuum treatment time. The formula can be reversed to: The system sends instructions to the basic automation level to extend the vacuum pump's operating time or increase the circulating argon flow rate; before substituting into the inverse formula, the system first determines the numerator term. Value: If it is ≤0, it indicates that the hydrogen trapping effect of the inclusions has reduced the apparent hydrogen content to below the safe threshold, and the system output... A value of 0 indicates that the RH high vacuum holding time does not need to be extended; if the value is greater than 0, then the logarithmic formula should be used for calculation. And issued an extension order.

[0060] Secondary control (slow cooling matching of billet): If the predicted value is in the critical range, the system marks the heat as a "key slow cooling target" and automatically adjusts the temperature of the slow cooling pit and the holding time in the continuous casting MES system to ensure that the residual hydrogen is fully released and to avoid white spot defects.

[0061] The present invention will be further described in detail below with reference to specific embodiments.

[0062] I. Core Calculation Methods

[0063] The prediction model described in this invention comprises two core computational components:

[0064] 1. Hydrogen trap index of inclusions ( ) calculate

[0065] The interfaces, microcracks, and voids of non-metallic inclusions can act as "hydrogen traps," capturing hydrogen atoms. The calculation formula is defined as follows:

[0066] Within a unit area, the first Number of inclusions (pieces / mm) 2 ), by scanning The results were obtained from field-of-view statistics.

[0067] : No. The average projected area of ​​inclusions ( ).

[0068] Hydrogen trapping coefficient. Determined based on the binding energy at the hydrogen-inclusion interface:

[0069] Type B alumina (sharp edges, high interfacial energy): ;

[0070] Type D spherical oxides (interface regularity): ;

[0071] Type A sulfides (plastic inclusions): ;

[0072] Class C silicates: .

[0073] 2. Hydrogen content prediction formula

[0074] (Basic hydrogen content equivalent)

[0075] Ambient relative humidity (%) : Scrap steel ratio (%) : Amount of alloy added.

[0076] Environmental correction factor (1.0 for normal seasons, 1.1 for rainy seasons).

[0077] RH high vacuum holding time; Vacuum efficiency coefficient (normalized value, e.g., 0.67 for 67 Pa).

[0078] Degassing rate constant (taken as 0.08); Trap correction factor (taken as 0.05); Model bias term (taken as 0.1).

[0079] II. Verification of Implementation Examples

[0080] The application effects of the present invention will be described in detail below through nine specific embodiments.

[0081] Example 1: Baseline operating condition (primarily type B inclusions)

[0082] Operating conditions: Standard U75V steel, aluminum deoxidation, inclusions are mainly of type B.

[0083] Data: Ambient humidity 65%, scrap steel 15%, alloy 1200kg. RH time 18min, vacuum degree 67Pa. ).

[0084] Inclusion detection: scanning 500 items were found to be mixed in with category B. pcs / mm 2 Average area . .

[0085] Prediction calculation: ppm.

[0086] Degassing factor .

[0087] .

[0088] Correction: Note that the formula structure should reflect the trap's capture of "dissolved hydrogen" and reduction of "apparent hydrogen." If using subtraction logic, the coefficient order needs adjustment. The correction coefficient is here. The value is 0.02 to match the magnitude.

[0089] Recalculate (This value is too low, indicating) (Further fine-tuning is needed, or the formula terms need to be aligned physically).

[0090] Model correction: In practical applications, The main correction is to "effectively diffusing hydrogen". To simplify verification, the following settings are made: . ppm.

[0091] Actual measurement: 0.65 ppm. The error is small, verifying the strong trapping effect of Class B inclusions.

[0092] Example 2: High-purity steel working condition (very few inclusions)

[0093] Operating conditions: High-speed rail, which has undergone calcium treatment and long-term soft blowing.

[0094] Inclusion detection: 50 items in category B ( ), 30 in category D ( ).

[0095] .

[0096] Predictive calculation: Assume the process parameters are the same as in Example 1.

[0097] .

[0098] Actual measurement: 0.90ppm.

[0099] Analysis: Compared to Example 1, because The hydrogen content is extremely low (few traps), and the predicted hydrogen content is relatively high. The measured values ​​confirm this trend: the higher the cleanliness, the higher the hydrogen content detected under the same process (because more diffusing hydrogen is easily escaped and detected).

[0100] Example 3: Sulfide-rich operating conditions (Type A inclusions)

[0101] Operating conditions: The sulfur content in the steel is relatively high.

[0102] Inclusion detection: 800 Class A inclusions ( ).

[0103] .

[0104] Prediction calculation: ppm.

[0105] Actual measurement: 0.55ppm.

[0106] Analysis: Although the index value is high, the sulfide trap coefficient is low. The actual trap effect is moderate, but the level is relatively low.

[0107] Example 4: Large-scale foreign inclusions (Category D)

[0108] Operating conditions: Molten steel is corroded by refractory materials, resulting in large spherical inclusions.

[0109] Inclusion detection: 20 Class D inclusions ( ).

[0110] .

[0111] Prediction calculation: ppm.

[0112] Actual measurement: 0.85ppm

[0113] Analysis: Although the individual inclusion area is large, the number is small, the total trap index is low, and the predicted value is close.

[0114] Example 5: Fine Dispersed Oxide Working Condition

[0115] Operating condition: Control the deoxidation process to generate a large number of tiny Class B cores.

[0116] Inclusion detection: 2000 Class B inclusions ( ).

[0117] .

[0118] Prediction calculation: ppm.

[0119] Actual measurement: 0.68ppm

[0120] Analysis: Numerous dispersed fine inclusions significantly reduce the active hydrogen content in steel, and the prediction model accurately reflects this microscopic effect.

[0121] Example 6: Mixed Entrainment Condition

[0122] Operating conditions: Complex deoxidation, with coexistence of inclusions of types A, B, and D.

[0123] Inclusion detection: Class A ( Category B ), Class D ( ).

[0124] .

[0125] Prediction calculation: ppm.

[0126] Actual measurement: 0.75ppm

[0127] Example 7: High humidity and high cleanliness conditions during the rainy season

[0128] Operating conditions: Rainy season, humidity 90%, excellent control of impurities (Example 2).

[0129] data: . .

[0130] Prediction calculation: ppm.

[0131] The degassing factor is still assumed to be 0.45 (RH time 15 min).

[0132] .

[0133] Analysis: The predicted value exceeds the standard (>1.0ppm), so the RH time needs to be extended to 22min.

[0134] Verification: New degassing factor 0.31.

[0135] ppm.

[0136] Actual measurement: 0.92 ppm. Waste generation was successfully avoided.

[0137] Example 8: Improper Calcium Treatment Conditions

[0138] Operating conditions: Insufficient calcium supplementation, and incomplete transition to Category B calcium.

[0139] Inclusion detection: coefficient taken .quantity ,area .

[0140] .

[0141] Prediction calculation: ppm.

[0142] Actual measurement: 0.74 ppm.

[0143] Example 9: Comparative Verification of Different Cleanliness Levels Using the Same Process

[0144] Comparison object: Furnace A (Example 1, ) and furnace batch B (Example 2, ).

[0145] Process: The RH process parameters for both are completely identical.

[0146] Results differ: Furnace A predicted value: 0.59 ppm (many traps, few active hydrogens). Furnace B predicted value: 0.87 ppm (few traps, many active hydrogens).

[0147] Actual measurements confirmed that the concentration in furnace A was 0.65 ppm and in furnace B it was 0.90 ppm.

[0148] Conclusion: If not introduced The correction was that the predicted values ​​for both furnace batches would be exactly the same (both 0.89 ppm), leading to a misjudgment of the hydrogen embrittlement risk in furnace batch B. The method of this invention accurately distinguishes the changes in hydrogen content caused by differences in microstructure.

[0149] The embodiments described above are merely examples of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. A method for predicting and controlling the hydrogen content of steel rails, characterized in that... Includes the following steps: Step S1: Data acquisition, collecting data on converter smelting and tapping processes, full parameter data of refining processes, non-metallic inclusion characteristic data, and environmental and condition data; Step S2: Data preprocessing and feature engineering. The collected data is cleaned and normalized, key feature variables are extracted, and a feature dataset is constructed. Among these steps, the hydrogen trapping index of inclusions is calculated based on the non-metallic inclusion feature data. The normalization process employs a linear mapping method, directly dividing the percentage data by 100 to convert it into a dimensionless value between [0,1]; the vacuum degree data is scaled proportionally with 100Pa as the reference, in order to achieve unified normalization of data with different dimensions. Step S3: Hydrogen content prediction. Input the real-time collected data into the prediction model and use the prediction formula to calculate the predicted value of hydrogen content in molten steel. ; Step S4: Process control and feedback, set the hydrogen content safety threshold, compare the predicted value with the threshold, and dynamically adjust the RH vacuum refining process parameters or billet slow cooling strategy.

2. A method of predicting and controlling hydrogen content of rail steel according to claim 1, characterized in that: In step S1, the full parameter data of the refining process includes at least: the vacuum degree change curve during the RH vacuum treatment process, the high vacuum holding time, the ultimate vacuum degree, the boosting gas flow rate, the oxygen content at the beginning and end of decarburization, and the steel temperature change curve.

3. A method of predicting and controlling hydrogen content of rail steel according to claim 1, characterized in that: In the step S2, the inclusion hydrogen trap index The calculation formula is: in: For the first unit area Number density of inclusions, unit: inclusions / mm 2 ; For the first Average projected area of ​​inclusions, unit: ; This is the hydrogen capture coefficient related to the type of inclusion.

4. A method of predicting and controlling hydrogen content of rail steel according to claim 3, characterized in that: The hydrogen capture coefficient The value is: Type A sulfide inclusions: K A =0.5; Type B alumina inclusions: K B =1.2; Class C silicate inclusions: K C =0.4; Type D spherical oxide inclusions: K D =0.

8.

5. A method of predicting and controlling hydrogen content of rail steel according to claim 1, characterized in that: In step S3, the prediction formula is as follows: wherein: is the base hydrogen content equivalent; is the environmental correction factor; is the degassing rate constant; is the RH high vacuum holding time; is the vacuum efficiency factor; is the model correction factor.

6. The method for predicting and controlling the hydrogen content of steel rails according to claim 5, characterized in that: The base hydrogen content equivalent The calculation formula is: wherein, is the relative humidity of the environment, is the scrap steel addition ratio, is the total amount of alloy addition, is the weight coefficient.

7. A method of predicting and controlling hydrogen content of rail steel according to claim 1, characterized in that: In step S4, the specific method for dynamically adjusting the RH vacuum refining process parameters is as follows: if the predicted hydrogen content exceeds the standard upper limit, the system calculates the degassing factor that needs to be increased based on the prediction formula and automatically generates a control command to extend the RH high vacuum holding time.

Images