Method for judging decarburization end point of RH refining process

Abstract

The invention relates to a method for judging a decarburization end point of an RH refining process, and belongs to the technical field of steelmaking production methods. According to the technical scheme, the method comprises the following steps: mounting a laser flue gas analyzer; an industrial control computer is adopted in the field data processing center, data processing software developed independently is operated, and collection and analysis of key data of a production field are achieved; constructing a decarburization end point judgment model, and performing model decision-making of a model decision-making layer; and determining a decarburization end point, and executing layer response. The method has the beneficial effects that the change of CO concentration in flue gas is monitored in real time through the laser gas analyzer mounted on the flue, meanwhile, vacuum degree data in the vacuum chamber fed back by the primary PLC system is collected, and the decarburization end point of the current heat is automatically and accurately judged in combination with the constructed decarburization end point judgment model.

Description

Technical Field

[0001] This invention relates to a method for determining the decarburization endpoint in the RH refining process, belonging to the technical field of steelmaking production methods. Background Technology

[0002] With the continuous improvement of the performance requirements of high-end manufacturing industries such as automobiles, home appliances, and electronics, ultra-low carbon steel (ULC steel) has become an important raw material in the fields of high-end cold-rolled sheets and high-grade electrical steel due to its excellent deep-drawing properties, weldability, and surface quality.

[0003] In the RH decarburization process, carbon in molten steel reacts with oxygen introduced under vacuum conditions to generate CO gas, thus achieving decarburization. However, in actual production, accurately determining whether the decarburization reaction is complete is a key issue affecting product quality stability. Traditional methods mainly rely on manual experience or offline sampling analysis to determine the decarburization endpoint. This approach suffers from problems such as response lag, strong subjectivity in judgment, and low control precision, making it difficult to meet the requirements of precise carbon content control for ultra-low carbon steel.

[0004] Therefore, the metallurgical field urgently needs to develop a method for automatically determining the decarburization endpoint in the RH refining process for ultra-low carbon steel products, in order to improve the accuracy, stability, and automation level of decarburization control and adapt to the trend of high-quality development in the modern steel industry. This is because:

[0005] 1. The need to improve the level of production automation

[0006] With the development of intelligent manufacturing and Industry 4.0, the steel industry is increasingly demanding process automation and intelligence. Traditional manual judgment methods can no longer meet the needs of high-paced, continuous production, and there is an urgent need for a method that can automatically identify the end of decarburization in order to achieve closed-loop control of the RH refining process.

[0007] 2. The need to improve the stability of molten steel quality

[0008] Insufficient decarburization will cause the carbon content in the steel to exceed the specified range, thus adversely affecting the mechanical properties of the subsequently rolled products. Excessive decarburization, on the other hand, will cause a significant drop in the temperature of the molten steel, necessitating extended oxygen blowing time and increased oxygen volume in subsequent processing stages to raise the temperature. This process can easily lead to an increase in the number of inclusions in the molten steel, thereby affecting the surface quality of the final rolled product. Therefore, accurately determining the end point of the decarburization stage in RH refining is of paramount importance for ensuring precise control of the molten steel composition, maintaining consistent product quality, and improving production efficiency.

[0009] 3. Optimize energy consumption and reduce costs

[0010] The RH (reverse carbonization) process consumes significant amounts of steam and electricity. Inaccurate decarburization time control can not only increase energy consumption but also potentially shorten equipment lifespan. Automatic identification technology helps shorten processing cycles and reduce the cost per ton of steel.

[0011] 4. The development of data acquisition and analysis technologies has driven...

[0012] With the maturity of sensor technology, big data analysis, and artificial intelligence algorithms, it has become possible to collect RH operating parameters in real time using online monitoring systems (such as gas flow, vacuum, and temperature changes) and predict the decarbonization endpoint through models.

[0013] In conclusion, developing an automatic method for determining the end of decarburization in the RH refining process based on multi-parameter fusion analysis not only helps improve the intelligence level of the metallurgical process, but also effectively ensures product quality and optimizes resource allocation, which is one of the important directions for the transformation and upgrading of the steel industry.

[0014] In recent years, some domestic steel companies have begun to focus on the research and development of ultra-low carbon steel products and have proposed a series of innovative methods. For example, patent CN119296664A introduces a method for determining the decarburization endpoint of automotive steel sheets using a dual decarburization judgment model. This method first monitors the change in vacuum level in a vacuum chamber, calculating the rate of change of vacuum level every 0.1 to 0.5 minutes. Once the actual measured vacuum level drops below a certain threshold, the first decarburization model will initiate its judgment process. If the calculated rate of change of vacuum level is also lower than the set critical value and reaches the required cumulative points, the second decarburization model will intervene.

[0015] The second decarburization model is constructed based on detailed statistical regression analysis, considering the influence of multiple factors, including oxygen blowing rate (V1), decarburization time (T1), maintenance time under vacuum conditions ≤266 Pa (T2), scrap steel quantity (M1), aluminum addition quantity (M2), carbon powder quantity (M3), and oxygen content (C1) in the final decarburization stage. By periodically collecting and calculating data on these variables, an estimated value of the current carbon content in the molten steel can be obtained. When this estimated value is equal to or lower than the preset target value, the decarburization process can be considered complete.

[0016] However, it is worth noting that since the above methods are mainly developed based on internal data from individual enterprises, their general applicability may be limited. This may lead to deviations in the application results in other steel enterprises, and further verification and adjustment of their accuracy and adaptability are needed to make them more universal in dealing with a wide range of RH refining scenarios.

[0017] Furthermore, during the RH refining process, some companies have noticed that changes in carbon monoxide (CO) concentration in the furnace flue gas most directly reflect the progress of the carbon-oxygen reaction within the furnace. Therefore, they have begun to experiment with installing laser flue gas analyzers on the flue to determine the endpoint of the decarburization process for ultra-low carbon steel. For example, patent CN119351677A proposes a method for controlling the smelting of ultra-low carbon steel based on a decarburization platform in the RH process. This method sets clear requirements for process parameters during the RH refining stage: the temperature of the molten steel at the RH arrival point should be controlled between 1615 and 1635°C; the evacuation time should not exceed 4 minutes; the vacuum degree should reach below 266 Pa; and the fluctuation and rebound amplitude of the vacuum degree should be controlled within 100 Pa throughout the entire vacuum treatment process. By real-time monitoring and analysis of the CO content in the flue gas, a correlation curve between the CO concentration at the RH decarburization endpoint and the carbon content in the molten steel was established, and the decarburization endpoint was determined accordingly. To accelerate the decarburization rate and ensure the decarburization effect, the circulating argon flow rate is adjusted during this process to ensure sufficient circulation of the molten steel. This method can reduce the carbon content of molten steel to ≤10ppm within 15 minutes. Subsequently, a low-carbon high-purity alloy is used for deoxidation and alloying, with the carbon increase strictly controlled to not exceed 5ppm. After alloying, the molten steel must undergo pure degassing treatment for at least 5 minutes and be allowed to stand for ≥5 minutes before being loaded onto the steelmaking process. The final target composition of the molten steel upon removal from the process is: [C]≤15ppm, [H]≤1.2ppm, [O]≤15ppm, [N]≤20ppm.

[0018] Although the patent proposes a technical approach to determine the endpoint of RH decarbonization by monitoring changes in CO concentration in flue gas, it does not specify the implementation details or data analysis methods.

[0019] Furthermore, some literature proposes using a CO concentration in the flue gas during RH refining below a certain set value (e.g., 0.1%) as the criterion for the end of decarburization in ultra-low carbon steel. However, in practice, the flue gas often contains a large amount of dust or a mixture of dust and water vapor, which can easily affect the laser transmission effect of the laser flue gas analyzer. This interference makes it difficult to ensure that the CO concentration stably drops below the set value in the later stages of RH decarburization in some heats, thus adversely affecting the accurate determination of the decarburization endpoint and threatening the stability of the entire heat production. Summary of the Invention

[0020] The purpose of this invention is to provide a method for determining the decarburization endpoint in the RH refining process. This method utilizes a laser gas analyzer installed on the flue to monitor changes in CO concentration in the flue gas in real time, while simultaneously collecting vacuum data from the vacuum chamber fed back by the primary PLC system. Combined with the constructed decarburization endpoint determination model, this method achieves automatic and accurate determination of the decarburization endpoint for the current furnace batch. This optimizes energy consumption, reduces costs, and enhances applicability and flexibility in different steel enterprises, providing possibilities for a wide range of applications. It meets the current needs of the steel industry's transformation and upgrading, and has significant practical significance and application value, effectively solving the aforementioned problems existing in the background technology.

[0021] The technical solution of this invention is: a method for determining the decarburization endpoint in an RH refining process, comprising the following steps:

[0022] (1) Install a laser flue gas analyzer;

[0023] (2) Construction of on-site data processing center: Construct a data processing layer model. The on-site data processing center adopts an industrial control computer and runs self-developed data processing software to realize the collection and analysis of key data on the production site.

[0024] (3) Construct a decarbonization endpoint determination model and conduct model decision-making at the model decision layer;

[0025] (4) Determine the decarbonization endpoint and execute the layer response.

[0026] In step (1), a laser flue gas analyzer is installed on the flue after the main valve of the vacuum chamber of the RH refining equipment, 3 to 5 meters away from the vacuum chamber. During installation, ensure that the analyzer probe is perpendicular to the flue axis and that the probe purging device is running continuously to avoid dust adhesion. Configure the vacuum degree data interface, connect the vacuum degree sensor of the first-level PLC system to the data transmission layer through hard wiring, and use the Modbus-RTU protocol to read the vacuum degree data in real time with a sampling frequency of 1Hz.

[0027] In step (2), the specific model is as follows:

[0028] (a) Data acquisition model, acquisition frequency: 1 time / second, synchronously acquiring CO concentration value from the laser analyzer and vacuum value from the PLC; data interface: the laser analyzer is connected via RS485 interface, and the vacuum value is connected via Ethernet interface, the data frame format is "timestamp + CO concentration + vacuum + check bit";

[0029] (b) Outlier removal model based on the 3σ criterion: Calculate the mean μ and standard deviation σ for CO concentration data from three consecutive periods. If the current data C... t Satisfy | C tIf -μ|>3σ, it is judged as an outlier and the previous valid data is used for interpolation; Vacuum degree anomaly judgment: if the vacuum degree data exceeds the range of 0 to 1000 mbar for 5 consecutive seconds, the sensor fault alarm is triggered and the model calculation is suspended.

[0030] (c) FIFO data storage model: A circular queue of length 10 is constructed to store the latest CO concentration data, denoted as: Q = [C t-9 C t-8 ,...,C t Each time a new data point is collected, the oldest data point is removed to ensure that the queue always maintains the CO concentration sequence of the most recent 10 seconds.

[0031] In step (3), when the vacuum degree P t When the temperature first drops to 20 mbar, triggering the threshold, the model starts and performs the following double verification:

[0032] (a) Slope Analysis Sub-model

[0033] The model determines whether the CO concentration has entered a stable phase, that is, the carbon-oxygen reaction tends to stop. Its calculation logic is as follows: the least squares method is used to perform linear regression on 10 data points in the FIFO list Q, and the fitting equation is C = k·t + b, where k is the slope and t is the time series.

[0034] Its stability determination condition is: when |k|≤0.01% / s, the counter S is incremented by 1; otherwise, the counter S is reset to 0. When S accumulates 10 times, the slope condition is met.

[0035] The formula is:

[0036] slope

[0037] in:

[0038] n = 10 (data window length)

[0039] x i This is a time series (unit: seconds, values ​​0-9).

[0040] y i The corresponding CO concentration value at that time (unit: %)

[0041] (b) Time Accumulation Sub-model

[0042] This model ensures that a low vacuum environment is maintained to guarantee thorough decarburization. Its timing logic is as follows:

[0043] When P t When P is ≤20 mbar, start timer T to begin accumulating time; if P t >20mbar, timer pauses; when Pt When the value is ≤20 mbar again, the timer continues to accumulate; when T≥300 seconds, the time condition is met.

[0044] (c) Logical Judgment Sub-model

[0045] When both the slope and time conditions are met, the model outputs the decarbonization endpoint signal; if either condition is not met, the model continues to cycle and detect until both conditions are met.

[0046] In step (4), after the system determines the decarbonization endpoint of the RH refining process through the decarbonization endpoint determination model, the system responds in the following way:

[0047] (a) The sound and light alarm module is activated, including a flashing red indicator light and a buzzer alarm, to notify the on-site operators to perform subsequent operations to detect the temperature and oxygen content of the molten steel.

[0048] (b) Send a digital signal to the on-site primary PLC system to trigger the RH refining process to automatically enter the next stage of operation.

[0049] The beneficial effects of this invention are as follows: By using a laser gas analyzer installed on the flue, the change in CO concentration in the flue gas is monitored in real time. At the same time, the vacuum level data in the vacuum chamber fed back by the primary PLC system is collected. Combined with the constructed decarburization endpoint determination model, the decarburization endpoint of the current furnace is automatically and accurately determined, which optimizes energy consumption, reduces costs, and enhances applicability and flexibility in different steel enterprises. It provides possibilities for a wide range of application scenarios, meets the current needs of the steel industry's transformation and upgrading, and has important practical significance and application value. Attached Figure Description

[0050] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0051] To make the purpose, technical solutions, and advantages of the invention's embodiments clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only a small part of the embodiments of the present invention, not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the protection scope of the present invention.

[0052] A method for determining the decarburization endpoint in an RH refining process includes the following steps:

[0053] (1) Install a laser flue gas analyzer;

[0054] (2) Construction of on-site data processing center: Construct a data processing layer model. The on-site data processing center adopts an industrial control computer and runs self-developed data processing software to realize the collection and analysis of key data on the production site.

[0055] (3) Construct a decarbonization endpoint determination model and conduct model decision-making at the model decision layer;

[0056] (4) Determine the decarbonization endpoint and execute the layer response.

[0057] In step (1), a laser flue gas analyzer is installed on the flue after the main valve of the vacuum chamber of the RH refining equipment, 3 to 5 meters away from the vacuum chamber. During installation, ensure that the analyzer probe is perpendicular to the flue axis and that the probe purging device is running continuously to avoid dust adhesion. Configure the vacuum degree data interface, connect the vacuum degree sensor of the first-level PLC system to the data transmission layer through hard wiring, and use the Modbus-RTU protocol to read the vacuum degree data in real time with a sampling frequency of 1Hz.

[0058] In step (2), the specific model is as follows:

[0059] (a) Data acquisition model, acquisition frequency: 1 time / second, synchronously acquiring CO concentration value from the laser analyzer and vacuum value from the PLC; data interface: the laser analyzer is connected via RS485 interface, and the vacuum value is connected via Ethernet interface, the data frame format is "timestamp + CO concentration + vacuum + check bit";

[0060] (b) Outlier removal model based on the 3σ criterion: Calculate the mean μ and standard deviation σ for CO concentration data from three consecutive periods. If the current data C... t Satisfy | C t If -μ|>3σ, it is judged as an outlier and the previous valid data is used for interpolation; Vacuum degree anomaly judgment: if the vacuum degree data exceeds the range of 0 to 1000 mbar for 5 consecutive seconds, the sensor fault alarm is triggered and the model calculation is suspended.

[0061] (c) FIFO data storage model: A circular queue of length 10 is constructed to store the latest CO concentration data, denoted as: Q = [C t-9 C t-8 ,...,C t Each time a new data point is collected, the oldest data point is removed to ensure that the queue always maintains the CO concentration sequence of the most recent 10 seconds.

[0062] In step (3), when the vacuum degree P t When the temperature first drops to 20 mbar, triggering the threshold, the model starts and performs the following double verification:

[0063] (a) Slope Analysis Sub-model

[0064] The model determines whether the CO concentration has entered a stable phase, that is, the carbon-oxygen reaction tends to stop. Its calculation logic is as follows: the least squares method is used to perform linear regression on 10 data points in the FIFO list Q, and the fitting equation is C = k·t + b, where k is the slope and t is the time series.

[0065] Its stability determination condition is: when |k|≤0.01% / s, the counter S is incremented by 1; otherwise, the counter S is reset to 0. When S accumulates 10 times, the slope condition is met.

[0066] The formula is:

[0067] slope

[0068] in:

[0069] n = 10 (data window length)

[0070] x i This is a time series (unit: seconds, values ​​0-9).

[0071] y i The corresponding CO concentration value at that time (unit: %)

[0072] (b) Time Accumulation Sub-model

[0073] This model ensures that a low vacuum environment is maintained to guarantee thorough decarburization. Its timing logic is as follows:

[0074] When P t When P is ≤20 mbar, start timer T to begin accumulating time; if P t >20mbar, timer pauses; when P t When the value is ≤20 mbar again, the timer continues to accumulate; when T≥300 seconds, the time condition is met.

[0075] (c) Logical Judgment Sub-model

[0076] When both the slope and time conditions are met, the model outputs the decarbonization endpoint signal; if either condition is not met, the model continues to cycle and detect until both conditions are met.

[0077] In step (4), after the system determines the decarbonization endpoint of the RH refining process through the decarbonization endpoint determination model, the system responds in the following way:

[0078] (a) The sound and light alarm module is activated, including a flashing red indicator light and a buzzer alarm, to notify the on-site operators to perform subsequent operations to detect the temperature and oxygen content of the molten steel.

[0079] (b) Send a digital signal to the on-site primary PLC system to trigger the RH refining process to automatically enter the next stage of operation.

[0080] In practical applications, this invention uses a laser flue gas analyzer installed in the RH refining flue to collect CO concentration in real time. Combined with vacuum data from the primary PLC system, the data is preprocessed at the on-site data processing center and then input into the decarbonization endpoint determination model for dual-condition verification. Finally, the endpoint determination result is output. For detailed steps, please refer to [link to detailed steps]. Figure 1 .

[0081] The specific implementation steps are as follows:

[0082] Step 1: Install the laser flue gas analyzer

[0083] 1. Installation of laser flue gas analyzer

[0084] Install a laser flue gas analyzer (selection parameters: measurement range 0-50% CO, accuracy ±0.01%, response time ≤1s, laser wavelength 1.57μm to reduce water vapor interference) on the flue after the main valve of the vacuum chamber of the RH refining equipment, 3-5 meters away from the vacuum chamber. During installation, ensure that the analyzer probe is perpendicular to the flue axis and that the probe purging device (compressed air pressure 0.4-0.6MPa) is continuously running to prevent dust adhesion.

[0085] 2. Vacuum Degree Data Interface Configuration

[0086] The vacuum sensor (measurement range 0-1000mbar, accuracy ±1mbar) of the primary PLC system is connected to the data transmission layer via hardwiring. The vacuum data is read in real time using the Modbus-RTU protocol with a sampling frequency of 1Hz.

[0087] Step 2: Construction of the on-site data processing center (data processing layer model)

[0088] The on-site data processing center uses industrial control computers and runs self-developed data processing software to collect and analyze key data from the production site. The specific model is as follows:

[0089] 1. Data Acquisition Model

[0090] Acquisition frequency: 1 time / second, synchronously acquiring CO concentration values ​​(denoted as C) from the laser analyzer. t (unit: %) and the vacuum level value on the PLC (denoted as P). t (unit: mbar).

[0091] Data interface: The laser analyzer is connected via RS485 interface, and the vacuum level is connected via Ethernet interface. The data frame format is "timestamp + CO concentration + vacuum level + check digit".

[0092] 2. Outlier Handling Model

[0093] Outlier removal based on the 3σ criterion: Calculate the mean μ and standard deviation σ for CO concentration data from three consecutive periods. If the current data C... t Satisfy | C t If -μ|>3σ, it is considered an outlier and the previous valid data is used for interpolation.

[0094] Vacuum degree anomaly detection: If the vacuum degree data exceeds the range of 0 to 1000 mbar for 5 consecutive seconds, a sensor fault alarm will be triggered, and the model calculation will be suspended.

[0095] 3. FIFO Data Storage Model

[0096] Construct a circular queue (FIFO list) of length 10 to store the latest CO concentration data, denoted as: Q = [C t-9 C t-8 ,...,C t Each time a new data point is collected, the oldest data point is removed to ensure that the queue always maintains the CO concentration sequence of the most recent 10 seconds.

[0097] Step 3: Decarbonization endpoint determination model (model decision layer)

[0098] When the vacuum degree P t When the temperature first drops to 20 mbar (the trigger threshold), the model automatically starts and performs the following double verification:

[0099] 1. Slope Analysis Sub-model

[0100] This model determines whether the CO concentration has entered a stable phase (the carbon-oxygen reaction tends to stop). Its calculation logic is as follows: A linear regression is performed on the 10 data points in the FIFO list Q using the least squares method, and the fitted equation is C = k·t + b, where k is the slope (unit: % / s) and t is the time series (0–9 seconds).

[0101] The stability condition is as follows: when |k|≤0.01% / s (slope threshold, determined based on field process tests), the counter S is incremented by 1; otherwise, the counter S is reset to 0. When S accumulates 10 times (i.e., the stability condition is met for 10 consecutive seconds), the slope condition is established.

[0102] The formula is:

[0103] slope

[0104] in:

[0105] n = 10 (data window length)

[0106] x iThis is a time series (unit: seconds, values ​​0-9).

[0107] y i The corresponding CO concentration value at that time (unit: %)

[0108] 2. Time Accumulation Sub-model

[0109] This model ensures a low vacuum environment is maintained to guarantee thorough decarburization. Its timing logic is as follows:

[0110] When P t When P is ≤20 mbar, start timer T to begin accumulating time; if P t >20mbar, timer pauses (cumulative time does not reset); when P t When the temperature drops below 20 mbar again, the timer continues to accumulate. The time condition is met when T ≥ 300 seconds.

[0111] 3. Logical Judgment Sub-model

[0112] When both the slope condition (S = 10) and the time condition (T ≥ 300 seconds) are met, the model outputs a decarbonization endpoint signal (high-level signal); if either condition is not met, the model continues to cycle and detect until both conditions are met.

[0113] Step 4: Determine the decarbonization endpoint and execute the layer response.

[0114] Once the system determines the decarburization endpoint of the RH refining process using the decarburization endpoint determination model, the system responds in the following manner:

[0115] (1) The sound and light alarm module is activated (red indicator light flashes + buzzer alarm) to notify the on-site operators to perform subsequent operations to detect the temperature and oxygen content of the molten steel.

[0116] (2) Send digital signals to the on-site primary PLC system to trigger the RH refining process to automatically enter the next stage of operation.

[0117] To better illustrate the application of this invention, three specific embodiments are provided. In these three embodiments, a laser flue gas analyzer is first installed after the main valve of the RH refining vacuum chamber, and a dedicated computer system is configured on-site. This system collects CO concentration data provided by the laser flue gas analyzer and vacuum level values ​​from the primary PLC system once per second.

[0118] Example 1:

[0119] Step 3: When the system detects that the vacuum level has dropped to 20 mbar, it automatically starts the decarbonization endpoint determination model to monitor and analyze the changing trends of CO concentration and vacuum level in real time.

[0120] 1. If the slope of the CO concentration change is zero for 10 consecutive times and the counter reaches the preset threshold;

[0121] 2. At the same time, ensure that the vacuum level remains stable below 20 mbar for more than 5 minutes;

[0122] Step 4: Once the above two conditions are met simultaneously, the system determines that the decarburization process of the current heat of ultra-low carbon steel has been completed and sends a completion signal to notify the operator to carry out subsequent processing.

[0123] Example 2

[0124] Step 3: Similarly, when the vacuum level drops to 20 mbar, the decarbonization endpoint determination model is activated to track the changes in CO concentration and vacuum level in real time.

[0125] 1. If the slope of the CO concentration change is found to be zero for 10 consecutive times and the counter reaches the threshold;

[0126] 2. However, the on-site requirements for maintaining a vacuum level below 20 mbar for 5 minutes were not met.

[0127] At this point, the system will temporarily postpone the final judgment and continue to execute step 3 in a loop until the conditions are fully met.

[0128] Step 4: Only when all conditions are met will the system issue a notification that decarbonization has ended and guide the operator to the next stage of the operation.

[0129] Example 3

[0130] Step 3: When the vacuum level drops to 20 mbar, activate the decarbonization endpoint determination model to dynamically analyze the changes in CO concentration and vacuum level.

[0131] 1. If the on-site conditions do not meet the standard that the slope of the CO concentration change is zero for 10 consecutive times (i.e., the counter has not reached the threshold);

[0132] 2. However, if the vacuum level is observed to remain below 20 mbar for more than 5 minutes;

[0133] In this situation, the system will pause making a final judgment and repeat step 3 until the condition is met.

[0134] Step 4: Finally, when both of the above criteria are met, the system confirms that the decarbonization operation is complete and issues instructions to the operator to proceed to the next step.

[0135] This invention achieves automatic and accurate determination of the decarburization process endpoint by real-time monitoring of CO concentration changes in flue gas and combining this with data analysis of vacuum levels within a vacuum chamber. A constructed decarburization endpoint determination model is used for dual verification (slope analysis and time accumulation). This method not only improves accuracy and avoids the subjectivity and lag issues associated with traditional methods relying on manual experience or offline sampling analysis, but also significantly improves production automation and efficiency, adapting to the trend of intelligent development in the modern steel industry. Simultaneously, precise control of the decarburization endpoint helps ensure accurate control of steel composition, maintains product quality consistency, and prevents inaccurate carbon content or improper temperature control from affecting the final product's high quality. Furthermore, this method optimizes energy consumption, reduces costs, and enhances applicability and flexibility in different steel enterprises, enabling a wide range of applications. It meets the current needs of the steel industry's transformation and upgrading, and has significant practical significance and application value.

Claims

1. A method for determining the decarburization endpoint in an RH refining process, characterized in that... Includes the following steps: (1) Install a laser flue gas analyzer; (2) Construction of on-site data processing center: Construct a data processing layer model. The on-site data processing center adopts an industrial control computer and runs self-developed data processing software to realize the collection and analysis of key data on the production site. (3) Construct a decarbonization endpoint determination model and conduct model decision-making at the model decision layer; (4) Determine the decarbonization endpoint and execute the layer response.

2. The method for determining the decarburization endpoint in the RH refining process according to claim 1, characterized in that: In step (1), a laser flue gas analyzer is installed on the flue after the main valve of the vacuum chamber of the RH refining equipment, 3 to 5 meters away from the vacuum chamber. During installation, ensure that the analyzer probe is perpendicular to the flue axis and that the probe purging device is running continuously to avoid dust adhesion. Configure the vacuum degree data interface, connect the vacuum degree sensor of the first-level PLC system to the data transmission layer through hard wiring, and use the Modbus-RTU protocol to read the vacuum degree data in real time with a sampling frequency of 1Hz.

3. The method for determining the decarburization endpoint in the RH refining process according to claim 1, characterized in that: In step (2), the specific model is as follows: (a) Data acquisition model, acquisition frequency: 1 time / second, synchronously acquiring the CO concentration value of the laser analyzer and the vacuum value on the PLC; data interface: the laser analyzer is connected via RS485 interface, and the vacuum value is connected via Ethernet interface, the data frame format is "timestamp + CO concentration + vacuum + check bit"; (b) Outlier removal model based on the 3σ criterion: Calculate the mean μ and standard deviation σ for CO concentration data from three consecutive periods. If the current data C... t Satisfy | C t If -μ|>3σ, it is judged as an outlier and the previous valid data is used for interpolation; Vacuum degree anomaly judgment: if the vacuum degree data exceeds the range of 0 to 1000 mbar for 5 consecutive seconds, the sensor fault alarm is triggered and the model calculation is suspended. (c) FIFO data storage model: A circular queue of length 10 is constructed to store the latest CO concentration data, denoted as: Q = [C t-9 C t-8 ,...,C t Each time a new data point is collected, the oldest data point is removed to ensure that the queue always maintains the CO concentration sequence of the most recent 10 seconds.

4. The method for determining the decarburization endpoint in the RH refining process according to claim 1, characterized in that: In step (3), when the vacuum degree P t When the temperature first drops to 20 mbar, triggering the threshold, the model starts and performs the following double verification: (a) Slope Analysis Sub-model The model determines whether the CO concentration has entered a stable phase, that is, the carbon-oxygen reaction tends to stop. Its calculation logic is as follows: the least squares method is used to perform linear regression on 10 data points in the FIFO list Q, and the fitting equation is C = k·t + b, where k is the slope and t is the time series. Its stability determination condition is: when |k|≤0.01% / s, the counter S is incremented by 1; otherwise, the counter S is reset to 0. When S accumulates 10 times, the slope condition is met. The formula is: in: n = 10 (data window length) x i This is a time series (unit: seconds, values ​​0-9). y i The corresponding CO concentration value at that time (unit: %) (b) Time Accumulation Sub-model This model ensures that a low vacuum environment is maintained to guarantee thorough decarburization. Its timing logic is as follows: When P t When P is ≤20 mbar, start timer T to begin accumulating time; if P t >20mbar, timer pauses; when P t When the value is ≤20 mbar again, the timer continues to accumulate; when T≥300 seconds, the time condition is met. (c) Logical Judgment Sub-model When both the slope and time conditions are met, the model outputs the decarbonization endpoint signal; if either condition is not met, the model continues to cycle and detect until both conditions are met.

5. The method for determining the decarburization endpoint in the RH refining process according to claim 1, characterized in that: In step (4), after the system determines the decarbonization endpoint of the RH refining process through the decarbonization endpoint determination model, the system responds in the following way: (a) The sound and light alarm module is activated, including a flashing red indicator light and a buzzer alarm, to notify the on-site operators to perform subsequent operations to detect the temperature and oxygen content of the molten steel. (b) Send a digital signal to the on-site primary PLC system to trigger the RH refining process to automatically enter the next stage of operation.

Images