Argon blowing tube anti-burning control method, system and computer program product for RH vacuum refining device

Abstract

The present application relates to the technical field of RH vacuum refining, and particularly relates to a method, system and computer program product for preventing burn loss of an argon blowing pipe of an RH vacuum refining device. The method comprises: establishing a mass balance model based on the mass conservation law according to the circulating flow of the molten steel at the immersion pipe and the ladle lifting speed, establishing a liquid level dynamic model based on the mass balance model, performing Kalman filtering estimation based on the mass balance model and the liquid level dynamic model to obtain a real-time effective liquid level of the molten steel in the ladle, calculating a real-time safety gap value between the end of the argon blowing pipe and the molten steel surface according to the real-time lifting height of the bottom of the ladle, the installation height of the end of the argon blowing pipe and the real-time effective liquid level of the molten steel in the ladle, and adjusting the ladle lifting action according to the real-time safety gap value. The present application realizes continuous monitoring of the dynamic safety gap not only in the lifting stage but also in the refining stage, prevents gradual burn loss caused by the rise of the liquid level, and significantly prolongs the average service life of the argon blowing pipe.

Description

Technical Field

[0001] This invention relates to the field of RH vacuum refining technology, specifically to a method, system, and computer program product for preventing burn-out of the argon blowing tube in an RH vacuum refining apparatus. Background Technology

[0002] The RH vacuum refining unit is the core equipment for the production of ultra-low carbon steel and clean steel.

[0003] Existing RH lifting control systems primarily rely on hydraulic cylinder stroke, encoder pulses, or fixed mechanical limits, using the insertion of the immersion tube as the main criterion for successful lifting. Current technologies generally focus on ensuring the insertion depth of the immersion tube, failing to recognize that the dynamic gap between the end of the argon blowing tube and the molten steel surface also requires closed-loop real-time protection. While ensuring the working depth of the immersion tube is the primary objective, effective monitoring methods for the dynamic safety gap between the end of the argon blowing tube and the molten steel surface are lacking.

[0004] Because some molten steel is drawn into the vacuum tank for circulation during RH operation, the actual remaining liquid level in the ladle is constantly changing. Existing technologies lack direct means to detect the real-time weight of the remaining molten steel in the ladle, making it impossible to accurately perceive the actual liquid level position during operation. Furthermore, direct methods for detecting the actual liquid level position in the ladle (such as laser ranging, visual recognition, and encoder positioning) face implementation challenges under the combined conditions of RH vacuum, high temperature, dust, and molten steel fluctuations, making them unreliable for real-time closed-loop control of the argon blowing tube gap. The safe clearance between the end of the argon blowing tube and the molten steel surface (typically requiring ≥150mm) relies entirely on manual experience or fixed stroke control throughout the lifting and refining process, leading to a risk of argon blowing tube burnout. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by providing a method for preventing burn-out of the argon blowing tube in an RH vacuum refining apparatus, comprising: Based on the molten steel circulation flow rate at the immersion tube and the ladle lifting speed, a mass balance model is established based on the law of mass conservation. A liquid level dynamic model is then established based on the mass balance model. Kalman filtering estimation is performed based on the mass balance model and the liquid level dynamic model to obtain the real-time effective liquid level of molten steel in the ladle. Calculate the real-time safety gap value between the end of the argon blowing pipe and the molten steel surface based on the real-time lifting height of the bottom of the ladle, the installation height of the end of the argon blowing pipe, and the real-time effective liquid level of the molten steel in the ladle. Adjust the ladle lifting action according to the real-time safety gap value.

[0006] Furthermore, the mass balance model is specifically represented as follows: In the formula, for The mass of molten steel remaining in the ladle at all times. The density of molten steel, for The circulation flow rate of molten steel at the immersion tube is constantly monitored. This represents the cross-sectional area of ​​the ladle's inner cavity. for The ladle lifting speed at all times for Process noise in the time-matter balance model; The specific representation of the liquid level dynamic model is as follows: In the formula, for Monitor the molten steel level in the ladle at all times. for Process noise in the dynamic model of liquid level at any time.

[0007] Furthermore, the specific method for obtaining the real-time effective liquid level of molten steel in the ladle by Kalman filtering estimation based on the mass balance dynamics model is as follows: Define the first mass Define a second mass to represent the mass of an empty ladle. Define the initial molten steel depth as the total mass of the ladle containing molten steel after steel is received and before it is lifted. For the second mass The corresponding molten steel level in the ladle Define the third mass for The real-time mass of molten steel remaining in the ladle under RH vacuum refining conditions will be measured. As observation data for Kalman filtering, ; Based on the mass balance model and the liquid level dynamic model, the discrete state transition equation of the Kalman filter is obtained, as shown in the following formula: In the formula, for The mass of molten steel remaining in the ladle at all times. for Monitor the molten steel level in the ladle at all times. The discrete state transition matrix, , For discrete control input matrix, , The sampling period; Prior estimation is performed based on discrete state equations: In the formula, for The optimal posterior estimate of the remaining molten steel mass in the ladle is obtained by using all sensor data as observations at all times. for Time based The optimal posterior estimate of the remaining molten steel mass in the ladle at time t is used to obtain the prior predicted value of the remaining molten steel mass in the ladle. for The optimal posterior estimate of the molten steel level in the ladle is obtained by using all sensor data as observations at all times. for Time based The optimal posterior estimate of the molten steel level in the ladle at time t is used to obtain the prior prediction of the molten steel level in the ladle. The prior covariance update formula is as follows: In the formula, for Covariance of the mass of remaining molten steel in the ladle under the optimal posterior estimate at time. for Covariance of the mass of remaining molten steel in the ladle as predicted a priori at each moment. for Covariance of molten steel level in ladle under the optimal posterior estimate at time. for Covariance of molten steel level in ladle predicted a priori at each moment. for The covariance between the estimation error of the remaining molten steel mass in the ladle under the optimal posterior estimation at a given time and the estimation of the molten steel level in the ladle. for The covariance between the prior prediction error of the remaining molten steel mass in the ladle and the estimation of the molten steel level in the ladle. for The linked covariance of the estimation error of the molten steel level in the ladle and the estimation of the mass of the remaining molten steel in the ladle under the optimal posterior estimation at the given time. for The covariance between the prior prediction error of the molten steel level in the ladle and the estimation of the remaining molten steel mass in the ladle. It is the transpose symbol. The process noise variance for the mass balance model. The process noise variance of the liquid level dynamic model; With the second mass As The initial value for the iteration is the initial depth of the molten steel. As The initial values ​​for iteration are used for posterior estimation, as shown in the following formula: In the formula, for Moment-time Kalman gain, This represents the best a posteriori estimate of the remaining molten steel mass in the ladle at the current moment. This is the post-hoc optimal estimate of the molten steel level in the ladle at the current moment. The posterior covariance update formula is: In the formula, It is the identity matrix. The variance of the posterior estimate of the mass of molten steel remaining in the ladle at the current moment. The posterior covariance of the estimation error of the remaining molten steel mass in the ladle with respect to the estimation of the molten steel level in the ladle. This is the posterior covariance of the estimation error of the molten steel level in the ladle with respect to the estimation of the mass of the remaining molten steel in the ladle. The variance of the liquid level at the current moment is estimated posteriorly. Through iteration, the final output is... and , As The effective mass of molten steel in the ladle at all times. As The effective liquid level of molten steel in the ladle at all times.

[0008] Furthermore, the third mass It is obtained by performing digital low-pass filtering and inertial compensation processing on the corresponding original quality signal.

[0009] Furthermore, Time Kalman Gain Specifically: In the formula, To measure the noise covariance.

[0010] Furthermore, the specific method for calculating the real-time safe gap value between the end of the argon blowing tube and the molten steel surface is as follows: In the formula, Let t be the dynamic safety gap value between the end of the argon blowing tube and the molten steel surface. The installation height of the argon blowing tube end. This refers to the real-time lifting height of the bottom of the ladle.

[0011] Furthermore, the specific method for adjusting the ladle lifting action based on the real-time safety gap value is as follows: when ≤ First gap threshold At the same time, control the ladle lifting speed. The first speed; When the first gap threshold < ≤Second gap threshold At the same time, control the ladle lifting speed. The second speed is less than the first speed. When the second gap threshold < ≤Third gap threshold At that time, the ladle lifting was stopped; when >Third gap threshold At that time, the ladle was controlled to descend at a third speed until... ≤ First gap threshold ; First gap threshold >Second gap threshold >Third gap threshold .

[0012] Furthermore, the method for preventing the argon blowing tube from burning out is divided into several execution stages. Each execution stage is completed within a preset execution time window. If an execution stage is not completed within the preset execution time window, the ladle is reset until... ≤ First gap threshold .

[0013] A burn-out prevention control system for the argon blowing tube of an RH vacuum refining unit includes: The effective liquid level calculation module is used to establish a mass balance model based on the law of conservation of mass, based on the molten steel circulation flow rate at the immersion pipe and the ladle lifting speed, and to establish a liquid level dynamic model based on the mass balance model. Based on the mass balance model and the liquid level dynamic model, Kalman filtering is performed to estimate the real-time effective liquid level of molten steel in the ladle. The real-time safety gap value calculation module is used to calculate the real-time safety gap value between the end of the argon blowing pipe and the molten steel surface based on the real-time lifting height of the bottom of the ladle, the installation height of the end of the argon blowing pipe, and the real-time effective liquid level of the molten steel in the ladle. The ladle lifting adjustment module is used to adjust the ladle lifting action according to the real-time safety gap value.

[0014] A computer program product includes a computer program / instructions that, when executed by a processor, implement the above-described method for preventing burn-out of the argon blowing tube in the RH vacuum refining apparatus.

[0015] The beneficial effects of this invention are as follows: 1. By detecting the real-time weight (third weight value) of the remaining molten steel in the ladle during the working state, the dynamic quantitative calculation of the actual molten steel level in the ladle during the RH cycle is realized, solving the problem of unmeasurable liquid level caused by the circulation of molten steel between the vacuum tank and the ladle.

[0016] 2. Dynamic safety gaps are continuously monitored not only during the lifting stage but also during the refining stage to prevent gradual burn-off caused by rising liquid levels. The average service life of the argon blowing tube can be significantly extended (based on process simulation calculations, it is expected to increase by about 2-3 times); the burn-off accident rate caused by argon blowing tube immersion is expected to decrease by more than 95% (based on simulation statistics of typical working conditions).

[0017] 3. Prevent refractory erosion and foreign matter intrusion caused by argon blowing tube burn-out, reduce the risk of secondary contamination of molten steel, and significantly reduce the molten steel return rate caused by argon blowing tube burn-out (based on process simulation calculations, it is expected to drop to below 0.1%).

[0018] 4. Liquid level calculation based on weight signals is not affected by factors such as refractory corrosion of the ladle, changes in molten steel density, and temperature fluctuations, and has higher accuracy and adaptability to working conditions compared with traditional position control. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the argon blowing tube anti-burn-off control method of the RH vacuum refining apparatus of the present invention.

[0020] Figure 2 This is a flowchart of the method for preventing burn-out of the argon blowing tube in the RH vacuum refining apparatus of the present invention.

[0021] Figure 3 This is a block diagram of the argon blowing tube anti-burn-off control system of the RH vacuum refining apparatus of the present invention.

[0022] Explanation of reference numerals in the attached diagram: 1-vacuum tank; 2-argon blowing tube; 3-left immersion tube; 4-right immersion tube; 5-steel ladle; 6-molten steel; L-steel ladle lifting direction. Detailed Implementation

[0023] To make the technical problems, technical solutions, and beneficial effects to be solved by this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit the scope of this application.

[0024] Example 1 Figure 1This is a schematic diagram of the argon blowing tube anti-burn-off control method of the RH vacuum refining device of the present invention. The vacuum tank 1 is located at the top of the device and is the core reaction chamber of RH refining. It is lined with a refractory lining and maintains a high vacuum environment during operation, providing a vacuum reaction space for steel degassing and metallurgical reaction. The argon blowing tube 2 is arranged in the middle of the lower part of the vacuum tank. Its lower end can be immersed in the steel. During operation, argon gas is blown into the steel, which is the power source for driving the circulation of the steel. The left immersion tube 3 and the right immersion tube 4 are symmetrically arranged on both sides of the argon blowing tube. Their upper ends are connected to the vacuum tank and their lower ends are immersed in the steel. They are the channels for the steel to enter and exit the vacuum tank. They respectively serve as the riser and the faller. The ladle 5 is located at the bottom of the device and is used to hold the molten steel to be refined. It can be lifted upward in the direction of arrow L so that the left immersion tube 3, the right immersion tube 4 and the argon blowing tube 2 are immersed in the steel. The steel liquid 6 is the high-temperature molten steel contained in the ladle and is the object of RH refining.

[0025] A ladle 5 containing molten steel 6 is transported to the bottom of the vacuum tank 1. The ladle 5 is slowly lifted upwards along the L direction, immersing the lower ends of the left immersion tube 3, the right immersion tube 4, and the argon blowing tube 2 into the molten steel 6 to a certain depth. At this time, the gap between the argon blowing tube 2 and the surface of the molten steel 6 is in a safe zone, and the equipment is ready for refining. The vacuum pump is started to evacuate the vacuum tank 1, creating a high negative pressure. Atmospheric pressure forces the molten steel 6 into the left immersion tube 3 and upwards into the vacuum tank 1. Simultaneously, the argon blowing tube 2 continuously blows argon gas into the molten steel 6. The argon gas bubbles expand due to the heat in the high-temperature molten steel 6, driving the surrounding molten steel 6 to flow upwards at high speed along the rising immersion tube and into the vacuum tank 1, creating a "bubble pump" effect. The molten steel 6 entering the vacuum tank 1 undergoes a series of metallurgical reactions under high vacuum. After vacuum treatment, the molten steel 6 flows back to the bottom of the ladle 5 along the right immersion tube 4 under the influence of gravity. The molten steel 6 continuously circulates between the ladle 5 and the vacuum tank 1. The entire ladle of molten steel 6 can undergo multiple circulation processes within a few minutes, ultimately achieving uniform purification of the entire ladle of molten steel 6. After achieving the refining target, the vacuum in the vacuum tank 1 is broken, the ladle 5 descends along the L direction, and the left immersion tube 3, right immersion tube 4, and argon blowing tube 2 detach from the surface of the molten steel 6. The ladle 5 is then transported to the next process.

[0026] Example 2 Current technology lacks a direct means to detect the real-time quality of the remaining molten steel in the ladle, making it impossible to accurately perceive the actual liquid level of the molten steel during operation. The safe clearance (typically required to be ≥150mm) between the end of the argon blowing pipe and the molten steel surface relies entirely on manual experience or fixed stroke control throughout the lifting and refining process, leading to the following typical burn-off risks: Failure Scenario 1: Initial Lifting Fluctuation. During the initial stage of establishing vacuum circulation in RH, the molten steel surface fluctuates violently due to the suction effect, and the end of the argon blowing pipe may have already touched the high-temperature molten steel before the control system responds.

[0027] Failure Scenario 2: Liquid Level Rise During Refining. After the ladle is fully jacked up, changes in the density of the molten steel due to alloying feed or temperature variations cause the liquid level in the ladle to rise continuously. Existing methods cannot identify this upward trend, resulting in the gradual immersion of the argon blowing tube.

[0028] Failure Scenario 3: Delayed Emergency Reset. When the RH is ventilated or descends urgently, the speed at which the argon blowing tube exits lags behind the speed at which the ladle descends, or the molten steel falls back down late, causing the argon blowing tube to be passively immersed in the molten steel.

[0029] To address the above failure scenarios, this invention proposes a method for preventing burn-out of the argon blowing tube in an RH vacuum refining unit, such as... Figure 2 As shown, the process includes: establishing a mass balance model based on the law of conservation of mass based on the circulating flow rate of molten steel at the immersion tube and the lifting speed of the ladle; establishing a liquid level dynamic model based on the mass balance model; and performing Kalman filtering estimation based on the mass balance model and the liquid level dynamic model to obtain the real-time effective liquid level of molten steel in the ladle. Calculate the real-time safety gap value between the end of the argon blowing pipe and the molten steel surface based on the real-time lifting height of the bottom of the ladle, the installation height of the end of the argon blowing pipe, and the real-time effective liquid level of the molten steel in the ladle. Adjust the ladle lifting action according to the real-time safety gap value.

[0030] As one specific implementation method, the mass balance model is specifically represented as follows: In the formula, for The mass of molten steel remaining in the ladle at all times. The density of molten steel, for The circulation flow rate of molten steel at the immersion tube is constantly monitored. This represents the cross-sectional area of ​​the ladle's inner cavity. for The ladle lifting speed at all times for Process noise in the time-matter balance model. This represents minor disturbances not fully covered by the mass balance model, such as slight steel burn-off, small amounts of alloy addition, localized refractory corrosion, and fluctuations in the steel surface. In this embodiment, it is assumed to be zero-mean Gaussian white noise.

[0031] This is a mass transfer term caused by cyclic pumping. This mass flows out of the ladle, resulting in a decrease in the mass of the remaining molten steel in the ladle. Therefore, this term is negative. Vacuum level and lifting speed The higher the vacuum level and the greater the insertion depth of the impregnation tube (the higher the jacking), the stronger the suction capacity and the greater the circulation flow rate.

[0032] The term represents the change in geometric volume caused by the jacking. From the perspective of the mass-liquid level correspondence, the reduction in the effective volume of the ladle is equivalent to the increase in the depth of the molten steel for the same mass. The rate of change in the remaining mass shows an increasing trend, so this term is positive.

[0033] The constant relationship between mass and liquid level Taking the derivative with respect to time yields the dynamic model of the liquid level, which is specifically represented as follows: In the formula, for Monitor the molten steel level in the ladle at all times. for Process noise in the dynamic model of liquid level at any time. The small disturbances that the liquid level dynamic model does not fully cover, such as slight loss of molten steel, small amount of alloy addition, and local erosion and slag removal of refractory materials, which change the total mass of molten steel, are synchronously reflected in the small offset of liquid level, as well as the disturbances of molten steel surface surge and fluctuation, the disturbances of ladle cross-section geometric deviation, the disturbances of liquid surface flow field deformation, and the instantaneous fluctuations of molten steel density. In this embodiment, it is assumed to be zero-mean Gaussian white noise.

[0034] Based on the law of conservation of mass, the derived laws of mass and liquid level evolution perfectly match the real physical process of RH molten steel circulation. The mass balance model and the liquid level dynamic model form a strict one-to-one mapping relationship through the density of molten steel and the cross-sectional area of ​​the ladle. The two state variables are physically self-consistent, providing a natural coupling basis for the two-dimensional joint estimation.

[0035] As one specific implementation method, the method for obtaining the real-time depth of molten steel by Kalman filtering estimation based on the mass balance dynamics model is as follows: (1) Define the first mass Define a second mass to represent the mass of an empty ladle. Define the initial molten steel depth as the total mass of the ladle containing molten steel after steel is received and before it is lifted. For the second mass The corresponding molten steel level in the ladle Define the third mass for The real-time mass of molten steel remaining in the ladle during RH vacuum refining operation. These three mass values ​​are obtained sequentially using mass sensors installed on the crane hoisting mechanism (or ladle car trunnion, ladle carrying platform).

[0036] The aforementioned third quality value This is the core detection quantity of this invention. It differs from the first mass. Second mass Static weighing, third mass value It is a dynamic sensing of the mass of the remaining molten steel in the ladle under the specific operating condition of RH circulation, and its numerical change directly reflects the rise and fall of the actual liquid level in the ladle. Third mass value Reduce first quality What we get is the mass of molten steel remaining in the ladle. The reduction in the mass of molten steel remaining in the ladle is exactly equal to the amount of molten steel circulating into the vacuum tank and immersion tube (mass conservation).

[0037] Among these factors, the force exerted on the wire rope or trunnion during the lifting process under RH vacuum refining conditions is affected by acceleration and deceleration inertial forces, and there is mechanical vibration interference, which affects the third mass. The corresponding raw quality signal is subjected to digital low-pass filtering and inertia compensation processing to obtain an effective value representing the actual mass of molten steel remaining in the ladle. The specific processing procedure is as follows: For the third quality The corresponding original quality signal is subjected to a first-order Butterworth low-pass filter: a cutoff frequency of 1Hz is set to filter out high-frequency noise generated by mechanical vibration of the equipment and violent fluctuations of molten steel, resulting in the filtered original quality signal. Inertial compensation processing specifically involves deducting the interference of the rebalancing signal caused by the inertial forces generated during the acceleration and deceleration of the ladle lifting process. , It is gravitational acceleration.

[0038] (2) Define the state vector Initialize the initial state Before the jacking, all the molten steel in the ladle remains inside the ladle, and no molten steel enters the vacuum tank for circulation.

[0039] Initialize the error covariance matrix , This represents the initial quality estimation error. This represents the initial liquid level estimation error.

[0040] Setting the process noise covariance matrix of the Kalman filter In the formula, the diagonal elements represent the variance of the noise in the corresponding state process. For variance operators, the off-diagonal elements are the covariance of the noise in the two-state process. For covariance operators, The process noise variance for the mass balance model. The process noise variance of the liquid level dynamic model. The coupling covariance of the noise from the mass process of the remaining molten steel in the ladle and the noise from the liquid level process of the molten steel in the ladle describes the linear correlation strength between the random disturbances on the mass side and the random disturbances on the liquid level side. The coupling covariance of the noise from the molten steel level process in the ladle to the noise from the noise from the mass process of the remaining molten steel in the ladle is represented by the reverse dimension of the same coupling relationship. Due to the symmetry of the covariance matrix, it is always numerically consistent. In this embodiment, a diagonal matrix is ​​used, i.e. Assuming that the process noise of mass and liquid level is statistically independent, Process noise covariance matrix It is used to cover minor disturbances that are not accurately modeled in the model, including uncertainties such as minor steel burn-off, minor alloy addition, and small fluctuations in operating conditions.

[0041] Set measurement noise covariance , In this embodiment, the measured noise standard deviation of the mass sensor is 20 kg, which is used to balance the weights of the model predictions and the measured values.

[0042] The actual mass of the remaining molten steel in the ladle was measured. The sole observation data used in this invention's Kalman filtering is employed to correct prediction errors. .

[0043] For the iterative process of Kalman filtering, the definition is... Let be the posterior state vector of the previous time step, based on Corrected from all observation data up to and before the time. The time-optimal state estimation is the starting point for the calculation of this round of prediction; The prior state vector at the current moment is obtained by extrapolation based solely on the posterior state at the previous moment and the mass balance dynamics model. The predicted state at any given moment has not yet been incorporated into the current observation data. The posterior state vector at the current time step is incorporated into the state vector at time step t. The data, and the optimal state estimate at time t obtained after Kalman gain correction, serve as the input benchmark for the next iteration. Definition Let be the posterior error covariance matrix of the previous time step, based on All moments and prior to After the observation data is corrected, the uncertainty quantification result of the optimal estimation of the system state is the final error benchmark output of the previous round of filtering iteration; The prior error covariance matrix at the current time is the prediction error at time t obtained solely from the extrapolation of the posterior benchmark and mass balance dynamics model at the previous time. It has not yet incorporated the observation data at the current time and represents the state uncertainty of the pure model prediction. The posterior error covariance matrix at the current time is the uncertainty quantification result of the optimal estimation of the system state at the current time after incorporating the quality observation data at time t and completing the state correction through Kalman gain. It is the final error benchmark output of this round of filtering iteration and will serve as the input starting point for the next round of prediction iteration.

[0044] (3) The standard form of the continuous state transition equation is: Comparison with mass balance model: Liquid level dynamic model: It can be seen that the continuous state transition matrix The zero matrix indicates that the present invention has a stateless self-evolutionary characteristic, and the state changes are entirely driven by external inputs, continuously controlling the input matrix. , control input vector Kalman filtering process noise .

[0045] Regarding the sampling period The general formula for discretizing linear continuous systems is: Discrete state transition matrix Discrete control input matrix In the present invention Since it is a zero matrix, the discrete state matrix in the present invention is... , It is the identity matrix, and because Since it is an identity matrix, Substitute the continuous control input matrix The discrete control input matrix in the present invention is obtained. .

[0046] Based on the mass balance model and the liquid level dynamic model, the discrete state transition equation of the Kalman filter is obtained, as shown in the following formula: In the formula, for The mass of molten steel remaining in the ladle at all times. for Monitor the molten steel level in the ladle at all times; Decomposing the discrete state transition equation yields two independent equations: an iterative equation for the effective remaining molten steel mass in the ladle and an iterative equation for the effective molten steel level in the ladle. Since... Given the identity matrix, the iterative equation for the effective remaining molten steel mass in the ladle is: The iterative equation for the effective liquid level of molten steel in a ladle is: In this embodiment, and It is Gaussian white noise with a mean of 0 and a covariance that conforms to the Q-constraint matrix.

[0047] (4) Based on the discrete state equations, perform prior estimation: Expanding, we get: In the formula, for The optimal posterior estimate of the remaining molten steel mass in the ladle is obtained by using all sensor data as observations at all times. for Time based The optimal posterior estimate of the remaining molten steel mass in the ladle at time t is used to obtain the prior predicted value of the remaining molten steel mass in the ladle. for The optimal posterior estimate of the molten steel level in the ladle is obtained by using all sensor data as observations at all times. for Time based The optimal posterior estimate of the molten steel level in the ladle at time t is used to obtain the prior prediction of the molten steel level in the ladle. The above equation can be broken down into a mass prediction equation and a liquid level prediction equation, because... The identity matrix is ​​the mass prediction formula, which corresponds to the discrete integral of the continuous mass balance equation, specifically: The liquid level prediction formula corresponds to the discrete integral of the auxiliary equation for the rate of change of liquid level, specifically: .

[0048] The prior covariance update formula is as follows: After unfolding, we get: because The identity matrix simplifies to: In the formula, for Covariance of the mass of remaining molten steel in the ladle under the optimal posterior estimate at time. for Covariance of the mass of remaining molten steel in the ladle as predicted a priori at each moment. for Covariance of molten steel level in ladle under the optimal posterior estimate at time. for Covariance of molten steel level in ladle predicted a priori at each moment. for The covariance between the estimation error of the remaining molten steel mass in the ladle under the optimal posterior estimation at a given time and the estimation of the molten steel level in the ladle. for The covariance between the prior prediction error of the remaining molten steel mass in the ladle and the estimation of the molten steel level in the ladle. for The linked covariance of the estimation error of the molten steel level in the ladle and the estimation of the mass of the remaining molten steel in the ladle under the optimal posterior estimation at the given time. for The covariance between the prior prediction error of the molten steel level in the ladle and the estimation of the remaining molten steel mass in the ladle. It is the transpose symbol; Based on the matrix operation rules, the independent covariance iteration formulas of mass and liquid level are decomposed, which correspond to the state uncertainty update of the continuous mass balance equation and the auxiliary equation of liquid level change rate, respectively.

[0049] The iterative calculation of the covariance of the remaining mass of molten steel corresponds to the update of the perturbation of the mass equilibrium state quantity, specifically as follows: The iteration of the steel molten liquid level covariance corresponds to the update of the perturbation of the liquid level change rate state variable, specifically: The mass-level coupled covariance iteration, with dual-state cross-perturbation update, is as follows: (5) Observation matrix Defined as: This indicates that the observation matrix maps to the physical observation relationship, representing only the mass of the remaining molten steel in the ladle. The molten steel level in the ladle can be observed by a mass sensor. Without direct observation, the results are derived solely from mass. The observed residuals are: The standard formula for Kalman gain is: In this invention, the following is obtained: By decomposing the vector elements, we obtain independent gain expressions for the two dimensions of mass and liquid level. The mass observation correction gain is as follows: The value ranges from 0 to 1. The greater the prior uncertainty and the smaller the observation noise, the greater the gain and the more confident the observation. The indirect gain correction for liquid level is as follows: Since the molten steel level cannot be directly observed, the quality observation residuals are indirectly transmitted to the correction coefficient of the molten steel level state through the quality-level coupling covariance.

[0050] With the second mass As The initial value for the iteration is the initial depth of the molten steel. As The initial values ​​for iteration are used for posterior estimation, as shown in the following formula: After unfolding, we get: In the formula, The optimal posterior estimate of the remaining molten steel mass in the ladle at the current moment, i.e., fusion. After observing the data and comparing it with the mass balance dynamics model, the statistically smallest error in the actual molten steel mass inside the ladle is the final filtered output result in terms of mass dimension. The posterior optimal estimate of the molten steel level in the ladle at the current moment is the actual molten steel depth in the ladle obtained after indirectly correcting the predicted level based on the residuals of mass observation. It is the final filtered output result of the level dimension. By decomposing the vector elements, we obtain independent update formulas for the two state variables: mass and liquid level. The a posteriori update of the remaining mass of molten steel in the ladle is as follows: Based on the prior prediction of the molten steel mass in the ladle, the Kalman gain-weighted mass observation residuals are used to correct the model bias and obtain the optimal estimate of the molten steel mass in the ladle. The updated steel level in the ladle is as follows: Since the molten steel level inside the ladle cannot be directly observed, the predicted molten steel level inside the ladle is indirectly corrected by using the residuals of mass observation and Kalman gain, thereby simultaneously improving the accuracy of the molten steel level estimation inside the ladle.

[0051] The posterior covariance update formula is: In the formula, It is a 2×2 identity matrix, which expands to: The variance of the posterior estimate of the mass of molten steel remaining in the ladle at the current moment is used to characterize the weight of molten steel remaining in the ladle at the current moment. After time-based observation correction, the uncertainty of the estimated mass of the remaining molten steel in the ladle is smaller than the prior variance, demonstrating the effect of observation data on improving the accuracy of mass estimation. The posterior covariance of the mass estimation error of the remaining molten steel in the ladle with respect to the liquid level estimation of the molten steel in the ladle characterizes the degree of positive linear correlation between the mass estimation deviation and the liquid level estimation deviation at the current moment, and reflects the transmission strength of the mass-side error to the liquid level side. Let $\mathbf{a}$ be the posterior covariance of the molten steel level estimation error in the ladle with respect to the remaining molten steel mass estimation. This covariance characterizes the degree of inverse linear correlation between the current time-to-time level estimation bias and mass estimation bias. Equal numerical values ​​indicate the reverse dimension of the same association. The variance of the posterior estimate of the liquid level at the current moment represents the variance obtained through [further context needed]. After indirect correction based on real-time observations, the uncertainty of the estimated molten steel level decreases in tandem with the improvement in the accuracy of the mass estimation.

[0052] After completely decomposing the matrix using the matrix multiplication rule, we obtain four independent iterative expressions with different elements. The posterior variance of the remaining mass of the molten steel is as follows: The post-covariance of the mass-level positive coupling is as follows: Post-hoc covariance of level-mass inverse coupling: Post-test variance of molten steel level: The covariance matrix is ​​naturally symmetric and numerically always satisfies With a single observation The model predictions are corrected to obtain the optimal estimated state at the current moment.

[0053] Through Kalman filtering iteration, the final output is... and , As The effective mass of molten steel in the ladle at all times. As The effective liquid level of molten steel in the ladle at all times.

[0054] The only direct observable in this invention is the mass signal output by the mass sensor, i.e., the only observable. This signal is unaffected by the density of molten steel. Cross-sectional area of ​​the inner cavity of the ladle The parameter error is the purest true reference. If a single liquid level condition scheme is used, the mass observation value must first be converted into the liquid level observation value. Parameter errors are directly added to the observation noise, effectively increasing the observation noise covariance and reducing the estimation accuracy and robustness of the filter. In the two-state framework, the off-diagonal elements of the covariance matrix are used... , The Kalman gain can quantify the transmission strength of mass estimation error to liquid level estimation error. Based on this coupling relationship, it can achieve statistically optimal indirect correction of liquid level status, theoretically guaranteeing the optimality of liquid level estimation. A single liquid level status scheme can only directly correct the liquid level after converting the observed values, failing to reflect the physical process of error transmission. The optimality of the estimation results lacks rigorous theoretical support, and the source of uncertainty in liquid level estimation cannot be quantified.

[0055] As one specific implementation method, In the formula, Let t be the dynamic safety gap value between the end of the argon blowing tube and the molten steel surface. The installation height of the argon blowing tube end. The real-time lifting height of the bottom of the ladle (obtained by a lifting displacement sensor such as an absolute encoder or a hydraulic cylinder stroke sensor).

[0056] It can realize continuous calculation of the gap throughout the entire process of lifting and refining. It not only covers the lifting stage, but also continuously monitors the gap reduction caused by the slow rise of liquid level during refining, thus solving the protection blind spot of the gradual burn-out of the argon blowing tube during the refining stage.

[0057] As a specific implementation method, the specific method for adjusting the ladle lifting action according to the real-time safety gap value is as follows: when ≤ First gap threshold At the same time, control the ladle lifting speed. The first speed; When the first gap threshold < ≤Second gap threshold At the same time, control the ladle lifting speed. The second speed is less than the first speed. When the second gap threshold < ≤Third gap threshold At that time, the ladle lifting was stopped; when >Third gap threshold At that time, the ladle was controlled to descend at a third speed until... ≤ First gap threshold ; First gap threshold >Second gap threshold >Third gap threshold .

[0058] The two-state covariance matrix can output the uncertainty of mass estimation, the uncertainty of liquid level estimation, and the degree of coupling between the two, which can serve as the first gap threshold. Second gap threshold and the third gap threshold The design and margin provisions provide a quantitative basis. For example, when quality observation is disturbed or the prior covariance increases, the system can automatically determine that the reliability of the liquid level estimation has decreased, and actively increase the safety margin and reduce the jacking speed. This enables the protection logic to have adaptive fault tolerance capabilities, further improving the system's fault safety.

[0059] Multi-threshold graded speed regulation achieves a smooth transition during deceleration, avoiding mechanical shock and molten steel surge caused by direct braking with a single threshold; it reduces mechanical wear of the lifting mechanism and also reduces the risk of secondary burn-off caused by violent fluctuations in the liquid surface.

[0060] As a specific implementation method, to ensure the real-time and deterministic nature of argon blowing tube protection, constraints are imposed on the entire link time from physical quantity detection to actuator action. A layered time-triggered architecture is adopted, with each execution stage completed within a preset execution time window. For example, each execution stage is divided into raw signal sampling, raw signal filtering, calculation of the effective liquid level of molten steel in the ladle, dynamic safety clearance judgment, and adjustment of the ladle lifting action based on the real-time safety clearance value.

[0061] To achieve hard real-time protection, the response time of each execution phase must be less than a safety threshold. This safety threshold is determined by the argon blowing tube intruding into the danger zone at a typical jacking speed (second gap threshold). To the third gap threshold The required time is determined (for example, at a lifting speed of 0.2 m / min, it takes about 45 seconds to enter the dangerous area, but considering sudden changes in molten steel fluctuations, the response time should be much less than 45 seconds, preferably controlled within the range of several hundred milliseconds). This solution uses high-frequency sampling, lookup table method for rapid calculation, and hardware interlocking output to control the total response time within the optimal value (e.g., reaching the range of 80 ms), ensuring that the protective action is completed before the sudden change in the molten steel level.

[0062] Configure a watchdog timer. If a certain execution phase fails to complete the control loop within a preset execution time window, a reset is triggered, and the system enters a safe zone (ladle descent + air breaker). ≤ First gap threshold .

[0063] As a specific implementation method, in scenarios requiring rapid jacking or where the molten steel level fluctuates drastically, a dynamic prediction algorithm based on the rate of mass change is employed. Mass sensors acquire mass signals at a high sampling rate, the control unit calculates the rate of mass change, and predicts the future molten steel position height through trend extrapolation, thereby calculating the predicted safety clearance. When the predicted clearance is about to reach a threshold, a deceleration control signal is generated in advance to smoothly reduce the jacking speed and avoid mechanical shock caused by emergency braking.

[0064] Dynamic prediction algorithms based on the rate of change of mass, such as the Kalman filter state multi-step extrapolation prediction algorithm, are as follows: Reuse the mass balance model, based on The effective liquid level of molten steel in the ladle is constantly updated, and the prediction time is iterated forward by one prediction period. For prediction The effective liquid level of molten steel in the ladle at all times. , To predict safety gaps.

[0065] As a specific implementation method, a multi-sensor fusion strategy is adopted to improve system reliability. A mass sensor and an absolute encoder (for detecting the lifting and lowering position of the ladle) are configured simultaneously.

[0066] The real-time molten steel depth in the ladle is calculated using the mass sensor signal as the main control parameter. The absolute lifting height of the ladle detected by the encoder is the necessary input for gap calculation and also serves as a verification signal. The theoretical liquid level height calculated based on the encoder position should meet the consistency relationship with the liquid level calculated by mass (e.g., deviation threshold ≤ 30mm). If the deviation exceeds the limit, it is judged as a sensor malfunction, triggering a reset and entering a safe zone. ≤ First gap threshold In emergency reset conditions, when the mass sensor experiences temperature drift due to environmental interference, the system automatically increases the encoder signal weight and uses a fixed travel limit as backup protection.

[0067] In this embodiment, the lifting displacement sensor (encoder) is not only a verification method, but also a necessary technical means to calculate the dynamic safety clearance: without the real-time lifting height of the ladle, it is impossible to convert the relative liquid depth inside the ladle into an absolute liquid level elevation, and thus it is impossible to compare it with the elevation at the end of the argon blowing tube fixed in the vacuum tank.

[0068] Example 3 like Figure 3 As shown, a burn-out prevention control system for the argon blowing tube of an RH vacuum refining apparatus includes: The effective liquid level calculation module is used to establish a mass balance model based on the law of conservation of mass, based on the molten steel circulation flow rate at the immersion pipe and the ladle lifting speed, and to establish a liquid level dynamic model based on the mass balance model. Based on the mass balance model and the liquid level dynamic model, Kalman filtering is performed to estimate the real-time effective liquid level of molten steel in the ladle. The real-time safety gap value calculation module is used to calculate the real-time safety gap value between the end of the argon blowing pipe and the molten steel surface based on the real-time lifting height of the bottom of the ladle, the installation height of the end of the argon blowing pipe, and the real-time effective liquid level of the molten steel in the ladle. The ladle lifting adjustment module is used to adjust the ladle lifting action according to the real-time safety gap value.

[0069] Example 4 A computer program product includes a computer program / instructions that, when executed by a processor, implement the method for preventing burn-out of the argon blowing tube in the RH vacuum refining apparatus of Embodiment 2.

[0070] The contents not described in detail in this specification are prior art known to those skilled in the art. Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0071] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0072] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1The function specified in one or more boxes.

[0073] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0074] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit its scope of protection. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that after reading the present invention, they can still make various changes, modifications or equivalent substitutions to the specific implementation of the invention, but these changes, modifications or equivalent substitutions are all within the scope of protection of the pending claims of the invention.

Claims

1. A method for preventing burn-out of the argon blowing tube in an RH vacuum refining apparatus, characterized in that, include: Based on the molten steel circulation flow rate at the immersion tube and the ladle lifting speed, a mass balance model is established based on the law of mass conservation. A liquid level dynamic model is then established based on the mass balance model. Kalman filtering estimation is performed based on the mass balance model and the liquid level dynamic model to obtain the real-time effective liquid level of molten steel in the ladle. Calculate the real-time safety gap value between the end of the argon blowing pipe and the molten steel surface based on the real-time lifting height of the bottom of the ladle, the installation height of the end of the argon blowing pipe, and the real-time effective liquid level of the molten steel in the ladle. Adjust the ladle lifting action according to the real-time safety gap value.

2. The method for preventing burn-out of the argon blowing tube in the RH vacuum refining apparatus according to claim 1, characterized in that, The mass balance model is specifically represented as follows: In the formula, for The mass of molten steel remaining in the ladle at all times. The density of molten steel, for The circulation flow rate of molten steel at the immersion tube is constantly monitored. This represents the cross-sectional area of ​​the ladle's inner cavity. for The ladle lifting speed at all times for Process noise in the time-matter balance model; The specific representation of the liquid level dynamic model is as follows: In the formula, for Monitor the molten steel level in the ladle at all times. for Process noise in the dynamic model of liquid level at any time.

3. The method for preventing burn-out of the argon blowing tube in the RH vacuum refining apparatus according to claim 2, characterized in that, The specific method for obtaining the real-time effective liquid level of molten steel in the ladle by Kalman filtering estimation based on the mass balance dynamics model is as follows: Define the first mass Define a second mass to represent the mass of an empty ladle. Define the initial molten steel depth as the total mass of the ladle containing molten steel after steel is received and before it is lifted. For the second mass The corresponding molten steel level in the ladle Define the third mass for The real-time mass of molten steel remaining in the ladle under RH vacuum refining conditions will be measured. As observation data for Kalman filtering, ; Based on the mass balance model and the liquid level dynamic model, the discrete state transition equation of the Kalman filter is obtained, as shown in the following formula: In the formula, for The mass of molten steel remaining in the ladle at all times. for Monitor the molten steel level in the ladle at all times. The discrete state transition matrix, , For discrete control input matrix, , The sampling period; Prior estimation is performed based on discrete state equations: In the formula, for The optimal posterior estimate of the remaining molten steel mass in the ladle is obtained by using all sensor data as observations at all times. for Time based The optimal posterior estimate of the remaining molten steel mass in the ladle at time t is used to obtain the prior predicted value of the remaining molten steel mass in the ladle. for The optimal posterior estimate of the molten steel level in the ladle is obtained by using all sensor data as observations at all times. for Time based The optimal posterior estimate of the molten steel level in the ladle at time t is used to obtain the prior prediction of the molten steel level in the ladle. The prior covariance update formula is as follows: In the formula, for Covariance of the mass of remaining molten steel in the ladle under the optimal posterior estimate at time. for Covariance of the mass of remaining molten steel in the ladle as predicted a priori at each moment. for Covariance of molten steel level in ladle under the optimal posterior estimate at time. for Covariance of molten steel level in ladle predicted a priori at each moment. for The covariance between the estimation error of the remaining molten steel mass in the ladle under the optimal posterior estimation at a given time and the estimation of the molten steel level in the ladle. for The covariance between the prior prediction error of the remaining molten steel mass in the ladle and the estimation of the molten steel level in the ladle. for The linked covariance of the estimation error of the molten steel level in the ladle and the estimation of the mass of the remaining molten steel in the ladle under the optimal posterior estimation at the given time. for The covariance between the prior prediction error of the molten steel level in the ladle and the estimation of the remaining molten steel mass in the ladle. It is the transpose symbol. The process noise variance for the mass balance model. The process noise variance of the liquid level dynamic model; With the second mass As The initial value for the iteration is the initial depth of the molten steel. As The initial values ​​for iteration are used for posterior estimation, as shown in the following formula: In the formula, for Moment-time Kalman gain, This represents the best a posteriori estimate of the remaining molten steel mass in the ladle at the current moment. This is the post-hoc optimal estimate of the molten steel level in the ladle at the current moment. The posterior covariance update formula is: In the formula, It is the identity matrix. The variance of the posterior estimate of the mass of molten steel remaining in the ladle at the current moment. The posterior covariance of the estimation error of the remaining molten steel mass in the ladle with respect to the estimation of the molten steel level in the ladle. This is the posterior covariance of the estimation error of the molten steel level in the ladle with respect to the estimation of the mass of the remaining molten steel in the ladle. The variance of the liquid level at the current moment is estimated posteriorly. Through iteration, the final output is... and , As The effective mass of molten steel in the ladle at all times. As The effective liquid level of molten steel in the ladle at all times.

4. The method for preventing burn-out of the argon blowing tube in the RH vacuum refining apparatus according to claim 3, characterized in that, The third mass It is obtained by performing digital low-pass filtering and inertial compensation processing on the corresponding original quality signal.

5. The method for preventing burn-out of the argon blowing tube in the RH vacuum refining apparatus according to claim 3, characterized in that, Time Kalman Gain Specifically: In the formula, To measure the noise covariance.

6. The method for preventing burn-out of the argon blowing tube in the RH vacuum refining apparatus according to claim 1, characterized in that, The specific method for calculating the real-time safe gap value between the end of the argon blowing tube and the molten steel surface is as follows: In the formula, Let t be the dynamic safety gap value between the end of the argon blowing tube and the molten steel surface. The installation height of the argon blowing tube end. This refers to the real-time lifting height of the bottom of the ladle.

7. The method for preventing burn-out of the argon blowing tube in the RH vacuum refining apparatus according to claim 6, characterized in that, The specific method for adjusting the ladle lifting action based on the real-time safety gap value is as follows: when ≤ First gap threshold At the same time, control the ladle lifting speed. The first speed; When the first gap threshold < ≤Second gap threshold At the same time, control the ladle lifting speed. The second speed is less than the first speed. When the second gap threshold < ≤Third gap threshold At that time, the ladle lifting was stopped; when >Third gap threshold At that time, the ladle was controlled to descend at a third speed until... ≤ First gap threshold ; First gap threshold >Second gap threshold >Third gap threshold .

8. The method for preventing burn-out of the argon blowing tube in the RH vacuum refining apparatus according to claim 7, characterized in that, The method for preventing the argon blowing tube from burning out is divided into several execution stages. Each execution stage is completed within a preset execution time window. If an execution stage is not completed within the preset execution time window, the ladle is reset until... ≤ First gap threshold .

9. A burn-out prevention control system for the argon blowing tube of an RH vacuum refining apparatus, characterized in that, include: The effective liquid level calculation module is used to establish a mass balance model based on the law of conservation of mass, based on the molten steel circulation flow rate at the immersion pipe and the ladle lifting speed, and to establish a liquid level dynamic model based on the mass balance model. Based on the mass balance model and the liquid level dynamic model, Kalman filtering is performed to estimate the real-time effective liquid level of molten steel in the ladle. The real-time safety gap value calculation module is used to calculate the real-time safety gap value between the end of the argon blowing pipe and the molten steel surface based on the real-time lifting height of the bottom of the ladle, the installation height of the end of the argon blowing pipe, and the real-time effective liquid level of the molten steel in the ladle. The ladle lifting adjustment module is used to adjust the ladle lifting action according to the real-time safety gap value.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the method for preventing burn-out of the argon blowing tube in the RH vacuum refining apparatus as described in claims 1 to 8.

Images