High-efficiency continuous casting intelligent control method based on multi-source sensor monitoring

Abstract

The present application belongs to the technical field of continuous casting machine control, and relates to a high-efficiency continuous casting intelligent control method for a continuous casting machine based on multi-source sensor monitoring. The method is based on Fourier heat conduction law to reversely calculate the instantaneous heat flow density of each monitoring point, to inversely obtain the thermal resistance distribution of the liquid slag layer along the direction of strand drawing, and to obtain the discrete index distribution of the circumferential heat transfer at different positions along the direction of strand drawing. The target parameters of the physical state of the liquid slag layer adapted to the current continuous casting process condition are called to compare, the deviation type of the current physical state of the liquid slag layer is identified and its spatial position is located, so as to control the microwave generating device to emit a specific mode of microwave energy field to the corresponding deviated area in the mold, to close-loop adjust or stop the microwave output according to the re-inverted state parameters after the application, to realize the online sensing and close-loop regulation and control of the viscosity of the liquid slag layer in the mold, and to improve the surface quality and process stability of the continuous casting strand through the real-time feedback of the accurate intervention.

Description

Technical Field

[0001] This invention belongs to the field of continuous casting machine control technology, and relates to an efficient intelligent control method for continuous casting based on multi-source sensor monitoring. Background Technology

[0002] During continuous casting, the protective slag in the mold is heated to form a liquid slag layer. This liquid slag layer seeps between the copper plate and the billet shell, and its physical state, especially its viscosity, directly determines the lubrication effect and heat transfer uniformity. During high-speed casting, process conditions change rapidly, and maintaining the viscosity of the liquid slag layer within a suitable range is crucial to ensuring the surface and internal quality of the billet.

[0003] Currently, the control of flux in continuous casting production mainly relies on static pre-setting and manual experience. Specifically, a fixed flux formula is selected in advance based on the target steel grade and preset casting speed; its viscosity characteristics are determined before use. During production, operators make adjustments afterward based on indirect information such as observing the surface quality of the cast billet and monitoring cooling water parameters, combined with their personal experience, such as changing the casting speed or replacing the flux type.

[0004] The main drawback of existing technologies lies in the lack of direct monitoring and quantification of the real-time viscosity of the liquid slag layer within the crystallizer, and the inability to establish online closed-loop control logic based on real-time viscosity information. Because the thermal resistance distribution of the liquid slag layer along the casting direction and the degree of heat transfer dispersion in the circumferential direction cannot be perceived, it is impossible to promptly detect mismatches such as excessively high or low viscosity of the liquid slag layer caused by high casting speeds or fluctuations in operating conditions. Remedial measures can only be taken after quality defects appear in the cast billet, through reverse reasoning, which reduces production efficiency and product quality. Summary of the Invention

[0005] In view of this, in order to solve the problems mentioned in the background technology, a high-efficiency intelligent control method for continuous casting machines based on multi-source sensor monitoring is proposed.

[0006] The objective of this invention can be achieved through the following technical solution: This invention provides an efficient intelligent control method for continuous casting based on multi-source sensor monitoring, including: real-time acquisition of temperature data from multiple monitoring points embedded in the copper plate of the crystallizer, and inverse calculation of the instantaneous heat flux density of each monitoring point based on Fourier's law of thermal conductivity.

[0007] Based on the instantaneous heat flux density, parameters characterizing the current physical state of the liquid slag layer are obtained by inversion. These parameters include the thermal resistance distribution of the liquid slag layer along the drawing direction and the circumferential heat transfer discrete index distribution at different positions along the drawing direction.

[0008] The target parameters of the physical state of the liquid slag layer adapted to the current continuous casting process conditions are invoked. The target parameters include the target thermal resistance range and the target circumferential heat transfer discrete threshold for each thermal state zone.

[0009] By comparing parameters, the deviation type of the current physical state of the liquid slag layer is identified, and the spatial location of the deviation area in the crystallizer is located.

[0010] A microwave generator is controlled to emit a specific mode of microwave energy field into a corresponding deviation region within the crystallizer, wherein the energy parameters of the specific mode are associated with the deviation type.

[0011] Based on the physical state parameters of the new liquid slag layer obtained by re-inversion after the microwave energy field is applied, the microwave application is adjusted or stopped in a closed loop.

[0012] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The present invention collects the temperature data of the copper plate of the crystallizer, calculates the instantaneous heat flux density in reverse based on Fourier's thermal conductivity law, and then inversely obtains the thermal resistance distribution of the liquid slag layer and the circumferential heat transfer dispersion index along the billet pulling direction. The two respectively characterize the thickness structure and distribution uniformity of the liquid slag film, and together constitute a quantitative characterization of the real-time viscosity state of the liquid slag layer, solving the defects of the prior art that cannot directly obtain the real-time viscosity-related parameters of the liquid slag layer and can only passively discover the problem after the billet has defects.

[0013] (2) This invention compares the real-time inversion parameters of the monitoring points with the target parameters that are adapted to the current process conditions, identifies the deviation type based on the deviation judgment logic of over-viscosity or low-viscosity, and locks the boundary of the deviation area through connected domain search. This helps to overcome the shortcomings of the existing technology in that it cannot detect the mismatch of the liquid slag layer state in time, and cannot locate the abnormal area.

[0014] (3) The present invention configures microwave energy field mode parameters according to the deviation type and deviation amplitude, controls the microwave beam to align with the geometric center of the deviation region based on phased array beamforming, and re-inverts the parameters after application to evaluate the effect, dynamically adjust or stop application, thereby realizing fixed-point closed-loop control based on real-time feedback, avoiding the defects of existing technologies that can only perform global coarse adjustment, cannot evaluate the intervention effect and cannot form closed-loop optimization. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating the implementation steps of the method of the present invention.

[0017] Figure 2 This is a flowchart illustrating the steps involved in analyzing the thermal resistance distribution of the liquid slag layer along the billet pulling direction according to the present invention.

[0018] Figure 3 This is a flowchart illustrating the analysis steps for the circumferential heat transfer dispersion index distribution at different positions along the billet pulling direction in this invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 As shown, the present invention provides an efficient intelligent control method for continuous casting based on multi-source sensor monitoring, including: S1. Real-time acquisition of temperature data from multiple monitoring points embedded in the copper plate of the crystallizer, and inverse calculation of the instantaneous heat flux density of each monitoring point based on Fourier's law of thermal conductivity.

[0021] To monitor the temperature field of the copper plate in the crystallizer in real time and provide high-quality source data to help with subsequent heat flux density inversion, the real-time acquisition of temperature data from multiple monitoring points embedded in the copper plate in the crystallizer includes: dividing the hot surface of the copper plate in the crystallizer into multiple continuous grid regions along the drawing direction and circumferential direction, with each grid region covering at least one monitoring point.

[0022] Temperature acquisition devices, such as thermocouples, are deployed at each monitoring point to synchronously poll all monitoring points at a high sampling frequency of 10 to 50 Hz. Transient temperature data of the monitoring points are collected and processed by spatiotemporal denoising and discrete point compensation to output temperature data of multiple monitoring points embedded in the copper plate of the crystallizer.

[0023] The spatiotemporal denoising process involves correcting the transient temperature value of any monitoring point at the current moment. This correction value is obtained by linearly weighting the historical weighted average temperature of that monitoring point over multiple sampling periods with the weighted average temperature of spatially adjacent monitoring points at the same moment, where the sum of the time dimension weight and the spatial dimension weight is 1. The weights are determined based on the fluctuation characteristics of sensor signals acquired under non-bundling conditions: the variance of the time series of each monitoring point and the variance of the difference between it and its spatially adjacent points are statistically analyzed, denoted as the time variance and the spatial variance, respectively. The time dimension weight is set as the ratio of the time variance to the sum of the two variances, and the spatial dimension weight is set as the ratio of the spatial variance to the sum of the two variances. The final values ​​are determined through normalization.

[0024] The discrete point compensation process is as follows: For each monitoring point, the average and standard deviation of the temperature values ​​of its eight neighboring points at the same time are calculated. The sum of the average and three times the standard deviation is used as the upper limit of the discrete decision threshold for the monitoring point, and the difference between the average and three times the standard deviation is used as the lower limit of the discrete decision threshold. If the measured temperature value of a monitoring point exceeds the discrete decision range, it is determined as a discrete point and discarded. Simultaneously, based on the temperature values ​​of the four nearest valid monitoring points in the four orthogonal directions in the discrete point space, a bilinear interpolation algorithm is used to calculate a replacement temperature value to fill the data gap.

[0025] The inverse calculation of the instantaneous heat flux density at each monitoring point based on Fourier's law of thermal conductivity includes: using the copper plate wall of the crystallizer as the heat conduction medium, establishing an inverse analytical equation for heat conduction based on Fourier's law of thermal conductivity, wherein the inverse analytical equation is expressed as: .

[0026] In the equation, Let represent the instantaneous heat flux density to be determined. This represents the real-time equivalent thermal conductivity of the copper plate. The negative sign indicates that the coordinate axis of the copper plate thickness in the crystallizer is defined as pointing from the cold side to the hot side as the positive direction. According to Fourier's law of thermal conductivity, the direction of the heat flux density vector is opposite to the direction of the temperature gradient, so a negative sign is needed for directional correction.

[0027] The right side of the equation follows the basic form of Fourier's law, that is, the magnitude of the heat flux density is equal to the thermal conductivity (referring to the heat transfer capacity). ) and heat transfer driving force (referring to The product of these two terms, denoted as , represents the physical meaning that when a temperature difference exists within the copper plate, heat will spontaneously flow from the high-temperature region to the low-temperature region. The amount of heat flowing depends on the size of the temperature difference and the thermal conductivity of the medium; the greater the thermal conductivity and the smaller the thermal resistance, the more heat can flow under the same temperature difference.

[0028] The reverse analysis of the equation is reflected in the fact that: usually, the internal temperature field is solved when the instantaneous heat flux density and the real-time equivalent thermal conductivity of the copper plate are known. This invention cannot directly measure the instantaneous heat flux density, but the internal temperature field can be obtained by measuring with a sensor, and the real-time equivalent thermal conductivity of the copper plate can be solved by the total heat flow, and the instantaneous heat flux density can be deduced in reverse.

[0029] At least two temperature data points of the same period adjacent to each monitoring point are selected. Based on the spatial location difference and temperature difference, the temperature gradient of the monitoring point is calculated using the central difference method. Specifically, based on the grid area divided by the hot surface of the crystallizer copper plate, each monitoring point is located at the center of the corresponding grid area. For any internal monitoring point... ,in Indicates the position index along the throwing direction. This indicates the position index along the circumference, and retrieves the temperature values ​​of the current monitoring point and its adjacent monitoring points at the same sampling time.

[0030] In the throwing direction, select the monitoring point Two adjacent monitoring points and The temperature values ​​of the three were recorded as follows: , , The spatial distance between adjacent monitoring points is determined based on the side length of the grid area along the throwing direction. ,Will and The difference, divided by twice , obtain monitoring points Temperature gradient in the throwing direction.

[0031] Similarly, select monitoring points in the circumferential direction. Two adjacent monitoring points on the left and right and Based on the temperature gradient acquisition method described above for the throwing direction, the monitoring points are determined. Temperature gradient in the circumferential direction.

[0032] Furthermore, for monitoring points located at the boundaries of the copper plate, if an adjacent monitoring point is missing in a certain direction, forward or backward differential is used instead of center differential. For example, for monitoring points on the starting boundary of the drawing direction, the temperature gradient in the drawing direction is determined by forward differential. For monitoring points on the ending boundary of the drawing direction, the temperature gradient in the drawing direction is determined by backward differential. The same principle applies to circumferential boundary monitoring points.

[0033] Ultimately, the temperature gradient amplitude at each monitoring point is obtained by combining its directional gradient and circumferential gradient using the Euclidean norm.

[0034] Based on the real-time collected cooling water flow rate and inlet / outlet temperature difference of the crystallizer, the heat flow rate lost by the copper plate through the cooling water per unit time is calculated. The total heat flow rate transferred through the hot surface of the copper plate is obtained by superimposing the rate of change of the internal energy of the copper plate itself within the same time period.

[0035] It should be noted that the heat loss of the copper plate per unit time through the cooling water is specifically the cumulative calculation result of the cooling water flow rate of the crystallizer, the inlet and outlet temperature difference, and the preset specific heat capacity and preset density of the cooling water.

[0036] The process for obtaining the rate of change of the internal energy of the copper plate within the same time period is as follows: Based on the temperature data of the monitoring points collected at the current moment, calculate the average temperature of all monitoring points at the current moment, and record it as the first average temperature. Calculate the average temperature of all monitoring points at the previous moment, and record it as the second average temperature. Divide the difference between the first average temperature and the second average temperature by the sampling time interval to obtain the time rate of change of the average temperature of all monitoring points.

[0037] The rate of change over time is accumulated with the total mass of the copper plate in the crystallizer and the preset specific heat capacity of the copper plate material to obtain the rate of change of the internal energy of the copper plate itself within the same time period.

[0038] The overall temperature gradient characteristic quantity is obtained by integrating the area of ​​the temperature gradient field at all monitoring points.

[0039] The real-time equivalent thermal conductivity of the copper plate is determined by negatively taking the ratio of the total heat flux to the overall temperature gradient characteristic.

[0040] Substitute the real-time equivalent thermal conductivity of the copper plate and the temperature gradient calculated at each monitoring point into the inverse analytical equation to solve for the instantaneous heat flux density at each monitoring point.

[0041] It should be noted that the central difference method used in calculating the temperature gradient is based on the quasi-steady-state assumption, which assumes that the temperature field changes slowly within an extremely short sampling period, approximating steady-state heat conduction. Simultaneously, by superimposing the rate of change of the copper plate's internal energy, the method compensates for the heat accumulation caused by temperature field changes, thus obtaining a more accurate total heat flux. This method simplifies calculations in engineering while also reflecting transient effects.

[0042] S2. Based on the instantaneous heat flux density, parameters characterizing the current physical state of the liquid slag layer are obtained by inversion. The parameters include the thermal resistance distribution of the liquid slag layer along the drawing direction and the circumferential heat transfer discrete index distribution at different positions along the drawing direction.

[0043] like Figure 2 As shown, the thermal resistance distribution of the liquid slag layer along the billet pulling direction is obtained by the following method: obtaining the calculated temperature of the hot surface of the copper plate of the crystallizer corresponding to the target monitoring point at the current moment, and subtracting it from the temperature of the cold surface on the cooling water side to obtain the interface temperature difference.

[0044] Divide the absolute value of the interface temperature difference by the instantaneous heat flux density to obtain the total thermal resistance of the target monitoring point.

[0045] The thermal resistance of the copper plate body and the inherent thermal resistance of the cooling water film are subtracted from the total thermal resistance at the target temperature measurement point to extract the comprehensive filling thermal resistance between the hot surface of the crystallizer and the billet shell.

[0046] The overall filling thermal resistance is calculated by subtracting the ratio of the estimated thickness of the solid slag layer at the target monitoring point to the preset thermal conductivity of the solid slag film. This separates the local thermal resistance contributed solely by the state of the liquid slag layer. The monitoring points are then arranged in sequence along the billet pulling direction to form a one-dimensional liquid slag layer thermal resistance distribution sequence. Specifically, the monitoring points can be sorted from smallest to largest according to their vertical coordinates along the billet pulling direction within the crystallizer.

[0047] It should be noted that the calculated temperature of the hot surface of the copper plate of the crystallizer corresponding to the target monitoring point at the current moment is obtained by forward calculation using Fourier's law of thermal conductivity. The implementation process is as follows: Based on the quasi-steady-state thermal conductivity assumption, considering that the thickness of the copper plate of the crystallizer is much smaller than its width and height, and that the heat flow is mainly transferred along the thickness direction, the slight heat conduction along the drawing direction and circumferential direction is ignored. One-dimensional Fourier's law is used for forward calculation, and the calculated temperature of the hot surface of the copper plate of the crystallizer is expressed as the sum of the cold surface temperature term and the temperature rise term. The temperature rise term is obtained by multiplying the transient heat flux density of the target monitoring point by the thickness of the copper plate at the location of the target monitoring point and then dividing by the real-time equivalent thermal conductivity of the copper plate. The cold surface temperature term is obtained by retrieving the measured temperature value of the cold side of the copper plate closest to the target monitoring point, which can be directly collected by configuring thermocouples.

[0048] The above-mentioned estimated solid slag layer thickness is dynamically output based on the current casting speed and the vertical distance of the monitoring point from the meniscus. According to solidification theory, the solid slag layer thickness is usually proportional to the square root of the residence time. The residence time is approximately equal to the vertical distance of the monitoring point from the meniscus divided by the casting speed. Therefore, the estimated solid slag layer thickness is calculated as follows: the square root of the ratio of the vertical distance of the monitoring point from the meniscus to the current casting speed is taken, and the result is multiplied by a thickness coefficient related to the physical properties of the protective slag to obtain the estimated solid slag layer thickness. The thickness coefficient is determined by: for the type of protective slag and the grade of steel being cast, the solid slag layer thickness samples at different casting speeds and locations are measured through offline experiments, and the thickness coefficient value is fitted and calibrated using the least squares method.

[0049] The preset thermal conductivity of the solidified slag film is an inherent physical property of the type of protective slag used, provided by the supplier, and is typically between 0.8 and 1.5. Within the range.

[0050] like Figure 3 As shown, the circumferential heat transfer discrete index distribution at different positions along the billet pulling direction is obtained in the following way: the monitoring points distributed along the billet pulling direction are grouped according to their circumferential positions, and each group constitutes a circumferential monitoring point set.

[0051] For each circumferential set of monitoring points, the instantaneous heat flux density corresponding to all monitoring points within the same sampling time is obtained, forming an instantaneous heat flux density sequence.

[0052] The standard deviation of the instantaneous heat flux density sequence is calculated and normalized. Specifically, the normalization method is to divide the calculated standard deviation by the arithmetic mean of all instantaneous heat flux densities in the same sequence to obtain the heat transfer dispersion index corresponding to each circumferential monitoring point set.

[0053] All heat transfer discrete indices of the set are arranged in order along the throwing direction to form a heat transfer discrete index distribution sequence.

[0054] S3. Call the target parameters of the physical state of the liquid slag layer that are adapted to the current continuous casting process conditions. The target parameters include the target thermal resistance range and the target circumferential heat transfer discrete threshold of each thermal state zone.

[0055] The process of calling the target parameters of the physical state of the liquid slag layer that are adapted to the current continuous casting process conditions is as follows: obtain the current billet pulling speed, steel type, and protective slag model information of the continuous casting machine.

[0056] Accessing a preset process knowledge base, the system matches the optimal slag film working window for the steel grade and protective slag type at the current billet pulling speed. The crystallizer is divided into multiple thermal state zones along the pulling direction, each zone defined by its height from the meniscus. And the definition of heat transfer mechanism, specifically including: meniscus region ( Heat transfer is mainly through the gap between the initial blank and the copper plate. The heat transfer mechanism is interfacial heat exchange under extremely high heat flux density. The liquid slag layer needs to form an extremely thin and uniform liquid film in this area to inhibit adhesion.

[0057] Central heat transfer zone ( The heat transfer mechanism is mainly based on the heat conduction of the solid slag film. The heat flow is controlled by the slag film crystal layer. The liquid slag layer needs to maintain a stable solid-liquid two-phase structure in this region to achieve a balance between heat transfer and lubrication.

[0058] Lower lubrication zone ( Heat transfer is mainly achieved through the air gap between the billet and the copper plate. The heat transfer mechanism is insulation and lubrication under low heat flux density. The liquid slag layer needs to maintain a sufficient solid slag film thickness in this area to reduce frictional resistance.

[0059] The data for the optimal slag film working window is stored in thermal state partitions.

[0060] Determine the thermal state zone of each monitoring point, and extract the target thermal resistance range and target circumferential heat transfer discrete threshold of the liquid slag layer in the corresponding thermal state zone.

[0061] S4. Identify the deviation type of the current physical state of the liquid slag layer by parameter comparison, and locate the spatial position of the deviation area in the crystallizer.

[0062] The process of identifying the deviation type of the current physical state of the liquid slag layer and locating the spatial position of the deviation region in the crystallizer includes: if the local thermal resistance of the liquid slag layer at a certain monitoring point is greater than the upper limit of the target thermal resistance range of its thermal state zone, and the heat transfer dispersion index of its circumferential monitoring point set continuously increases within the sampling period window of the most recent preset number of sampling points and exceeds the target circumferential heat transfer dispersion threshold, then the physical state of the liquid slag layer at that monitoring point is classified as the first deviation type. Here, "continuously increasing" is manifested in the circumferential heat transfer dispersion index of each sampling period window being greater than the circumferential heat transfer dispersion index of the previous sampling period window.

[0063] If the local thermal resistance of the liquid slag layer at a certain monitoring point is less than the lower limit of the target thermal resistance range for its thermal state zone, and the heat transfer dispersion index of the circumferential monitoring point set to which it belongs increases sharply relative to the sampling period window of the previous preset number and exceeds the target circumferential heat transfer dispersion threshold, then the physical state of the liquid slag layer at that monitoring point is classified as the second deviation type. The sharp increase is determined as follows: the moving average of the heat transfer dispersion index of the sampling period window of the previous preset number is calculated; the difference between the heat transfer dispersion index of the circumferential monitoring point set to which the monitoring point belongs and the moving average is calculated; this difference is then divided by the moving average to obtain the relative rate of change. If the relative rate of change is greater than the preset abrupt change rate threshold, then a sharp increase has occurred.

[0064] The preset mutation rate threshold Different zones are defined according to different thermal states: meniscus region Central heat transfer zone Lower lubrication area The preset mutation rate threshold is obtained by statistically analyzing historical data of steel grades under normal production conditions and taking the quantile of the rate of change at a 95% confidence level.

[0065] It should be noted that the first deviation type refers to the over-viscosity type, which is usually characterized by a slow change process that leads to deterioration and uneven heat transfer, such as excessively thick slag film, excessively high crystallization rate, or formation of slag nodules.

[0066] The second type of deviation points to the low viscosity type, which is usually characterized by an excessively thin slag film or local rupture, or even instantaneous direct contact between molten steel and copper plate, resulting in a sudden increase in local heat flux density. At the same time, due to the random location of the rupture, the circumferential heat transfer dispersion index increases sharply, showing a sudden and localized heat transfer anomaly.

[0067] Based on the coordinates of the grid region to which the monitoring point belongs, grid regions with the same deviation type are searched through connected components to locate the spatial boundary of the deviation region in the three-dimensional space of the crystallizer. The spatial boundary of the deviation region refers to the closed polygonal region enclosed by consecutive adjacent monitoring points of the same deviation type.

[0068] S5. Control the microwave generator to emit a specific mode of microwave energy field into the corresponding deviation area inside the crystallizer, wherein the energy parameter of the specific mode is associated with the deviation type.

[0069] When controlling the microwave generator to emit a microwave energy field of a specific mode, the energy parameters of the specific mode are configured as follows: The deviation amplitude of the local thermal resistance of each deviation region liquid slag layer relative to the target thermal resistance interval of its thermal state zone, and the deviation amplitude of the heat transfer dispersion index of its circumferential monitoring point set relative to the target circumferential heat transfer dispersion threshold are calculated respectively, denoted as thermal resistance deviation amplitude and dispersion deviation amplitude. The specific calculation process is as follows: When the local thermal resistance of the deviation region liquid slag layer is less than the lower limit of the target thermal resistance interval of its thermal state zone, the local thermal resistance is subtracted from the lower limit of the target thermal resistance interval of the thermal state zone where the deviation region liquid slag layer is located, and the ratio is calculated with the lower limit of the target thermal resistance interval to obtain the thermal resistance deviation amplitude.

[0070] When the local thermal resistance of the liquid slag layer in the deviation area is greater than the upper limit of the target thermal resistance range of its thermal state zone, the local thermal resistance of the liquid slag layer in the deviation area is subtracted from the upper limit of the target thermal resistance range of its thermal state zone, and the ratio is calculated with the upper limit of the target thermal resistance range to obtain the thermal resistance deviation amplitude.

[0071] The discrete deviation amplitude is obtained by subtracting the target circumferential heat transfer discrete threshold from the heat transfer discrete index of the set of circumferential monitoring points, and then performing a ratio calculation with the target circumferential heat transfer discrete threshold.

[0072] For the first type of deviation, the physical cause is mostly slag film thickening or crystallization, requiring a gentle and continuous energy input to promote its melting and thinning. Therefore, determining the power density by linearly weighting the amplitude of thermal resistance deviation and discrete deviation requires the microwave generator to switch to continuous wave modulation mode to output the determined power density in a fixed low frequency band (e.g., 2.45 GHz).

[0073] For the second type of deviation, the physical causes are mostly slag film rupture or instantaneous closure of the air gap, requiring rapid and high-intensity energy impact to reconstruct the local slag film or suppress the non-uniform shrinkage of the billet shell. Therefore, the pulse width is determined by the discrete deviation amplitude, and the average power density is determined by the thermal resistance deviation amplitude. Specifically, the ratio of the deviation amplitude to the maximum deviation amplitude can be linearly mapped to the parameter control range to obtain a specific value. This requires the microwave generator to switch to pulse wave control mode to output the average power density at the corresponding pulse width in a fixed high-frequency band (e.g., 5.8 GHz).

[0074] The microwave generator is a high-temperature resistant phased array, installed between the cooling water jacket and the outer shell of the crystallizer's copper plate. Each radiating unit faces the hot surface of the crystallizer through a quartz window or a ceramic wave-transmitting window. The wave-transmitting window material is alumina or boron nitride, which can withstand temperatures above 1200℃ and allows 2.45GHz or 5.8GHz microwaves to penetrate.

[0075] The process of controlling the microwave generator to emit a specific mode of microwave energy field into the corresponding deviation area within the crystallizer is as follows: a three-dimensional coordinate system is established with the geometric center of the microwave generator array as the origin, and the spatial position coordinates of each radiation unit in the array are pre-calibrated.

[0076] Based on the spatial boundary of the deviation area, calculate the position of its geometric center in the three-dimensional coordinate system to obtain the target pointing vector from the origin of the coordinate system to the position of the geometric center.

[0077] The target pointing vector is decomposed into horizontal azimuth and vertical elevation angles. Combined with the calibration position of each radiating element, and based on the phased array beamforming theory, in order to ensure that all electromagnetic waves emitted by the array are superimposed in phase at the target point, thus forming the main energy lobe, the transmitted signal of each radiating element needs a specific phase delay relative to the reference point. The phase compensation value required for beam directional focusing of each radiating element in the array can be calculated using the following formula: .

[0078] in, This is the phase compensation value required for the radiating element. Let be the wave number, and its value is . Divide by the microwave wavelength determined by the operating frequency, For the calibration spatial position coordinate components of the radiating element, It is the unit direction cosine of the target pointing vector.

[0079] Refers to the radiating unit in The position of the direction and the target pointing vector The product of direction cosines represents the radiating element in Similarly, the projected distance relative to the reference point in the direction. They respectively refer to the radiating units in The projected distance relative to the reference point in the direction.

[0080] This represents the projected length of the radiating element's calibrated spatial position onto the target pointing vector. It reflects the relative spatial delay of the radiating element's emitted wavefront relative to the reference point when it reaches the isophase surface of the target pointing vector. The sign of the spatial delay corresponds to the need for phase lead or lag compensation of the element's emitted wavefront to offset the path difference caused by positional differences and ensure the consistency of the wavefront in the target direction.

[0081] Furthermore, the above formula is based on the far-field plane wave assumption and is applicable when the target distance is greater than twice the array aperture. When the target is in the near-field region, a phase compensation formula based on the distance difference can be used instead. Specifically, the distance between the radiating element's calibration spatial position and the target pointing vector is calculated, the distance between the reference point and the target pointing vector is subtracted, and finally multiplied by the wavenumber.

[0082] The calculated set of phase compensation values ​​is loaded into the microwave waveguide controller, which drives the microwave generator array to generate a phase-synchronized microwave beam. The main lobe of the microwave beam is aligned with and focused on the geometric center of the off-region. The focusing means that the energy density reaches its maximum at the target point and the focal spot size is less than one-third of the size of the off-region.

[0083] S6. Based on the physical state parameters of the new liquid slag layer obtained by re-inversion after the microwave energy field is applied, adjust or stop the microwave application in a closed loop.

[0084] The criteria for determining whether to adjust or stop microwave application in a closed loop include: re-execute the temperature acquisition and parameter inversion process within the set observation period after microwave application.

[0085] Calculate the deviation of the current local thermal resistance and the heat transfer dispersion index of its corresponding circumferential monitoring point set from their respective target parameters.

[0086] If the microwave transmission has returned to the target range, then microwave transmission will cease.

[0087] If the deviation is still present but the magnitude of the deviation is decreasing, then adjust the microwave output power density according to the current magnitude of the deviation.

[0088] If there is no improvement within multiple consecutive observation periods, specifically if the local thermal resistance and heat transfer dispersion index of the liquid slag layer in the deviation area do not show a trend of convergence to the target range within three or more consecutive periods, that is, the current deviation amplitude has not decreased or has increased compared to the deviation amplitude at the time of the first detection, it can be determined that there is no improvement in microwave control, triggering equipment fault warning and switching to the conventional process intervention procedure.

[0089] It should also be noted that if multiple deviation areas exist at the same time and the deviation type is the same, the geometric center coordinates of each area are linearly weighted and fused according to the predefined position weights to obtain the comprehensive target pointing point, and the microwave parameters are configured according to the maximum deviation amplitude.

[0090] If multiple deviation areas exist simultaneously and have different types of deviation, the second type of deviation area should be dealt with first because it is sudden and high-risk. The first type of deviation area should be dealt with after it recovers.

[0091] If multiple regions are too far apart to be effectively covered by a single microwave focusing, a time-division polling method is used to apply microwave energy fields to each region sequentially.

[0092] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A method for efficient intelligent control of continuous casting machines based on multi-source sensor monitoring, characterized in that, include: Temperature data from multiple monitoring points embedded in the copper plate of the crystallizer are collected in real time, and the instantaneous heat flux density of each monitoring point is calculated in reverse based on Fourier's law of thermal conductivity. Based on the instantaneous heat flux density, parameters characterizing the current physical state of the liquid slag layer are obtained by inversion. These parameters include the thermal resistance distribution of the liquid slag layer along the drawing direction and the circumferential heat transfer discrete index distribution at different positions along the drawing direction. The target parameters of the physical state of the liquid slag layer adapted to the current continuous casting process conditions are called. The target parameters include the target thermal resistance range and the target circumferential heat transfer discrete threshold of each thermal state zone. By comparing parameters, the deviation type of the current liquid slag layer's physical state is identified, and the spatial location of the deviation region within the crystallizer is determined, including: If the local thermal resistance of the liquid slag layer at a certain monitoring point is greater than the upper limit of the target thermal resistance range of the thermal state zone in which it is located, and the heat transfer dispersion index of the circumferential monitoring point set to which it belongs continues to increase within the sampling period window of the most recent preset number and exceeds the target circumferential heat transfer dispersion threshold, then the physical state of the liquid slag layer at that monitoring point is classified as the first deviation type. If the local thermal resistance of the liquid slag layer at a certain monitoring point is less than the lower limit of the target thermal resistance range of the thermal state zone in which it is located, and the heat transfer dispersion index of the circumferential monitoring point set to which it belongs increases sharply relative to the sampling period window under the previous preset number and exceeds the target circumferential heat transfer dispersion threshold, then the physical state of the liquid slag layer at that monitoring point is classified as the second deviation type. Based on the coordinates of the grid region to which the monitoring point belongs, grid regions with the same deviation type are searched through connected components to lock the spatial boundary of the deviation region in the three-dimensional space of the crystallizer; The microwave generator is controlled to emit a microwave energy field of a specific mode into a corresponding deviation region within the crystallizer, wherein the energy parameters of the specific mode are associated with the deviation type. The energy parameters for the specific mode are configured in the following way: Calculate the deviation amplitude of the local thermal resistance of the liquid slag layer in each deviation region relative to the target thermal resistance range of its thermal state zone, and the deviation amplitude of the heat transfer dispersion index of its circumferential monitoring point set relative to the target circumferential heat transfer dispersion threshold, denoted as thermal resistance deviation amplitude and dispersion deviation amplitude. For the first type of deviation, the power density is determined by the linear weighted fusion result of the thermal resistance deviation amplitude and the discrete deviation amplitude. The microwave generator is required to switch to continuous wave modulation mode to output the determined power density in a fixed low frequency band. For the second type of deviation, the pulse width is determined by the discrete deviation amplitude, and the average power density is determined by the thermal resistance deviation amplitude. The microwave generator is required to switch to pulse wave modulation mode to output the average power density at the corresponding pulse width in a fixed high frequency band. Based on the physical state parameters of the new liquid slag layer obtained by re-inversion after the microwave energy field is applied, the microwave application is adjusted or stopped in a closed loop.

2. The intelligent control method for high-efficiency continuous casting based on multi-source sensor monitoring for continuous casting machines according to claim 1, characterized in that, The real-time acquisition of temperature data from multiple monitoring points embedded in the copper plate of the crystallizer includes: The hot surface of the copper plate in the crystallizer is divided into multiple continuous grid areas along the drawing direction and circumference, and each grid area covers at least one monitoring point; The transient temperature data of the monitoring points is collected and processed through spatiotemporal denoising and discrete point compensation, and finally outputs the temperature data of multiple monitoring points embedded in the copper plate of the crystallizer.

3. The intelligent control method for high-efficiency continuous casting based on multi-source sensor monitoring for continuous casting machines according to claim 1, characterized in that, The inverse calculation of the instantaneous heat flux density at each monitoring point based on Fourier's law of thermal conductivity includes: Using the copper plate wall of the crystallizer as the heat conduction medium, an inverse analytical equation for heat conduction is established based on Fourier's law of thermal conductivity. At least two temperature data points of the same period adjacent to each monitoring point are selected, and the temperature gradient of the monitoring point is calculated by the central difference method based on the spatial location difference and temperature difference. Based on the real-time collected cooling water flow rate and inlet / outlet temperature difference of the crystallizer, the heat flow rate lost by the copper plate through the cooling water per unit time is calculated. The total heat flow rate transferred through the hot surface of the copper plate is obtained by superimposing the rate of change of the internal energy of the copper plate itself within the same time. By integrating the temperature gradient field at all monitoring points, the overall temperature gradient characteristic quantity is obtained. The real-time equivalent thermal conductivity of the copper plate is determined by negatively taking the ratio of the total heat flux to the overall temperature gradient characteristic. Substitute the real-time equivalent thermal conductivity of the copper plate and the temperature gradient calculated at each monitoring point into the inverse analytical equation to solve for the instantaneous heat flux density at each monitoring point.

4. The intelligent control method for high-efficiency continuous casting based on multi-source sensor monitoring for continuous casting machines according to claim 3, characterized in that, The thermal resistance distribution of the liquid slag layer along the billet pulling direction is obtained through the following method: Obtain the calculated temperature of the hot surface of the copper plate of the crystallizer corresponding to the target monitoring point at the current moment, and obtain the interface temperature difference by subtracting it from the temperature of the cold surface on the cooling water side. Divide the absolute value of the interface temperature difference by the instantaneous heat flux density to obtain the total thermal resistance of the target monitoring point; Subtracting the thermal resistance of the copper plate body and the inherent thermal resistance of the cooling water film from the total thermal resistance at the target temperature measurement point, the comprehensive filling thermal resistance between the hot surface of the crystallizer and the billet shell is extracted. The overall filling thermal resistance is calculated by subtracting the ratio of the estimated thickness of the solid slag layer at the target monitoring point to the preset thermal conductivity of the solid slag film. This separates the local thermal resistance contributed solely by the state of the liquid slag layer, and the monitoring points are arranged sequentially according to the billet pulling direction to form a one-dimensional liquid slag layer thermal resistance distribution sequence.

5. The intelligent control method for high-efficiency continuous casting based on multi-source sensor monitoring for continuous casting machines according to claim 3, characterized in that, The circumferential heat transfer dispersion index distribution at different positions along the billet pulling direction is obtained through the following method: The monitoring points distributed along the throwing direction are grouped according to their circumferential positions, and each group constitutes a circumferential monitoring point set; For each set of monitoring points in the perimeter, the instantaneous heat flux density corresponding to all monitoring points within the same sampling time is obtained, forming an instantaneous heat flux density sequence; The standard deviation of the instantaneous heat flux density sequence is calculated and normalized to obtain the heat transfer discrete index corresponding to each circumferential monitoring point set; All heat transfer discrete indices of the set are arranged in order along the throwing direction to form a heat transfer discrete index distribution sequence.

6. The intelligent control method for high-efficiency continuous casting based on multi-source sensor monitoring for continuous casting machines according to claim 1, characterized in that, The process of calling the target parameters of the physical state of the slag layer that are adapted to the current continuous casting process conditions is as follows: Obtain information on the current billet casting speed, steel type, and mold flux type of the continuous casting machine; Access the preset process knowledge base to match the optimal slag film working window for the steel type and protective slag model at the current billet pulling speed. The crystallizer is divided into multiple thermal state zones along the billet pulling direction. Each thermal state zone is defined according to its height from the meniscus and the heat transfer mechanism. The data of the optimal slag film working window is stored according to the thermal state zones. Determine the thermal state zone of each monitoring point, and extract the target thermal resistance range and target circumferential heat transfer discrete threshold of the liquid slag layer in the corresponding thermal state zone.

7. The intelligent control method for high-efficiency continuous casting based on multi-source sensor monitoring for continuous casting machines according to claim 1, characterized in that, The process of controlling the microwave generator to emit a specific mode of microwave energy field into the corresponding off-center region within the crystallizer is as follows: A three-dimensional coordinate system is established with the geometric center of the microwave generator array as the origin, and the spatial position coordinates of each radiation unit in the array are pre-calibrated. Based on the spatial boundary of the off-region, calculate the position of its geometric center in the three-dimensional coordinate system to obtain the target pointing vector from the origin of the coordinate system to the position of the geometric center. The target pointing vector is decomposed into horizontal azimuth and vertical elevation angles. Based on the calibration position of each radiating element and the phased array beamforming theory, the phase compensation value required for each radiating element in the array to achieve beam directional focusing is calculated. The calculated set of phase compensation values ​​is loaded into the microwave waveguide controller, which drives the microwave generator array to generate a phase-synchronized microwave beam. The main lobe of the microwave beam is aligned with and focused on the geometric center of the off-region.

8. The intelligent control method for high-efficiency continuous casting based on multi-source sensor monitoring for continuous casting machines according to claim 1, characterized in that, The criteria for determining whether to adjust or stop microwave application in a closed loop include: Within the set observation period after microwave application, the temperature acquisition and parameter inversion process is repeated. Calculate the deviation of the current local thermal resistance and the heat transfer dispersion index of its circumferential monitoring point set relative to their respective target parameters; If the microwave transmission has returned to the target range, then microwave transmission should be stopped. If the deviation is still present but the magnitude of the deviation is decreasing, then adjust the microwave output power density according to the current magnitude of the deviation. If there is no improvement within several consecutive observation periods, an equipment failure warning will be triggered and the process will switch to the regular process intervention procedure.

Images